Experimental study on the deformation and failure mechanism of overburden rock during coal mining using a comprehensive intelligent sensing method

Gang Chng,Wntao Xu,Bin Shi,Jinghong Wu,Binyang Sun,Honghu Zhu*

a School of Computer Science,North China Institute of Science and Technology,Beijing,101601,China

b School of Earth Sciences and Engineering,Nanjing University,Nanjing,210023,China

c State Key Laboratory of Geohazard Prevention and Geoenvironment Protection,Chengdu University of Technology,Chengdu,610059,China

d School of Civil Engineering,Suzhou University of Science and Technology,Suzhou,215009,China

e State Key Laboratory of Mining Response and Disaster Prevention and Control in Deep Coal Mines,Anhui University of Science and Technology,Huainan,232001,China

Keywords:Similarity model Distributed fiber optic sensing(DFOS)Overburden deformation Failure mechanism Coal seam mining

A B S T R A C T Understanding the spatiotemporal evolution of overburden deformation during coal mining is still a challenge in engineering practice due to the limitation of monitoring techniques.Taking the Yangliu Coal Mine as an example,a similarity model test was designed and conducted to investigate the deformation and failure mechanism of overlying rocks in this study.Distributed fiber optic sensing(DFOS),highdensity electrical resistivity tomography(HD-ERT)and close-range photogrammetry(CRP)technologies were used in the test for comprehensive analyses.The combined use of the three methods facilitates the investigation of the spatiotemporal evolution characteristics of overburden deformation,showing that the mining-induced deformation of overburden strata was a dynamic evolution process.This process was accompanied by the formation,propagation,closure and redevelopment of separation cracks.Moreover,the key rock stratum with high strength and high-quality lithology played a crucial role in the whole process of overburden deformation.There were generally three failure modes of overburden rock layers,including bending and tension,overall shearing,and shearing and sliding.Shear failure often leads to overburden falling off in blocks,which poses a serious threat to mining safety.Therefore,realtime and accurate monitoring of overburden deformation is of great significance for the safe mining of underground coal seams.

1.Introduction

Currently,coal resources are still an essential energy source in many countries.In China,coal is a basic energy and important industrial raw material,and its consumption accounts for more than half of the total energy consumption for a long time,making a positive contribution to the development of the national economy.The statistics of China’s coal output from 2012 to 2021(Fig.1)show that the coal industry has maintained an average production of approximately 3.5 billion tons per year in the past decade and has gradually increased in recent years.Taking China’s energy mix in 2021 as an example,coal consumption still accounts for 56%of total energy consumption(National Bureau of Statistics of the People’s Republic of China,2022).

Fig.1.China’s coal output during 2012-2021.

Compared with the reserves of crude oil and natural gas,China’s coal resources are relatively rich.However,most of the coal resources are deeply buried underground.Therefore,the primary coal mining method adopted in China is underground mining,accounting for more than 90% of China’s coal output.Extensive coal exploitation has led to challenging environmental problems and energy utilization issues.For instance,a large amount of methane is generated in longwall mining operations,causing potential gas leakage risks(Schatzel et al.,2012;Sechman et al.,2020).Underground mining destroys the storage environment of water resources and adversely affects the groundwater quality in mining areas(Howladar,2013;Li et al.,2015;Yousefi et al.,2020).In longwall mining,sudden roof collapse often leads to surface subsidence and local seismic activities(Sasaoka et al.,2015;Konicek and Waclawik,2018;Salmi et al.,2019;Forbes et al.,2020).The strata movements caused by coal mining not only affect the viability of the mine and increase the mining cost but also hinder the effective exploitation of oil and gas resources(Zeynali,2012;Wang et al.,2020).The causes of these adverse factors can be attributed to overburden deformation,as well as the formation of separation and cracks.Therefore,understanding the deformation mechanism of overburden rocks and crack evolution due to coal mining is of great significance for the safe mining of coal seams in underground space.

However,it is a great challenge to identify the deformation characteristics of overburden rocks in engineering practice.As an effective method to investigate the failure mechanism in mining,tunneling and other engineering activities,physical model testing has been widely applied in recent years(Zhu et al.,2010;Ghabraie et al.,2017;Ju et al.,2017;Moussaei et al.,2019;Yan et al.,2020).Taking the Gongjiafang landslide in the Three Gorges Reservoir as an example,Huang et al.(2014)established a large physical model capable of producing impulse waves to study the whole procedure of impulsive waves generated by cataclastic rock mass failure.In the three-dimensional(3D)physical model test by Li et al.(2014),the formation pattern of the excavation damaged zone around a tunnel was explored,together with numerical simulation and field measurements.Kang et al.(2017)established a large-scale physical model to simulate the failure mode of sudden massive roof collapse in the process of longwall coal mining.Numerous theoretical studies(Palchik,2003;Qian et al.,2003;Kuang et al.,2019;Sun et al.,2019;De Santis et al.,2020)and in situ investigations(Cheng et al.,2015;Yu et al.,2017;He et al.,2021)show that the overburden deformation characteristics under the influence of coal mining have a strong correlation with the position of key stratum.The concept of“key stratum”,initially proposed by Qian and Li(1982),refers to the stratum that controls the deformation and movement of the whole or part of the overburden.Kuang et al.(2019)established the timespace correspondence between the crack and movement of the key stratum and ground pressure in stopes based on a super-thick coal seam in the Datong coalfield.However,the three-zone theory for overburden movement,i.e.the caved zone,fracture zone,and continuous deformation zone proposed by Palchik(2003),has been widely accepted and applied to engineering practice.There is a specific correlation between the three-zone theory and the key stratum theory.For instance,the fracture zone in the three-zone theory is mainly formed below the key stratum,while the continuous deformation zone is formed above the key stratum.Based on the characteristics of the rock deformation above the coal seam,Cheng et al.(2017)established a zoning method using roof strata movements in the vertical and horizontal directions.These studies are mainly based on the crack degrees and displacement magnitudes of overburden obtained from monitoring.However,few analyses have been performed on the inherent characteristics of stressstrain changes in overburden caused by mining.Therefore,it is necessary to further explore the roof failure modes and mechanisms using fully instrumented physical model tests.

Another critical challenge is the monitoring technologies of overburden deformation,whether for laboratory tests or field monitoring.Conventionally,the primary monitoring technique of overburden includes strain gages,earth pressure cells,dial indicators,and fiber Bragg gratings(Zhang et al.,2009;Zhu et al.,2010;Du et al.,2014;Wang et al.,2019;Li et al.,2020).These instrumentation methods all rely on discrete measuring points and are prone to omit specific regions with large local deformation.Especially when key locations,such as stratum interfaces,are neglected,the judgment of the overall stability of the overburden rock mass may have considerable errors.For laboratory-scale tests on coal mining,the deformation process inside the model is difficult to capture.To visually explore the deformation characteristics of overburden strata,technologies have been developed,e.g.closerange photogrammetry(CRP),microseismic monitoring,and highdensity electrical resistivity tomography(HD-ERT)(Ren et al.,2004;Huang et al.,2018;Sun et al.,2018;Ye et al.,2018).However,for small-scale physical models,it is difficult to install too many sensors to comprehensively monitor the deformation characteristics.This is because too many sensors will not only change the original physical characteristics of the model but also make the monitoring process cumbersome.In recent years,distributed fiber optic sensing(DFOS)technologies have been successfully applied to monitoring the deformation of geomaterials(Klar et al.,2014;Cheng et al.,2015;Zhu et al.,2016,2022;Madjdabadi et al.,2017;Zhang et al.,2017;Chai et al.,2019;Wu et al.,2020).These pilot studies have opened a window for developing new monitoring systems of overburden deformation and provide a new technical means for studying overburden failure mechanisms.However,DFOS has its own drawbacks,such as low spatial resolution,special protection requirements,and complex behavior of deformation compatibility(Zhu et al.,2015;Zhang et al.,2016).Therefore,to better understand the mining-induced overburden deformation and failure mechanism,other monitoring technologies should be integrated into the investigation.Fortunately,the nondestructive measurements of CRP and the inversion of rock physical properties by HD-ERT are also effective methods for dynamic monitoring of mining-induced overburden deformation.They can be used to establish an integrated system for obtaining multisource strain and displacement data during the process of coal mining.

The primary purpose of this study is to investigate the deformation and failure characteristics of overburden rocks under the influence of coal mining,as well as the evolution process of cracks and separation layers.According to the field geological conditions of the Yangliu Coal Mine,China,a similarity model test is designed.During the test,the DFOS,HD-ERT and CRP techniques are used to comprehensively analyze the deformation characteristics of overburden rock.The formation and evolution of cracks and separations,as well as the failure mode and mechanism of the goaf roof,are investigated.The test results provide improved insight into the characteristics of stratum deformation in the coal mining process.

2.Study area

The study area is located in the Yangliu Coal Mine of the Huaibei Mining Group in Anhui Province,China(Fig.2a and b).The Yangliu Coal Mine area is approximately 9 km long from north to south and 3-9 km wide from east to west.The terrain in the mining area is flat,and the ground elevation is 25.98-28.26 m.The surface water belongs to the Huai River system.According to historical hydrological data,the current surface water is not harmful to coal mining.There are seven minable coal seams in the mine,of which coal seam 10 is the main minable coal seam,which is the focus of this monitoring.Its upper part is approximately 55 m away from the mudstone,and the lower part is approximately 60 m away from the top interface of the first limestone.The average thickness of the coal seam is 3.05 m,belonging to a medium-thick to thick coal seam.The roof of the coal seam is primarily composed of sandstone and mudstone,and siltstone is sporadically distributed.The original deposition of coal seam 10 is well developed.Due to magmatic intrusion and stratigraphic deposition,two layers of igneous rocks(diorite)with high hardness formed above the coal seam,of which the average thickness of the upper igneous rock is 25.9 m,and the thickness of the lower igneous rock is 47 m(Fig.2c).The burial depth of coal seam 10 and the bedrock coverage thickness are large.Therefore,the internal water of the loose aquifer has no impact on coal seam mining,and water production from sandstone fractured aquifers at the top and bottom of the coal seam is weak.Igneous rocks,faults,and fracture zones are relatively developed in the mining area.In this study,the simulated coal mining is the 10414 working face with a depth of 614.9 m,an average thickness of 3 m,and a strike length of 1080 m(Fig.2c).

According to a specific proportional relationship or similarity theory,the field prototype can be simulated via the experiment of similar materials(Zhang et al.,2011,2017).In the similarity test of simulating rock strata and coal seam excavation using physical materials,the simulated geological conditions and physical quantities must comply with certain ratios,including the length similarity ratio(αL),stress similarity ratio(ασ),density similarity ratio(αγ)and time similarity ratio(αt)(Sun et al.,2021):

whereLis the generalized length,which can be distance or displacement;γ is the density;σ is the stress;Sis the strength;cis the cohesion;Eis the elastic modulus;and the subscripts R and S represent the prototype and model,respectively.

Considering the actual test conditions,the geometric(length)ratio of the similarity model test was determined to be 1:200.Then,the time ratio was defined as 14,and the stress ratio was defined as 333.To simulate the overlying formation pressure,the equivalent load pressure applied to the model was 18 kN.

3.Instrumentation

To comprehensively study the deformation process of overburden rocks during coal seam mining,three measuring technologies were used in the model test,including DFOS,HD-ERT,and CRP.

3.1.Pulse pre-pump brillouin optic time domain analysis(PPPBOTDA)technology

BOTDA technology,one of the DFOS technologies,is based on the Brillouin optic time domain analysis of stimulated Brillouin scattering(Horiguchi and Tateda,1989).The principle of BOTDA is to inject continuous light and pump light into opposite ends of the fiber optic to form stimulated Brillouin scattering through a change in the pump laser pulse structure.Stimulated Brillouin amplification occurs when the Brillouin frequency shift of a region of the fiber optic is the frequency difference between continuous light and pump light.Therefore,the frequency difference corresponding to the maximum Brillouin that gains on each segment of the fiber can be determined by continuously adjusting the frequency of the detection pump light and detecting the continuous light power output at the other end of the fiber.According to the linear relationship between the Brillouin frequency shift and strain or temperature,the strain or temperature of each point along the fiber can be determined by

Fig.2.Location and geological conditions of the Yangliu mine.

whereTis the ambient site temperature,T0is the initial temperature,?vB(ε)／?ε denotes the Brillouin frequency shift-strain coefficient,and?vB(T)／?Tstands for the Brillouin frequency shifttemperature coefficient.

If the temperature variation during strain measurements is negligible,Eq.(5)can be simplified as

When one-dimensional(1D)deformation occurs,the strain measurements can be easily converted into displacement by

whereΔLis the relative displacement between two points,and ε(x)is the strain of a sampling point betweenl1andl2.

PPP-BOTDA is an improved version of the conventional BOTDA technology.Before the sensing pulse light enters the system,the pre-pump pulse light with a wider pulse width and lower power is injected into the system.The gain spectrum and phase shift spectrum of the Brillouin signal are obtained by changing the frequency of continuous light to realize the accurate measurement of temperature and strain.

To verify the performance of distributed strain sensing using PPPBOTDA,a series of pullout tests were conducted on sand-embedded fiber optic cables.Fig.3 shows the setup of the pullout test,which was performed in a test box of 1.5 m in length,2 m in width,and 2 m in height.The 2 m thick foundation was prepared using a sand raining technique.When the mid-height of the test box was reached,the fiber optic cable was placed on the sand carefully.The relative density and water content of the sand were 23% and 1.92%,respectively.The length of the sand-embedded cable section was 1.2 mm,whereas a 2 mm long free section was prepared to minimize boundary effects.The cable head was fixed on a special clamp having two rubber pieces.The clamp was then connected to a tensile force gage for pullout force measurement.The other end of the cable was always kept loose during testing.Finally,both ends of the cable were connected to a 6050A PPP-BOTDA interrogator.

During testing,pullout displacements were applied to the cable tip in stages.One strain measurement was recorded at each displacement step of 1 mm.A total of 11 steps were applied in the test.The details of the test setup can be found in Zhang et al.(2020).

The strain measurements at each stage are shown in Fig.4.The test results show that the strains were almost constant at the 0-0.2 m segment(free length),and those of the 0.2-1.4 m segment can be fitted by linear functions.An obvious progressive failure phenomenon was observed during the pullout of the cable.It is noted that the strains emerged at the cable tip and then propagated towards the cable end with increasing pullout displacement.The mobilized length gradually increased to 1.2 m with increasing pullout force.At a displacement of 8.7 mm,the strains were fully mobilized,and the linear distribution of axial strain was observed.Because the differentiation of strain with respect to location is related to interfacial shear stress,the fiber optic cable monitoring data can be used to characterize the stress-strain relationship of the cable-sand interface.The results indicate that during the cable pullout process,an ideal elastoplastic model can be used to describe the interfacial behavior(Zhang et al.,2016).

Fig.3.Setup of the pullout test on the sand-embedded fiber optic cable.

3.2.HD-ERT

The purpose of HD-ERT technology is to survey the resistivity distribution of the measured medium.Similar to the conventional resistivity method,HD-ERT combines vertical resistivity sounding and apparent resistivity profile measurements.Based on the conductivity difference of different electrolytes,an artificial direct current is supplied to the electrodes to measure the resulting potential difference to retrieve information on the medium heterogeneity and electrical characteristics(Xu et al.,2021).In practical applications,the inversion results and the accuracy and resolution of the final results are closely related to the collection parameters,including device type,electrode spacing,and electrode number.In this study,the Wenner configuration(dipole power supply-ABM method)layout scheme and 64-bit electrode data acquisition mode were adopted.Electrodes A and B(current electrodes)are power poles with an electric current(I),and the other two electrodes M and N(potential electrodes)are employed to record the resulting potential difference(ΔU=UM-UN).Thus,the conductivity of the medium,which is usually reflected by resistivity,can be easily calculated from the measured current and voltage by

where ρsis the apparent resistivity andKis a coefficient related to the electrode configurations.

3.3.CRP

Fig.4.Strain distribution of the fiber optic cable under pullout.

CRP is mainly based on digital image processing technology to monitor the deformation characteristics of objects and capture surface features(Zhang et al.,2017;Xu et al.,2020).To perform image numerical calculations,the following observation model was established(Fig.5).Measurement pointsA,B,CandDare four fixed points on the same horizontal plane with the same spacing.The camera system was placed at the optimal observation pointPfor monitoring.The horizontal projections ofPA(andPB)andPC(andPD)wereL1andL2,respectively.The angle betweenLandL1is α,and β is the angle betweenLandL2.

Suppose that the perpendicular angles of pointsAandBare θ and γ,respectively,and the vertical distance between pointsAandBisH;then,we have

Similarly,if the perpendicular angles of pointsCandDare φ and ω,respectively,then we have

Fig.5.Observation model for CRP.

For any measurement pointM(x,y)on the model,the horizontal observation angles are α and β,and the vertical angle is δx.According to the triangle sine theorem,we can obtain

wherelis the distance from pointOAto pointOB,Lis the distance from pointPto pointO,∠Ais∠PAB,and∠Bis∠PBA.

Therefore,the horizontal distancelxbetween any measurement pointMand fixed pointAis

In addition,the vertical distancehxbetween any measurement pointMand fixed pointAcan be expressed as

where∠Cis∠DCP,H0is the distance from pointAto pointOA,and δxis∠MPO.

Thus,the vertical displacementΔHand horizontal displacement ΔUof pointMcan be obtained by observing the changes in coordinates before and after mining as

wherehis the distance from pointAto pointN,xiis the abscissa value of pointM,andx0is the abscissa value of pointF.

3.4.Calculation of overburden deformation using monitoring data

Accurate measurement of the stress states of geomaterials in physical model tests has always been a bottleneck.This is because both the normal stress and the shear stress should be monitored.However,the monitoring of shear stress is still a challenge.According to the conventional force analysis of an object,the maximum strain usually occurs at the surface layer of the material component.Therefore,based on the analytical relationship between displacement and strain,the overlying rocks can be considered to be elastic materials before failure.The evolution law of the stress field can be evaluated through strain measurements.Assuming that a componentABCis deformed toA′B′C′(Fig.6),the corresponding relation between displacement and strain can be expressed as

where εxis the linear strain in thex-axis direction,εyis the linear strain in they-axis direction,γxyis the shear strain,anduandvare the displacement variables along thex-axis direction and they-axis direction,respectively.

Thus,according to the obtained strain component,the principal strain can be calculated by

where εmaxand εminare the maximum and minimum linear strains,respectively.

Based on the coordinates before and after the deformation of three adjacent displacement measuring points arranged perpendicular to each other in the model,the horizontal and vertical displacements caused by deformation can be obtained by CRP results(Eq.(14)).Under the condition that the displacement data are known,the main strain at each measuring point can be calculated by Eqs.(15)and(16),and then strain values of all the measuring points in the model can be obtained.Therefore,the deformation characteristics inside the model can be characterized by monitoring the surface displacement variation of the model.By capturing the strain characteristics of the model during deformation,the displacement can be obtained by integrating the strain monitored by the fiber optic cable using Eq.(7).Therefore,the spatiotemporal evolution characteristics of mining-induced deformation are established by combining the monitoring of surface displacements and internal strains.

Fig.6.Relation between the displacement and strain.

The evolution of overburden deformation characteristics caused by mining is closely related to strata movement.The key strata,which are mostly hard and thick rock layers,are the bearing structures that control strata movement.Because the key layer has high strength,the deflection after mining can be equivalent to an elastic foundation beam.Assume that the load on each stratum is uniformly distributed,and there aremlayers in the overlying strata of the coal seam.Then(n≤m)layers overlying the coal seam are in synchronous deformation from bottom to top,the thickness of each stratum ishi(i=1,2,3,…,m),and the corresponding volume force is γi(i=1,2,3,…,m).As thenlayers can be deformed synchronously,considering the weak shear force on the strata in the layered rock mass,it can be obtained from the beam theory by

whereMiis the bending moment of thei-th layer,Eidenotes the elastic modulus of thei-th layer,andIirepresents the moment of inertia of thei-th layer.In terms of the rectangular moment of inertiaIi=bh3/12,bstands for the cross-section width of the beam,andhis the thickness of the layers.

For the first floor beam,the bending momentM1can be expressed as

whereq=Eε is the deadweight load of allnstrata.

According to the bending beam principle,the following expression can be obtained(Qian et al.,1996):

Thus,the locations of the key stratum can be inferred from the strain measurement ε and the density γi.

4.Model test

4.1.Test materials

According to the similarity principles,mixtures of sand(S),lime(L)and gypsum(G)with various proportions(S:L:G)were determined to accurately simulate the field geological conditions.In this test,sand with a particle size of 0.5-2 mm was used as the primary aggregate of similar materials.Lime and gypsum were used as the jointing and cementitious materials,respectively.Mica powder,as a simulation material to weaken the connection between rock strata,was used to separate strata with different lithologies.Considering the geological conditions of in situ strata,physico-mechanical parameters of similar materials with different specifications and ratios can be obtained by uniaxial compression and tension tests(Table 1).Therefore,the optimal proportion of S:L:G for simulating different lithologic strata can be determined via multiple tests.Through the transformation of the similarity ratio,the material properties used to simulate each rock stratum were obtained,as listed in Table 1.It should be noted that taking mudstone as an example,the ratio of sand,lime and gypsum is 10:0.5:0.5,indicating that sand accounts for 10/11 of the total weight,while the remaining two materials account for 1/11 of the total weight(lime is 1/22 and gypsum is 1/22).The water consumption is 10% of the total mass of sand,lime and gypsum.

4.2.Model construction and test procedures

Based on the actual engineering-geological conditions and the principle of similarity transformation,a physical model with a length of 300 cm,a width of 25 cm and a height of 160 cm was designed.Similar materials for the in situ strata were prepared and configured according to the parameters presented in Table 1.The installation of distributed strain sensing cables was performed during model construction.In this process,similar materials should be in full contact with the fiber optic cables,and the cables should be carefully protected during the compaction of the strata.The procedure for preparing the physical model is as follows(Fig.7):

Table 1Physical parameters,composition materials and related properties of simulated strata.

(1)Bonding and fixing the fiber optic cables,i.e.fixing the cables at the bottom hole of the model frame with epoxy resin(Fig.7a);

(2)Pre-stretching the fiber optic cables.A horizontal pole was placed above the top of the model for pre-stretching of the fiber optic cable to ensure that these cables remained vertical during the similar material filling process to avoid cable relaxation and bending;

(3)Preparing similar materials.The weight of the materials was determined according to the similar proportions of each rock stratum.Then,the similar materials were poured into a mixer machine to be evenly mixed;

(4)Compacting the rock strata.According to the design thickness of each rock layer,the prepared similar materials were filled in sequence and compacted with a hammer,and the flatness of the compacted rock layer was checked by a spirit level;

(5)Fixing the protective formwork.Install and fix the formwork every 20 cm of height of the similar material filling to ensure the stability of the model.

(6)Sprinkling mica.After a rock stratum was built,mica powder was evenly sprinkled on the surface of the rock stratum to ensure good stratification between strata.The constructed model was cured for 14 d at room temperature(20°C)for the setting and stabilization of similar materials.

During the model test,the coal mining process was simulated by stepped excavation,and the deformation of the overlying rock strata was measured continuously.To eliminate the boundary effect of the model test in simulating coal seam mining,30 cm“pillars”were reserved on the left and right sides of the model.When simulating coal mining,an“open-off cut”was first made on the right side of the model(at 270 cm of the scale).Then,the coal seam was gradually excavated to the left by 30 cm away from the right side of the model,advancing 5 cm each time.The DFOS,HD-ERT,and CRP data were collected simultaneously by the end of each excavation step.For the three aforementioned measurements,the working principle and layout technique are described in the following sections in detail.

4.3.Monitoring scheme

To acquire the spatiotemporal distribution of overburden deformation,a monitoring scheme was designed.PPP-BOTDA technology was used to perform distributed strain sensing along with fiber optic cables,which can be used to investigate the deformation and failure evolution of strata during coal mining.HDERT technology was employed to invert the stress change characteristics of overburden rocks.CRP technology was adopted to characterize the displacement field at the model surface.The layout of the sensors is shown in Fig.8.

Considering that the deformation and failure of overlying rocks were primarily affected by vertical compression,tension,and shear deformations,BOTDA strain sensing cables were vertically installed to form three U-shaped loops(i.e.A12,B12,and C12)in the model.A1(at the position ofx=240 cm)and A2(at the position ofx=220 cm)sensing cables were deployed at the right side of themodel with a spacing of 20 cm.B1 and B2 sensing cables were deployed in the middle of the model atx=160 cm andx=140 cm,respectively.C1 and C2 sensing cables were deployed on the left side of the model at the locations ofx=80 cm andx=60 cm,respectively(Fig.8).The three U-shaped fiber optic cables were connected together to achieve the dynamic monitoring of model deformation.For the HD-ERT investigation scheme,one horizontal and two vertical monitoring lines were adopted.A 32-channel survey line with a measuring point interval of 3.5 cm was arranged horizontally(at the depths of 76 cm),and another two with the same interval(3.5 cm)were vertically arranged to form a 64-channel survey line(at the location ofx=110 cm andx=170 cm).In terms of CRP,four displacement measuring lines were set at depths of 61 cm,71 cm,91 cm,and 111 cm.The interval between two measuring points on the measuring line was 5 cm(Fig.8).

Fig.7.Construction of the physical model:(a)Bonding and fixing the fiber optic cables,(b)Pre-stretching the cables,(c)Preparing the similar materials,(d)Compacting the rock strata,(e)Fixing the protective formworks,and(f)Sprinkling mica.

During the model test,a 6050A PPP-BOTDA interrogator,a 64-channel parallel electric device(CPED),and a single lens reflex(SLR)camera with a fixed focal length were used to record the measurements.The measurement parameters of these instruments are given in Table 2.The advantages and limitations of the three technologies are given in Table 3.

Table 2Instruments used in the model test and their main measurement parameters.

Table 3Advantages and limitations of the three technologies.

5.Results and analysis

5.1.DFOS results

In the model test,the simulated coal seam was excavated for 5 cm each time.In Fig.9,the model remained stable in the process of advancing at the initial mining stage according to the monitoring results of the six vertical strain sensing cables(A1,A2,B1,B2,C1,and C2).The stress change in the fine sandstone formation overlying the coal seam roof was not obvious,and the bending and deformation zone was small.With the continuous advancing mining of the coal seam,the deformation amount of the fine sandstone group and mudstone group above the coal seam roof gradually increased,and the stress inside the model redistributed,which contributed to the subsequent deformation and failure of the overburden.Here,the tensile strain value was defined as positive(tensile strain＞0),and the compressive strain value was defined as negative(compressive strain＜0).The spatial sampling interval of the strain points of the sensing cable was 0.1 m.

Fig.9a shows that the strata under a lower igneous rock(at depths of 62-84 cm)presented a trend of compression with the continuous excavation of the coal seam,and the strains constantly increased,with a maximum value of approximately 7800×10-6.The upper strata of the lower igneous rock were first compressed.With continuous mining,cracks and separation layers formed in the lower strata under the lower igneous rock.At this time,the tension phenomenon occurred.In the continuous advancing process,the separation layers in the overlying rock body kept closing,and the upper rock mass covering the lower igneous rock gradually indicated the compression trend,which was more obvious in survey line A2(Fig.9b).For survey lines B1 and B2 in the middle of the model,the stress deformation mechanism was the most complex;however,on the whole,it still presented a certain trend(Fig.9c and d).That is,the deformation of rock layers covering the lower igneous rock was different from that of the rock layers under the lower igneous rock due to the existence of a key stratum.The complex deformation was caused by the formation and closure of cracks or fractures in the separation layers.Tensile strains occurred during the formation of the separation layers,while compressive strains occurred in the process of separation closure.The formation and closure of the separation layers during mining was a dynamic evolutionary process,and it was also a process of the rock strata being subjected to tensile bending and shear failure,which formed the collapse debris of the rock strata.As far as survey lines C1 and C2 were concerned,they were located near the model’s left side,which was less affected by mining and had the same deformation trend(Fig.9e and f).In the whole experimental process,the adjacent area of C1-C2 was in a compression state.The compressive stress far from the mining position was less than that near the mining position,revealing that the transfer of deformation gradually attenuated from right to left.When mining was completed,the key strata lost the support of the lower original stratum and bore more pressure from the upper strata.According to the data of the six survey lines,it can be concluded that the lower igneous rock was the key layer(located at 62-84 cm of the model scale),which was the dividing line of stress deformation of the rock mass and controlled the deformation of overburden caused by coal mining.The calculation results of Eq.(19)also supported this result.According to the three-zone theory,the key layer position determined by the monitoring results was the boundary between the fracture and continuous deformation zones.The concentrated crack development area was mainly located 30 cm above the coal seam roof.In other words,the height of the well-developed fracture zone(also known as the maximum height of the water-flowing crack)was approximately 30 cm at the roof of the coal seam.This conclusion was supported by the empirical formula that appeared in Cheng et al.(2017)for determining the height of the fracture zone.In addition,by calculating the settlement displacement of the coal seam roof using Eq.(7),the maximum settlement was approximately 4 cm.

Fig.8.Layout of the monitoring system.

5.2.HD-ERT results

In this study,HD-ERT prospecting was used to retrieve the characteristics of the medium through the distributioncharacteristics of the resistivity.The resistivity distribution at the initial stage was different due to the complex properties of the simulated rock materials and the varied degrees of compaction at different positions.The distribution characteristics of the resistivity at the initial stage are shown in Fig.10a.Before mining,the overlying strata were not disturbed and had good integrity,resulting in a low resistivity value of 50-400 Ω m.In addition,as the rock stratum approached the bottom of the model,it became denser under the action of overlying rock strata and the effect of gravity,and the resistivity value gradually increased.After mining,the rock stratum deformed,resulting in changes in its density.Therefore,the resistivity distribution of the overlying strata showed an increasing trend.The measured resistivity distribution is shown in Fig.10b.By comparison,it can be found that the resistivity of the overlying strata changed significantly under the influence of mining.The overall resistivity increased by 2-5 times,which was mainly shown in the area of 62-32 cm above the roof of the coal seam.The results show that the influence range of crack or fracture concentration development caused by coal mining was also located 30 cm above the coal seam,which was consistent with the results from the fiber optic test.

Fig.9.Overburden strains measured by six strain sensing cables:(a)A1;(b)A2;(c)B1;(d)B2;(e)C1;(f)C2.

For the vertical electrical survey lines,Fig.11 shows the resistivity inversion results at different advancing positions.In the initial state,the overburden was undisturbed and well consolidated,generally showing a low resistance value of approximately 50 Ω m(Fig.11a).With the continuous advance of coal mining(advancing to 50 cm),the resistivity distribution changed,and the resistivity increased significantly at the coal seam roof(Fig.11b).This is because the rock stratum within the control range of the electrical survey line was subject to the action of advance bearing pressure and certain shear action.The resistivity value of the local overburden in the lower part increased significantly.Nevertheless,the overall resistivity value was still low,indicating that the rock stratum was not greatly deformed,but cracks appeared in the overburden of the coal seam.As the mining of the working face gradually approached the electrical survey line(Fig.11c),the rock strata within the monitoring range were greatly affected by mining,and the resistivity values of the upper and lower overburden layers increased significantly.Among them,the resistivity of the mudstone varied the most,and the resistivity of the rock stratum between the two layers of igneous rock can achieve 800 Ω m.This indicated that mudstone with good plasticity and low strength was easily affected by mining,resulting in separation cracks.Compared with the upper and lower igneous rocks,the resistivity of the other rock strata generally increased,which was more obvious in Fig.11d;at this time,the working face was advanced to 2 m.The deformation degree of the overburden strata reached the maximum and gradually stabilized.Caving and filling of the roof strata of the coal seam increased the resistivity.In addition,the resistivity of the mudstone under the upper igneous rock also changed greatly,up to 800-900 Ω m locally,forming large separation cracks.Overall,based on the comprehensive analysis of resistivity and lithologic distribution,the resistivity of the upper and lower igneous rocks had a small change during the whole excavation period,indicating that the two layers of igneous rocks had good integrity with a few cracks.Meanwhile,the resistivity of the rock strata between the two igneous rocks gradually increased.It can be concluded that the upper and lower igneous rocks had a specific control effect on the mining-induced overburden deformation and were the locations of the key strata.The high-resistivity zone in the lower layer of igneous rock extended upwards in the progress of excavation.Then,the dynamic development process of the waterconducting fracture zone interface can be described effectively.From Fig.11d,the development interface of the water-conducting fracture zone was located 30 cm above the coal seam,which was consistent with the results in Fig.10b.

5.3.CRP results

Fig.11.Inversion of resistivity distribution for vertical survey lines under different advancing distances.

Fig.12.Displacement variations of the four survey lines during coal mining:(a)Survey line 1,(b)Survey line 2,(c)Survey line 3,and(d)Survey line 4.

To analyze the displacement variation during mining via CRP technology,each 20 cm of mining advance was taken as an observation point,and the mining was advanced 12 times in total.Fig.12 shows the dynamic change in the subsidence of each survey line caused by coal mining.The movement range of overburden rock expanded with increasing burial depth.With the advancement of mining,the influence of coal mining on the front deformation became significant.The effect of mining on the deformation of the back,i.e.the overlying strata in the goaf gradually reduced.The influence of the deformation close to the model at 270 cm was almost unchanged.After the completion of the mining,the maximum displacement and deformation of the overall model occurred at 155 cm,which was in the middle of the model goaf.CRP monitoring lines 1 and 2(Fig.8)were barely affected by mining because they were located far from the coal seam.Significant displacement changes only occurred when the excavation reached a certain distance.The displacement variations along survey lines 3 and 4 were similar because the two survey lines were close to each other,and both were located between the coal seam and the igneous rock.As the position of survey line 4 was closer to the coal seam,the corresponding rock deformation was larger than that of survey line 3.With the continuous mining of the coal seam,the increase in the overall displacement was accompanied by a small mutation,in which the local displacement decreased first and then increased slightly.This is because the stress state of the rock strata changed,and the previously generated separation layers were gradually closed under pressure.New separation layers formed due to retension,indicating that there was still a closed intermediate process from the formation to the failure of separation layers.The maximum subsidence displacement of the coal seam roof was approximately 4 cm.

5.4.Overall deformation characteristics

By combining the three monitoring techniques and performing a comprehensive analysis,the deformation characteristics of the overburden during coal mining can be well understood.To visually demonstrate the deformation characteristics of the overlying rock,the strain contours were depicted from the fiber optic monitoring data by Kriging interpolation.The same method was adopted to obtain the displacement contours based on the data of the four CRP survey lines.At the early stage of coal mining,the deformation of the whole overlying rock was not significant(Fig.13a),and only slight compressive strains occurred in the middle of the model and above the goaf(Fig.13b).In terms of displacement,the strata above the goaf showed a small amount of settlements(Fig.13c).When the coal seam advanced to 170 cm,the first periodic pressure was generated with a weighting step of 28.2 cm,leading to the collapse of the old roof(Fig.13d).The strain contour indicated two significant compressive stress areas.The height of compressive stress area A behind the working face was between 90 cm and 130 cm,which was more obvious than the compressive stress area located at 30 cm above the coal seam(Fig.13e).From Fig.13e,the compressive deformation value was relatively large,while the influence area of compression deformation was relatively small.The rock layer in this region was broken,and the process of compression deformation had not yet been stabilized.The rock mass had a significant degree of elastic compression,which contributed to the overall settlement.Specifically,large displacements occurred in the goaf from 220 cm to 250 cm(Fig.13f).In the process of coal seam advancement,the redistribution of stress should be a dynamic evolution process.Due to the development,formation,and failure of the crack separations in the mining process,the tensile and compressive stress areas changed constantly,forming stress concentrations in some areas and resulting in large displacement deformation.In particular,this phenomenon was more pronounced when the working face advanced to 90 cm(Fig.13g-i).For the overall deformation characteristics in the mining process,the compressive stress area A kept expanding laterally with the advancing of the working face.The location of the newly generated compressive stress in area A tended to rise,while the stress variable decreased.The strain was transmitted to the top of the model,and the strain increased.Therefore,area A was a caving zone,and area C was a bending zone.Tensile strain area C gradually moved towards area B forming at the upper part of the model,which was considered the main cause of ground subsidence during underground mining(Fig.13j-l).

The depth of area C was 60-80 cm,resulting in a large tensile strain(approximately 6000×10-6).Therefore,it can be concluded that there was large subsidence in this area without generating cracks,and this layer can be considered a key layer.

5.5.Analysis of the deformation mechanism

In the mining process of the coal seam,the overburden deformation is caused by stress redistribution,which is primarily manifested as the generation of cracks in the overburden as well as the formation,development,and failure of separation layers.To analyze the evolution mechanism of cracks and separation layers in the overburden rock,the monitoring data corresponding to the 170 cm working face were selected(Fig.14a).The propagation directions of cracks initiated at the working face and the mining starting location were different,and they basically developed in a symmetrical pattern,forming penetrating cracks that intruded into different rock layers(Fig.14b).Moreover,the crack inclination angle α increased gradually from the edge of both sides to the middle of the model,forming a transverse crack network.The above results explained why the overburden deformation zone affected by mining had a saddle shape.This is further explained as follows:as the volume of the goaf increased during mining,the overburden was subjected to large loads transmitted from the upper strata.At the same time,localized shearing emerged at the two boundaries of the goaf.On the one hand,the formation of new cracks intensified the stress concentration at the crack tip,forming penetrating cracks.On the other hand,the upwards shear force induced increased tensile stresses in the separation layers.Therefore,the growth and evolution of cracks and separation layers were affected by the coupled effect of shearing and tension(or compression),which was clearly demonstrated by the fiber optic strain measurements(Fig.13e).Furthermore,when the initial shearing stress was stronger than the tension and compression,the separation layers developed,and the separation cracks opened gradually.With the increase in the mining face,the separation layers gradually closed when the tension and compression were stronger than shearing.This phenomenon was also supported by the CRP and BOTDA results.During this process,the separation layers were destroyed by the coupled effect of shearing and tension(or compression).

Fig.14.Crack and separation of overburden rocks and the related deformation and failure mechanism:(a)Photograph of overburden deformation when mining to 170 cm,(b)Photograph of cracks and separation layers,(c)Failure process of a rock layer dominated by bending and tension,(d)Failure process of a rock layer dominated by overall shearing,and(e)Failure process of a rock layer dominated by shearing and sliding.

For the evolution mode of cracks and separation layers dominated by tension and bending(or compression),the upper rock strata above the goaf bent under the overburden pressure when the initial goaf formed.With the increase in goaf volume,microcracks appeared above the working face in the middle of the goaf.When the goaf reached a certain length,the rock layers experienced crack failure(Fig.14c).That is,with the advancement of mining,the overburden in the goaf reached the maximum breaking distance of the basic roof.Under the action of overburden load and self-weight,the rock stratum bent and broke down at a certain angle.Compared with hard rock,soft rock has a certain degree of plasticity,resulting in relatively stable separation layers under minor bending and tension.As shown in Fig.14b,there are continuous and good separation cracks in the mudstone distribution area.

Shear slip failure usually took place near the excavation and can be divided into two types.One is the crack failure caused by a shear action,which generally occurs under the following conditions:(1)the rock strata have high strength and stiffness;and(2)a large shear force is exerted on the rock strata.Due to the high strength of the whole rock strata,cracks initiated at both ends of the goaf.Under the action of strong shear stresses,the entire rock strata will be directly cut off,as shown in Fig.14d.At this time,significant separation cracks occurred in the overburden rock.This kind of failure mostly occurs in brittle surrounding rocks in deep mining sites,which is also the main cause of coal mining disasters.The other failure mode is for the rock strata,which have already undergone bending and tensile deformation.Influenced by the free working face of the goaf and mining disturbance,transverse and vertical cracks are formed in the overburden rock.After the cracks expand,close and develop again by bending and tension,the rock layer is chopped and slipped due to compression and shearing action,which is characterized by irregular small collapse fragments(Fig.14e).At this stage,the overlying strata are often damaged slowly and periodically.Overall,through this test,there are three modes of overburden deformation and failure caused by coal mining:(i)tensile and bending failure,(ii)overall shearing,and(iii)periodic shearing and sliding.This provides a reference basis for understanding the overburden failure mode caused by mining in practice.

Notably,the type of coalfield geological structure,the thickness of the overlying rock layer and lithology properties have an impact on the failure mechanism of the overlying rock during mining.In particular,when the mining depth of the coal seam is large,the mining rate is fast,and the thickness of the coal seam is large,the damage degree of overlying rock increases,which will aggravate the threat to the safe production of the mining,and it needs to be prevented in advance.

6.Conclusions

A similarity model test was conducted to investigate the characteristics of overburden deformation during underground coal mining.The spatiotemporal evolution process of the deformation field,as well as the formation,development and failure mechanism of cracks and separation layers during coal mining,were comprehensively analyzed by using DFOS,HD-ERT,and CRP monitoring techniques.The main conclusions are as follows:

(1)In the process of coal mining,the deformation of the overburden strata was a dynamic process,and the concentration areas of tensile and compressive stresses continuously changed,resulting in the corresponding evolution of separation layers and cracks.The whole deformation influence area gradually expanded and finally presented a saddle shape.Symmetrical stepped cracks were generated at the starting and ending locations of coal mining due to the coupling of shearing and tension(compression).

(2)The high-strength rock strata played a crucial role in controlling the deformation and failure mode of the overburden rock.The upper rock mass deformation of the key layers was small,while that in the lower part was relatively large.The maximum displacement of the coal seam roof was approximately 4 cm according to the monitoring data,which occurred in the middle of the goaf.

(3)There are three failure modes of overburden separation layers,including bending-tension,overall shearing,and periodic shearing and sliding.When the brittle rock was subjected to the coupled effects of high-intensity bendingtension and shearing-sliding,the whole rock mass collapsed and formed large rock blocks.This kind of failure mode had a significant and rapid impact on coal seam mining.Therefore,special attention should be given to such mechanisms in engineering practice,and rational prevention and treatment measures should be carried out in a timely manner.

(4)The deformation obtained by the BOTDA strain sensing cables is consistent with that calculated by the CRP technique.The deformation distribution characteristics also fit well with the resistivity variation inverted from HD-ERT technology.The development height of the water-flowing fracture zone detected by BOTDA and HD-ERT technologies was approximately 30 cm above the coal seam roof.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

We acknowledge the funding support from the National Natural Science Foundation of China(Grant No.42225702),the Central Government Guided Local Science and Technology Development Fund(Grant No.226Z5404G),and the Natural Science Foundation of Hebei Province,China(Grant No.D2022508002).