Spectral characteristics of micro-seismic signals obtained during the rupture of coal

Mining Science and Technology( China)21(2011)641-645Contents lists available at SciVerse Science Directand lesh氵容Mining Science and Technology(China)ELSEVIERjournalhomepagewww.elsevier.com/locate/mstcSpectral characteristics of micro-seismic signals obtained during the rupture of coalLiu Jikun, b, * Li Chengwu Wang Cuixia. b, Zhang Ruming a, b, Zhang HaofAculty of Resource and Safety Engineering. China University of Mining 6 Technology, Beiing 100083, chinaState Key Laboratory of Coal Resources and Safe Mining, China University of Mining 8 Technology, Beijing 100083, China"Shenyang Branch of China Coal Research institute, Shenyang 110016, ChinaARTICLE INFOABSTRACTReceived 22 January 201This study was performed to investigate the spectral characteristics of micro-seismic signals observedReceived in revised form 18 February 2011luring the rupture of coal. Coal rupture micro-seismic observations were obtained on a test system thatccepted 16 March 201ncluded an electro-hydraulic servo pressure tester controlled by a YAw microcomputer, a micro-seismicAvailable online 26 October 2011ensor,a loading system, and a signal collection system. The results show that the micro-seismic signalincreases with increasing compressive stress at the beginning of coal rupture the signal remains stableKeywordsfor a period at this stage. A large number of micro-seismic signals appear immediately before the mainrupture event. The frequency of micro-seismic events reaches a maximum immediately after the coalruptures Micro-seismic signals were decomposed into several Intrinsic Mode Functions(IMFs)by theMicro-seismic signalempirical mode decomposition(EMD)method using a Hilbert-Huang transform(HHT). The main fre-pectrum characteristicency band of the micro-seismic signals was found to range from 10 to 100 Hz in the Hilbert energypectrum and from marginal spectrum calculations. the advantage of applying an hht is that this canextract the main features of the signal. This fact was confirmed by an hht analysis of the coal micro-seis-mic signals that shows the technique is useful in the field of coal rupture.e 2011 Published by Elsevier B.V. on behalf of China University of Mining Technology2. Uni-axial compressive destruction tests on coal rock samplesCoal rock fracture disasters have a dynamic effect and disas- 2.1. Test designtrous consequences. They happen in a short time when coal rockbears an external stress [1]. Wide-band micro-seismic monitoringThe loading device was a servo controlled material testing mtechnology rich in spectral components can be used to monitor and chine with controlled loading speed and adjustable oil pressureforecast regional coal rock dynamic disasters. This can be real time The uniaxial compressive strength, the impact strength, and themonitoring of the energy released by fracturing of the surrounding corresponding micro-seismic events were measured with this sys-rock masses on a longwall face, from roof movement, or from rock tem. A micro-seismic sensor a loading system, and a signal collec-burst [2-6]. Parameters such as amplitude, frequency, and the tion system constitute the test apparatus The loading system wasnumber of incidents provide information on micro-crack extended an electro-hydraulic servo pressure tester controlled by a YAW ser-number, size, and direction. The micro-cracks are strongly corre- ies microcomputer, a control box, and another computer runninglated with the degree of damage in a coal-rock mass [7, 8 PowerTestV3 4 software The yAw6106 electro-hydraulic servoMicro-seismic signals caused by different damage mechanisms pressure tester was manufactured by MTS Systems Corporationhave different spectral characteristics with each corresponding(China)and is controlled by electricity It can maintain a load fora characteristic spectrum[9-11]. Micro-seismic signals of the same a long time and complex compression test procedures may beintensity may have different spectra [12-15]. Therefore, the inter- programmed into it. The test machine comprises a servo pressnal behavior of the rock may be deduced through the analysis of and an automatic control system that can apply a 1000 kN maxi-the micro-seismic spectral characteristics. This helps with moni- mum axial load. The loading method, speed, time, and drift aretoring and forecasting coal rock dynamic disasters.automatically controlled. recorded data include stress, linearstrain, and transverse strain and are collected synchronously themeasurement accuracy of load, stress, and drift values is greatlyenhanced. The equipment can obtain the stress and strain changeduring the entire damage process. Fig. 1 shows a photograph oft Corresponding author TeL +86 15210567646the system.中国煤化工1674-5264/ssee front matter 2011 Published by Elsevier B V on behalf of China University of Mining TechnolCNMHGdoi: 10. 1016/ j. mstc. 2011.10.010J Liu et aL/Mining Science and Technology(China )21(2011)641-645sensors were attached to both sides of the samples. Then the soundand electrical monitoring collection systems were started and thetest parameters were set. The loading system and computer werestarted and the loading path and speed were chosen. The loadingequipment was activated and the data collection was begun asthe coal sample came within the range of the load platen. Information was collected until the coal sample had broken. Finally, theapparatus was stopped and the test data were stored on a harddisk. A uniaxial compression loading plot on a coal sample is3. Results and discussion3. 1. Fourier spectral analysis of micro-seismic signals from coal subjectto uniaxial compressionFig. 1. Uni-axial compressive test system used for destructive testing of the coalThe micro-seismic signal spectrum from the Laohutai coal sam-2.2. Coal sample preparationple under uniaxial compression is shown in Fig 3 as a function ofstress. The coal sample collapsed after 462 s of sampling atThe coal samples were taken from the number 83002 cross- 2.5 MHz, which is after 543 s of loading time. Eight signal sampleshead of the Laohutai coal mine. They were processed into a cubic were chosen: before loading: from 60 to 62 s, 120 to 122 s, 180 toshape 100 mm on a side by cutting with a cut-off machine The coal 182 s, 240 to 242 s, and 300 to 302 s during the loading cycle: 2ssamples were kept in their original state by taking big coal samples immediately following initial rupture; and, 2 s after complete rup-from the upper cross-head 100 m in front of the working face. All ture(only some data are shown here because of publication limits).samples were from the same layer within the vertical stratification Fig. 3 shows that the micro-seismic signals from the coal-massand were near the same place in a horizontal direction. Dry drilling. under load are mainly low frequency signals in the band less thandry cutting, and dry grinding were used during laboratory process- 1000 Hz. The frequency distribution of the signals is related to theing. The machine speeds were reduced as far as possible to pre number and size of the micro-seismic events during the compresserve the sample structure. Ultrasonic tests on the processed coal sion process. Generally speaking the lower frequency componentssamples were done to determine sample uniformity. Coal samples will increase with the time. The main frequency occurs aroundwith obvious micro-seismic fracture were rejected. More than 10 90 Hz. Minor frequencies are mainly of low frequency early incoal samples were chosen as satisfying the test requirementsthe loading sequence and these move toward lower frequency asthe load increases and fractures begin to develop in the coal sam-ple. These frequencies are concentrated in the 100-200 Hz region23. Experimental proceduresbefore and after failure. After sample destruction the micro-seismicsignal frequencies are the same as those observed under no loadThe coal samples were put on the pressure head, insulatedThis suggests micro-seismic events are the product of energy accu-the testing machine base by insulation paper. Micro-seisimulation, to a certain degree.0.0100.2040608101.2141.61.820TH中国煤化工CNMHGFig. 2 Stress versus loading time: uniaxial compression of a Laohutai coal sampleJ. Liu et aL. /Mining Science and Technology (China)21(2011)641-64500l002040608101.2141.61.82.002004006008001000120000100.2040.60.81.01.2141.61.820Time(s)Frequency(Hz)Fig- 3. Time domain signals, and their spectra, from coal under uniaxial compression at different stress levels.0020.40.60.8101.2141.61820060080010001200Time(s)0,000.2040608101.214161.8200200400600800100012000.20.50.2Time(s)中国煤化工Fig 4. EMD decomposition of micro-seismic signalsYHsCNMHGL Liu et aL/ Mining Science and Technology( china) 21(2011)641-6451Energy in the IMFs obtained by EMD decomposition of several micro-seismic signals.0-2s24546s6-8s8-10s10-12s12-14s14-16s16-18s18-20s20-22s22-24524-26526-28528-30IMF1606357788006058.55062211.700874394110072172761290009210MF26.6050849000909153007337.3315.30015.700IMF315501550439000216047.100178033300244011.9014001830124016904030042300MF4146011.10298000143023.10025.702700020.7011.8016.5013.902080193031.10027400MF5182035802380013.703.11014006140121018801230752.6902940IMF65991640064307641160997168096109441050IMF 70.50623021002.120825737187020.609.910.299IMF817.00465.17807780500758IMF90820420038317.800.2773.7204095641.6194414500167D290IMF104811.21000400.0900440.10981962.02002000543.2. Hilbert-Huang transform on the micro-seismic signals from coalFig. 4 shows that the original micro-seismic signalssamplescomposed into 10 or 11 IMF components where the mean alsowas one IMF component Signals in the 6-8 s period were decom-3. 1. Empirical mode decomposition of micro-seismic signalsposed into 10 IMF components. The signals were decomposed fromTwo second long micro-seismic signals from 30 s prior to and high frequency to low according to the distance between the twoimmediately after sample rupture were decomposed by empirical adjacent significant peaks. That is to say, the high frequency wasmode decomposition(EMD). The main rupture occurred at extracted first and then the lower frequencies. The IMF 1 compo24-26 s and the remaining sample periods all showed significant nent has the highest frequency. the subsequent IMF componentsmicro-seismic events except the 4-6s, 8-10 5, and 12-14 s sam- obtained will be lower in frequency, and their wavelengths willples before the main rupture and the 26-28 s and 28-30 s samples be longer until the last, lowest frequency, has been decomposedtaken after the main rupture. Fig. 4 shows the frequency of each Each IMF component reflects different time scales and displays sig-Intrinsic Mode Function(IMF) component calculated by Fourier nal features at a different resolution, which means the resolution intransform. This shows that EMD is a feasible way to analyze the EMd decomposition is self-adaptablemicro-seismic signals The micro-seismic signals were decomEach IMF component waveform and the energy in each IMFposed from high frequency to low and each decomposed IMF com- component were obtained from the original signal by decomposiponent has an independent physical significancetion with DataDemon software. As shown in table 1 the sampl1200100000.20.40.60.81.0121.41.61.82020040060080010001200Time(s)100000.204060.8101.21.4161.8200080010001200Time(s)1200608101.2Time(s中国煤化工Fig. 5. Hilbert energy and mean spectra of micro-seismicCNMHGJ. Lfu et al /Mining Science and Technology(China)21(2011)641-645taken from 4 to 6s. 8 to 105, 12 to 14 s, 26 to 28 5, and from 28 to30 s had no significant micro-seismic events. The percentage range (2)The main frequency band of the micro-seismic signals wasof the energy in components IMF 1, IMF 2, IMF 3, and IMF 4 wasbetween 10 and 100 Hz. An HHT obtained by calculating thefrom 87. 3% to 96.48%. In samples with significant micro-Hilbert energy and marginal spectra, by EMD, of signals fromseismic events this range was from 32. 55% to 67.04% The percentthe coal micro-seismic events illustrated this factenergy in components IMF5, IMF 6. IMF 7. IMF 8, IMF 9, and IMF 10increased from a range of 3.52% to 12.7% to a range of 32.96% to (3)The advantage of applying hht to the field of coal micro -seis67. 45% in the presence of micro-seismic events. This means thatthese six low frequency components can be used to characterizemic study is confirmed by an hht analysis of the coal micro-seismic signals. HHT can extract the main features from themicro-seismic events and can be said to be the main frequenciestransformed signal and is an effective tool for dealing withof the micro-seismic signalnon-stationary signals in general. This makes the techniqueTable 1 also shows that at the sampling time of the main rupadaptable for the analysis of micro-seismic signals containinge - re samples with more micro-seismic events and of larger ampli-sudden transients and rapid drops in amplitude.de were obtained. the energy percentage in these componentsincreases more clearly than in the others. Note that especiallyIMF 8 and IMF 9 increased nearly 10-fold. This means that the Acknowledgmentsmicro-seismic signal spectrum tends toward lower frequencieswith an increasing extent to the coal destructionFinancial support for this work provided by the National Sci-Fig 4 shows that the frequencies of all the IMF components are ence and Technology Planning Project(No. 2009BAK54B03)andwer than 200 Hz, although frequencies higher than 200 Hz are the National Natural Science Foundation of China(no 50834005also displayed A frequency higher than 200 Hz is considered back- is deeply appreciatedground noise according to this analysis. The frequencies of thedominant bands that appear in components IMF 5, IMF 6. IMF 7, ReferencesMF8, IMF9, and IMF 10 range from 10 to 100 Hz. the main energyin the signals from coal rupture lies between 10 and 100 Hz[1]Nie BS. He XQ, Wangz, Sa ZY. Study on electromagneticeaning process of coal. J China Univ3.2.2. Hilbert spectra of micro-seismic signalsning Technol 2002: 31(nca simulation ofThe Hilbert energy spectrum, as shown in fig. 5, was obtainedby performing a Hilbert transform on the IMF components withompression. J China Univ Mining Technol 2010: 39(5): 648-51 [in Chinesthe EMD method. This indicates the energy distribution for each[3] Wang EY, He XQ, Liu ZT. Zhou SN Frequency spectrum characteristics ofagnetic emission of loaded coal. J China Univ MiIMF component. the energy is mainly concentrated in the rangebelow 100 Hz, which is the considered the main energy band from 4 w n& a, dex Lichi RIS st distin leof M during deforation and fracturemicro-seismic events, while above 100 Hz the spectra have verynarrow energy distribution. this is considered to be background[5] Dou LM, He XQ, Geophysics of mining. Beijing: China Science and CulturePress: 2002 (inby EMD from the original signal was plotted as a color-coded [71 Xie HP. Peng RD Ju y Ene eset hang XM. Mao ZY. Xu F]. A study onnoise acquired during the signal acquisition process(61 Jiang Fx, Yang SH, Cheng YH.The Hilbert energy spectrum of each IMF component obtainedmicroseismic monitoring of rock burst in coalof rock deformation and fractureChin J Rock Mech Eng 2004: 23(21): 3565-70 (in Chineselcolor the greater the energy of the datum in this plot. The instan- 8 m cP. Dou LM, wu xr. Wang HM, @in YH. Frequency spectrum analysis oftaneous frequency of each IMF component appears as a volumeeotech Eng 2005: 27(7): 772-5 lin Chinesethat fluctuates slightly around the center frequency but has little 19)Jiang Fx. Xun L Application of microseismic monitoring technology of strataoverlap with the other frequencies. the distribution of each IMFfracturing in underground coal mine. Chin J Geotech Eng 2002: 24(2 ): 147-9 [inChineselcomponent over time, frequency, and energy can be clearly [10] Zhang XM. Yu K. xi JD, Kong FM. Zhang XB, wang SI. The research anddescribed by this plot this is entirely consistent with the resultslogy in mine fractured and caving zonesfrom EMD decomposition and further shows the suitability of (11)Spetzler H. Sondergeld C Sobolev G Seismic and strain studies on largeHHT analysis of micro-seismic, coal rupture, signals.aboratory rock samples being stressed to failure. Tectonophysics 1987: 144(13):55-684. Conclusions[12] Ohnaka M. Acoustic emission during creep of brittle rock. Int J Rock MechMining Sci Geomech Abstr 1983: 20(3): 121-3.[13] Ohnaka M. Mogi K. Frequency characteristics of acoustic emission in rocksder uniaxial compression and their relation to the fracturing process ofilure. J Geophys Res 1982: 87: 3873-84(1)Micro-seismic signal strength increases as the compressive (14) Wang EY. He Xo, Liu ZT. LiZH Study on frequency spectrum characteristics ofstress increases at the beginning of coal rupture After a certainacoustic emission in coal or rock deformation and fracture. J China Coal Socstage, the signal strength is stable for a period. A large number2004:29(3)289-92| in Chinese.of micro-seismic signals appears before the main rupture event 15 Zhang H. The spectrum characteristics research of microseism signal in theand the rate of signal generation reaches a maximum when the2010 [in Chinese.coal ruptures中国煤化工CNMHG