Physical modeling of failure process of the excavation in horizontal strata based on IR thermography

Available online at www.sciencedirect.comMININGScienceDirectSCIENCE ANDTECHNOLOGYELSEVIERMining Science and Technology 19 (2009) 0689 0698www.elsevier.com/loeate/jcumtPhysical modeling of failure process of the excavation inhorizontal strata based on IR thermographyHE Man-chao'2, GONG Wei-li'2, LI De-jian1.2, ZHAI Hui-mingh2'State Key Laboralory for Geomechanics and Deep Underground Engineering, Bejing 100083, China"School of Mechanics and Civil Engineering, China University of Mining & Technology, Beijing 100083, ChinaAbstract: In order to capture the mechanism of roadway instability in deep mines, a new approach of Physically Finite ElementalSlab Assemblage (PFESA) is proposed in order to construct a large- scale physical model simulating the geologically borizontalstrata. We carried out physical modeling on the deformation and failure processes of roadways subjected to a plane loading scheme.Our laboratory tests were based on work which incorporated infrared (IR) detection, IR radiation temperature (IRT) statistics, imagefeature extraction and 2D Fourier transformation, from resulting thermographies. The IRT characterizes the mechanical responsesfrom the roadway after loading with two stages, i.e., IRT evolving at higher ievels corresponded to shallow mining (!!之Sf(,)(4)tical load was designed as a stepwise loading while，MN台台the lateral load was kept at an amplitude equal to thevertical load of 0.8 and 1.4 MPa (corresponding towhere MxN= 120x160 is the total number ofmining depths of 350 and 500 m) and then kept un-pixels in one frame of the image series. The meanchanged until the vertical load was stepped up to 5value < f> represents the transient IR power in-MPa (with respect to mining depths of 2200 m). Intensity emitted from the target at the time of sampling.Fig. 4, the solid line denotes the vertical stress σThe sample mean series < fi : k were calculatedand the dashed line the lateral stressσz. At everyusing Eq.(4) ftom the IR image series f(i,j)loading step, the stress level is indicated by the stress(k=0, 1, 2.. is the number of IR images), and thevalue in the vertical stress curve and the digits in theparenthesis next to the stresses indicate the corre-variable k can be converted into the discrete timesponding mining depth. For characterization of thevariable t by multiplying k with the imaging fre-rock behaviour under loading, the lateral pressurequency used in the test. The first frame of the IR im-coefficient was defined byλ=σ2/σ.age was taken when the model was at its initial state, .used as the basic reference for the calibration of thetemperature changes of the following images. LetVertical。6 MPe(2600 m)T=g,k=0,1, 2.，the normalized T is---- Horizontal o,5 MPa; (200 m)defined by4一4 MPa (1700m)工-元(5)T。-Tmx3 MPa (1300m)2.5 MPa (1000 m)In the following context, T is also referred to as1.8 MPa (800m) σ2-the“Infrared Radiation Temperature (IRT)” which1.4 MPa[ (500m)actually represents either the IRT variations relative0.8 MPa (350 m)SlageAStage B _to the initial state of the model or the energy dissipa-tion level relative to the maximum IRT mean value100102030405060708090T. Hence Tp (or IRT) was used in our study as aTime (s)measure of the mechanical responses for the rockFig. 42D loading schememass subjected to external loading. The characteris-The roof pressure on the roadway due to the stratatics of the IRT are presented in Fig. 5; Fig. 5a is theon the model itself can be calculated byσ= 2rhtemporal evolution of the IRT. According to the pro-gressive structural failure of the physical model with(kPa), where Y (kN/m') is the volumetric weight ofrespect to the loading steps, some special pointsa particular stratum (see Table 2) and h, (m) the(points of interest, POI) were selected on the IRTcurve marked by the capital letters A0-A5 andthickness of the stratum. The overlying strata for theB0-B8 respectively. Table 4 is the relation among theroadway on the physical model are composed ofmining depths, stress level and POIs. Fig. 5b showsstrata 1, 2 and 3 (see Fig. 1); thus the roof pressurethe photographs for these POIs.for the model was calculated as 8.99 kPa, much lessThe IRT curve in the regime A1-A5 correspondsthan the simulated vertical load o, by one order ofto the loading stage A for shallow depth (miningmagnitude. Its effect on the applied horizontal anddepth<500 m; h=σ, 1σ,=1). During this stage, thevertical stress is small and was therefore, not takenconfining pressure was uniform and relatively low.into account in the loading scheme of the test.The IRT evolved at a relatively higher level withsharp rises marked by A1-A5, depicted as related4 Characterization of IRTstructural changes in the photographs in the first rowin Fig. 5b. During stage A, the model deformedSince the IR image is the visualization of the in-gradually and slightly, but no internal breakage orfrared radiation temperature (IRT) distribution on thecollapse occurred. Therefore, stage A is denominatedviewed surface, its pixel matrixf(i,j) (iandj are"steady deformation stage”. A1 corresponds to thethe number of pixels, i=1, 2, .. M;j=l1,2, ... M) ismoment when the first step loading (0.8 MPa) wasthe IRT data set at a specific instant. From a stochas-appl中国煤化工second step load-tical point of view, the IR image matrix f(i,j) caning (; A4 are the localbe regarded as a 2D randomized field; its samplepeaksYHCNMHGthefactthatthemean can be calculated by:abrupt increase in the external load will result in anintense release of the elastic energy induced by the694Mining Science and TechnologyVol.19 No.6effect of microscopic fracturing. A2, A3 and A5 arelevel was much smaller than that in stage A and itsalso local IRT peaks, respectively within the first andprogress, as a function of time, exhibits a quasi-cy-second loading step, which are not caused by the in-clical fluctuation pattern with multiple local peaks.crease in load but by the time-dependent nature of theThe loading steps for σ and lateral pressure coeffi-engineered soft rocks.cienth for B1-B7 were: B1:1.8 MPa, 0.78; B2: 2.5MPa, 0.56; B3: 3 MPa, 0.47; B4:4 MPa, 0.35; B5: 51.0r AlMPa, 0.28; B6: 6 MPa, 0.33; B7: 6 MPa, 0.68. The0.8failure mode for the model during stage B is breakageand collapse. Accordingly, stage B is denominated0.6the“unsteady deformation stage". A sudden increasein loading stress caused a sharp rises in the IRT level,0.4-旷B3marked by peaks B1-B7, as seen in Fig. 5a. That is,, B6mechanical responses are sensitive to changes in0.2B7stress in deep rock masses and sudden increases istress induce breakage or collapse of the surrounding0.0Slage Arock mass (see Fig. 5b).Fig. 5b shows the structural responses related to1000 2000 3000 4000 5000the POL. For A1-A3 the two side walls convergedTime (S)towards each other, slightly at first with the inititia-(a) Temporal evolution of IRT with respect to loadingtion of cracks and then a marked convergence for A4and A5 with clearly observed propagating cracks. ForBI, breaking of the floor was seen with nucleation ofwing cracks on the left side wall. For B2, breakingoccurred on the left side wall due to the coalescence12A3A4of the cracks. For B3, the left side wall collapsed andthe cracks nucleated on the right side the wall. For B4and B5, propagation and coalescence of cracks wereseen on the right side wall, as well as the failure ofthe surrounding rocks in the vicinity of the left sidewall. For B6, floor heaving was seen and the roadwayAs31B2B3section converged prominently. For B7, the two sidewalls and floor broke down into the roadway spaceand the cracks coalesced on the roof. B7 correspondsto a complete failure at a vertical load equal to 6 MPaand a lateral pressure coefficienth equal to 0.68.BB:5 Results and discussion(b) Photographs showing progressive development of filureof the roadwayEnhanced understanding of the mechanical behav-Fig.5 Evolution of IRT as a function of progressiveiour for the roadway over the deformation and failuredevelopment of failure of the roadwayprocess was achieved by the description and charac-terization of the IR image and the 2D Fourier spec-Table 4 Mining depths, stress levels against POIstrum| F(u,v)l, which are computed with the algo-Depth (m)POIStress (MPa)Stess (MPa)rithm briefly mentioned in section 3.2 and 3.3.A1,A2,A3.8500A4,AS.45.1 Steady deformation800BI1.4Fig. 6 shows the figure sets for the steady deforma-10002.5tion process (i.e, loading stage A). The figures in1300333.0each row correspond to each POI A0 - A5 respectively.1700B44.0In Fig. 6, the figures in the first column are photo-2003s,.0graphs taken by the video camera; the second column60036s.02.0represent the denoised IR images; the third column is.0the 2D Fourier spectrum |F(u,川| and the origin ofthe co中国煤化工- the center of theThe IRT curves in the regime B1-B7 correspond to2D frfrequency com-the loading stage B for deep mining (mining depthponenCNMHGsthepresenceofincreased from 800 to 2600 m and λ=σ,/σ, var-wave components, the tourtn column is the sectionalied between 0.78 to 0.68). Over this stage B, the IRTdistribution of the 2D spectrum along the horizontalHE Man-chao et al695axis, ie. | F(u,0)|; finally, the fifth column is thecleation and localization of the cracks and the defor-sectional distribution of the 2D spectrum along themation of the roadway.vertical axis |F(0,v)|，For heterogeneous material,2) IR images A1-A2: the small scale IRT distribu-normal and shear stresses will occur at the same time,tion on the two side walls represents the damage ini-thus the P (primary, i.e, longitudinal) wave and Stiation. Images A3-A5: the growing area for IRT dis-(secondary, i.e., transversal) wave will exist simulta-tribution indicates the nucleation and localization ofneously as well. The P and S waves are coupled withdamage on the two side walls and floor (note that theeach other when reflection and refraction occurs ondamage on the floor is hardly discernable with thethe surface of an object or interface between two me~naked eyes from the photos but clearly seen on the IRdia. Therefore, the differences between the horizontalimages, indicating that the IR thermography outper-and vertical spectra are closely related to the stressforms the visible light camera in remote detection incapturing the damage on this microscopic scale).path and the orientation of the rock strata.The mechanical and structural responses for theStage A corresponds to shallow mining (miningmodel subjected to the external loading, ilustrated indepth≤500 m;A=1). The roadway section had re-the sets ofFig. 6 for A1-A5, can be understood andtained its original shape, deformed slightly andinterpreted as follows:gradually, without a macroscopic mechanical break-1) Photographs A1-A2 show the initiation of thedown, which was the reason for the denomination ofcracks on the two side walls. A3- -A5 depict the nu-the“steady deformation".Al: σ,=σ: =0.8 MPa; (mining depth of 350 m)Froquenory (出Freqpenay (H)A2: o, =σ; =0.8 MPa; (mining depth of350 m)Fesperay (HE).1....。A3: σ =σ: =0.8 MPa; (mining depth of350 m)Froaqutmny (H2)Faqumy CH)A4: σ,=0:=1.4 MPa; (mining depth of 500 m)Freapurnry (Hz)Froqpeney 02)A5: σ,=σ; =1.4 MPa; (mining depth of 500 m)Freqpency (Hz)Frepumacy (H2)(a) Photograph(b) IR image(C) 2D spetum|F(,0I中国煤化工() |F(0.nMIFig. 6 Characteristic maps for steady defMHCNMH G3) 2D and sectional spectral distributions: A1: theformed low frequency wave indicate the initial load-regularly-shaped 2D spectral distribution and pulse-ing state when the model had not undergone any in-696Mining Science and TechnologyVol.19 No.6trinsic damage. A2: the cross-line of high frequencywaves represent features in the frequency domain foron the 2D spectral plane and stress waves in the highthe steady deformation process of the modelband denote the initiation of the cracks on the model.5.2 Unsteady deformationA3- -A5: high frequency waves with large amplitudeand wide bands on the 2D and sectional spectralFig. 7 shows a set of figure for the unsteady de-plane represent damage propagation and localization.formation process related to B0- B8 (ie, the loadingNote that the bands in the horizontal and vertical di-stage B), which were arranged in the same order as inrections are almost the same, which validated the as-Fig.6. The understanding and interpretation of thesesumption that homogeneous and isotropic stressfigures are as follows:BI:可=1.8MPa，0: =1.4 MPa ; (mining depth of 800 m)Frequency 0Hz)Froqueney (HE)B2: o.=2.5 MPa. o, =1.4 MPa ; (mining depth of 1000 m)Frepenery (H)Frequency (H)B3: 0=3MPa，0: =1.4 MPa ; (mining depth of 1300 m)Frequenry (H2)Frequency (H2)B4: σ=4MPa. σ, =14 MPa ;(mining depbh of 1700 m)Fraqueney (Ha)Freqpeney (Hz)BS: o=5MPa. o; =14 MPa ; (mining depth of 2200 m)Frcquncy (0Hz)Froquency (HB6: σ=6 MPa. σ: =2 MPa ; (mining depth of 2600 m)Frequeney (H2)中国煤化工。B7: a=6 MPa, 0; =4 MPa; (mning depuh of 2600 m)MH.CNMHGFrequeney(Hhe)(a) Poograph(b) IR image(C) 2D spectrum|F(u,)I山ru,川I(e) |F(0.川Fig. 7 Characteristic maps for unsteady deformation processes (stage B)HE Man-chao etalPhysical modeling of filure process of the excavation in ...6971) Photographs:vertical bands, denoting the much higher frequencyBI (depth: 800 m; k =0.78): showing nucleationwaves propagated along the direction parallel to theof the wing cracks on the left side wall and breakagestrata, while the model had undergone macroscopicof the floor.fracturing and structural breakdown. In contrast to theB2 (depth: 1000 m, k =0.56): depicting nucleationsteady deformation (stage A), the frequency-domainof the wing cracks which lead to breakage of the leftfeatures for the unsteady deformation (stage B) areside wall.heterogeneous and anisotropic for stress waves.B3 (depth: 1300 m, k =0.47): ilustrating that coa-lescence of the cracks results in the collapse of the6 Conclusionsleft side wall, cracks propagating on the periphery ofthe left side wall, as well as nucleation of the cracks1) IRT characterizes the mechanical responses foron the right side wall in the surrounding rock mass.the model under loading by two stages, i.e, steadyB4 and B5 (depth: 1700 m, 2200 m; k =0.35,and unsteady deformation stages. Under the loading0.28): showing the breakdown of the surroundingconditions for shallow mining (≤500 m), the IRTrock mass on the left side and prominent floor heav-evolve at higher levels and the failure mode of theing which causes a dramatic convergence of themodel shows small-scale and gradual deformation,roadway section.thus denominating the“steady deformation stage”. InB6 and B7 (depth: 2600 m; k =0.33, 0.68): show-contrast, for the loading conditions simulating deeping that the model failed completely.mining (800- -2600 m), the IRT levels are much2) IR images:smaller than in shallow mining and evolve in aB1: nucleation of wing cracks is denoted by IRTquasi-cyclical manner with multiple local peaks. Theconcentration on the periphery of the left side wall;mechanical behavior of the model is sensitive toIRT concentration with deep red color on the leftstress changes and its structural responses are domi-lower cormer indicating intrinsic damage; floornated by breakage and collapse and is therefore re-breakage is ilustrated by the curved IRT distributionferred to as the“unsteady deformation stage".at the bottom of the roadway section.2) The denoised IR images provide a detailed deB2: growing area of IRT concentration on the leftscription and characterization of the mechanical re-side wall represents localization of damage, corre-sponses for the model, subjected to the external load-sponding to the breaking of the left side wall.ing in terms of initiation, nucleation and coalescenceB3: IRT concentration develops with larger extentof damage, as well as the eventual failure in a real-stretching far away from the left side wall, indicatingtime sense and over the entire field, which otherwisecoalescence of cracks; at the same time, nucleation ofwould have been undetectable and unobservable. Thewing cracks is seen on the right side wall.2D Fourier spectra characterize the rock responses byB4 and B5: increased area of IRT concentration ontheir 2D distribution patterns, the wave componentsthe two side walls and floor indicates the damageand the frequency bands, which provide features inlocalization due to the rock mass collapse and floorthe frequency domain and can be understood moreheaving, which causes a large convergence of theasily.B6 and B7: the large-scale connected low IRT areaAcknowledgementsindicates the collapse of the surrounding rock mass.3) 2D Fourier spectra: features of the 2D and sec-Financial support from the Special Funds for thetional spectral distributions over the unsteady defor-Major State Basic Research Project under Grant No.mation process can be summarized as:2006CB202200 and the Innovative Team Develop-a) 2D spectrum: the irregular degree of the 2Dment Project of the state Educational Ministry ofspectral distribution is closely related to the level ofChina under Grant No. IRT0656 are gratefully ac-the damage on the surrounding rock mass; the moreknowledged.irregular the 2D spectral distribution, the more intrin-sic damage the rock mass has undergone.Referencesb) Sectional spectra: wider bands and higher am-plitude for the high frequency waves represent the[1] Charlie C L. Disturbance of mining operations to a deepintrinsic damage on the rock mass. It is noted that theunderground workshop. Tunnelling and Undergroundwave bands in the horizontal and vertical directionsSpace Technology, 2006(21): 1-8.become different compared with stage A, indicating2]中国煤化IChina Universily ofI technology for min-that the coupling effects of the P and S waves areG226- 231.intense, due to the reflection and refraction of the3]MYHCNM H GwH, Yao P. Analyti-stress waves on the interfaces of the fractured strata.cal model and application of stress distribution on min-In most cases, the horizontal bands are wider than theing coal floor. Journal of China University of Mining &698Mining Science and TechnologyVoL.19 No.6Technology, 2008, 18(1): 13- 17.[12] He MC, Yong X J, Sun x M. Chinese Coal Mine Soft4] Blair S C, Cook N G W. Analysis of compressive frac-Rock- SIrudy on Characteristics of the Clay Minerals. Bei-ture in rock statistical techniques: part I: a non-linearjing: Coal Industry Press, 2006. (In Chinese)rule-based model. International Journal of Rock Me-[13] HeMC, LuX J, Jing H H. Characters of surroundingchanics and Mining Sciences, 1998(35): 837- -848.rock mass in deep engineering and its non-linear dy-5] Germnanovich L N, Dyskin A V. Fracture mechanismsnamic-mechanical design concept. Chinese Journal ofand instability of opening in compression. InternationalRock Mechanics and Engineering, 2002, 21(8): 1215-Journal of Rock Mechanics and Mining Science, 2000,1224. (In Chinese)37(3): 263 -284.[14] Grover P. Infrared can warm of an impending disaster.6] Tang C A, Kaiser P K. Numerical simulation of cumula-Chemical Engineering, 1990, 97(7): 141-146.tive damage and seismic energy release during brittle[15] Luong M P. Introducing infrared thermography in soilrock failure -part I : fundamentals. Internationallournaldynamics. Infrared Physics & Technology, 2007(49):of Rock Mechanics and Mining Sciences, 1998, 35(2):306- 311.1113-121.[16] Shi W Z, Wu Y H, Wu L x. Quantitative analysis of the7]ZhuWC,LiuJ,TangCA,ZhaoxD,BradyBHSimu-projectile impact on rock using infrared thermography.lation of progressive fracturing processes around under-International Joumnal of Impact Engineering, 2007(34):ground excavations under biaxial compression. Tunnel-990-1002.ling and Undenground Space Technology, 2005(20):17] Luong M P. Infrared thermovision of damage processes231- -247.in concrete and rock. EngineeFracture Mechanics,8] Zhu Q H, Feng M M, Mao X B. Numerical analysis of1990, 35(1/2/3): 291-310.water inrush from working- face floor during mining.[18] Wu L X, Wang J Z. Features of infrared thermal imageJournal of China University of Mining & Technology,and radiation temperature of coal rocks loaded. Science2008, 18(2): 159- -163.in China (Series D), 1998, 41(2): 158 -164.9] Jeon s, Kim J s, Hong Y C. Effect of a fault and weak[19] Rami H A, Wei B s, Shane J. Thermal elastic and infa-plane on the stability of a tunnel in rock-a scaled modelred-thernography methods for surface strains in crackedtest and numerical analysis. International Journal oforthotropic composite materials. Engineering FractureRock Mechanics & Mining Science, 2004, 41(1): 658-Mechanics, 2007, 75(1): 58 -75.663.[20] Gao X F, Lo Y T, Tam C M. Investigation of micro-[10] Kamata H, Masimo H. Centrifuge model test of tunnelcracks and microstructure of high performance light-face reinforcement by bolting. Tunelling and Under-weight aggregate concrete. Building and Environmen,2002, 37(5): 485- 489.[11] Nomoto T, Imamura S, Hagiwara T, Kusakabe O, Fujii N.[21] Pinsky M A. Introduction to Fourier Analysis and Wave-Shield tunnel construction in centrifuge. Journal oflets. Beijing: China Machine Press, 2003.Geotechnical and Geoenvironmental Engineering, 1999,125(4): 289 -300.中国煤化工MYHCNMHG