Characteristics of transmitting channel wave in a coal seam

Availableonlineatwww.sciencedirect.conMININGScienceDirectSCIENCE ANDTECHNOLOGYELSEVIERMining Science and Technology 19(2009)0331-0336www.elsevier.com/locate/jcumtCharacteristics of transmitting channel wave in a coal seamYANG Zhen", GE Mao-chen, WANG Shu-ganghool of Mines, China University of Mining Technology, Xuzhou, Jiangsu 221008, ChinaCollege of Earth and Mineral Scienc, The Penn State University, PA 16802, USAAbstract: An embedded underground coal seam caries channel waves of low seismic velocity along a stratigraphic rock-coal-rocksequence. In a homogeneous and isotropic seam, seismic waves propagate as trapped waves within the seam, which leads to propagation of channel waves. We describe how to set up a field test for transmission in order to acquire channel waves in a coal seamBecause channel wave signals are non-stationary in their frequencies and amplitudes, a necessary velocity spectrum and wavelettransformation analysis are applied to interpret the characteristics of channel waves. The advantage of using a wavelet transforma-tion is that different resolutions can be obtained at different times and different frequencies. According to analysis of the seismicsignals acquired in the 7 sensor hole, it was clearly shown that the characteristics of channel waves are lower frequencies andtenuation which can guide an effective wave for detecting voids, boundaries and faults in coal seams with strong roofs and floorsKeywords: in-seam seismic technique; channel wave; coal seam; time-frequency domain; Gabor wavelet; velocity spectrum1 IntroductionA coal seam embedded in the underground carriesa channel wave of low seismic velocity along a stratigraphic rock-coal-rock sequencel-2).The propagationof seismic waves can be described mathematically byintroducing some simplifying assumptions. In aAComogeneous and isotropic seam, seismic wavespropagate as trapped waves within the seam, whichleads to propagation of channel waves(Fig. 1). Theenergy can be kept in the coal seam when the incidentangle is larger than the critical angle . which provides a promising method to indentify the characterFig. 1 Schematic development of interference systems ofSH-waves leads to channel wavesistics of channel waves from seismic signals acquiredin a coal seam Fig. l shows each SH-wave contributesa certain phase of the complete wave train. Its phase function of time can be easily obtained by usingelocity depends on the incident angle on the wavelet transformations. As well. wavelet transfor-coal-rock interface. The range of A is a leak mode. Bmations are more accurate than Short Time Fourierand C are normal modes, which trap the energy in the Transforms (STFT) to focus on transient signalscoal seamTherefore, time frequency analyses based on waveletThe Fourier Transform is a tool widely used formany scientific purposes, but it is well suited only for persion characteristics of channel waves in a coal seam,the study of stationary signals where all frequencies which is applicable in the selection of an appropriatehave an infinite coherence time!4l. It is well-knowfilter to analyze reflection signals. We will discuss thethat the channel wave signals are non-stationary in characteristic of channel waves by means of a velocitytheir frequency and amplitude. Wavelet analysis isused to decompose a signal into time and frequencyReceived 12 September 2008; accepted 03 December 2008THvelet transformation中国煤化工CNMHGProject B2532532 supported by the U.S. Mine Safety and Health Administrationorresponding author. Tel: +001-8145743817; E-mail address: zzyl @psu.eduMining science and Technol. 19 Noout of the testing site consists of four sections: a sen-2 Field testsor section for reflection surveys, a blasting sectionfor reflection surveys, a sensor section for transmis2.1 Introduction to the condition of testing sitesion surveys and a blasting section for transmissionA field test was carried out to investigate the char- sentially focused on the signals of transmission whichdo not have very complicated reflected signals. It willtremely thin bituminous coal seam(36 inches/0.9 m) be easy to figure out the characteristics of channelin the Black King Mine on May 10, 2008. The testing wavessite at the black King Mine is a long pillar, approximately 98 feet(30 m)wide(Fig. 2). The general layQ Sensor section for tra22section for trarFig. 2 General layout of test fieldAnother difficult condition for the Iss (in-seamseismic)technique is the irregular rib surface, such asthe one shown in Fig. 3. The irregular rib will affectthe reflection of channel waves in the coal seam Theadvantages of this test site include good, strong coaland excellent roof/floor conditions. Furthermore, boththe roof and the floor are sandstone. These factors150contribute positively towards the development andpropagation of channel waves.ig. 4 Layout of sensor holes for transmission surveysquint measured by feet)Sensor holes are drilled in pairs as a pseudo-2Dsensor arrangement. Given this arrangement, thesensors on the left of each pair are more sensitive toP-waves while the sensors on the right are more sen-sitive to S-waves. The technique of angled sensorpairs provides a simple and efficient means to studysignal polarization. In this field test, the sensor holeFig3 Rugged rib condition created by mining practice of S7 was sensitive to S-waves. The sensor holes ofS8 and s10 were more sensitive to p-waves because2.2 Transmission surveysthe direction of the boreholes is parallel to the shortest ray path of the transmissionThe sensor section for transmission surveys conThree transmission surveys were performed at thissists of six sensor holes, which are 150 feet (46.7 m)The seismic sources for the38 andaway from the corner of the crosscut as shown in Fig. g of dynamite detonated at the blasting boreholes2. The sensor holes are arranged in pairs, 4 feet(1. 22BT6, BT5 and BT4. These boreholes are perpendicu-m) long and 1.75 inches(. 45 cm) in diameter, drilledlarnd drilled in the mid-in the middle of the coal seam. The sensor holes s7中国煤化工nFg2. The size ofS8 and S10 were used to acquire seismic signals. In thesthe following discussion the signals waveforms arCNMHGSame asthat of thesenillustrated in the sequence S7, S8 and S10(Fig. 4)YANG Zhen et alCharacteristics of transmitting channel wave in a coal3Table i Relevant information for survey eventsEqs. (3)and (4)Hole群Dynamite(gEventBT6BT5088y()=x@, exp-2lr/*io,1277The distance of ray paths associated with these中(o)=(2nno|expt2(ansurveys ranges from 159 feet(48.5 m)to 170 feet7/(a-ay28 m) with an average path distance of 164 feet(4)(50.0 m). Table 2 shows the average velocities ac- where is the center frequency and y a con-cording to the data acquired in our test fieldstant taken as y=T(2/In 2)2=5.336. From theTable 2 Velocities associated with Black King MineFourier Transform, the half-value frequency width ofWave type(fu/s)(m/s)the Gabor wavelet is equal to 20, /y and the half-(usy(m/s)Channel wave 2705/8248225value of time is equal to 2y /o,. The Gabor wP-wave15684/4780function satisfies the orthonormality conditionRoof (sandstone)S-wave78622396WT coefficients wf(u, s)are obtained by summingthe products, which indicates the correlation betweenBecause the first arrival signal is from the roof and the signal and the Gabor wavelet function y(r).As afloor, the velocities of the P-and S-waves in the coal result, at high frequencies, a good time result is ob-seam cannot be identified by using traditional methtained, whereas at low frequencies, a good frequencyods, except for the waves coming from the roof or resolution is achievedfloor. Therefore, we have to use wavelet transforma-tion to separate the seismic signals from the time and 3.1 Original waveformsdomainsThe original seismic signals recorded during theransmission surveys are illustrated in Fig. 5. The first3 Seismic signal analysisimpression of the waveforms shown in Fig. 5 is thevery high frequency associated with the signal withinWavelet transformation is a tool for separating data, the first 20 milliseconds for each channel. The frefunctions or operators into different frequency com- quencies for these signals vary from 1500 to 3000 Hz.ponents, so that each can be studied with a resolution These high frequency signals are the P-waves andmatched to its scale s. That means that the wavelet S-waves coming from the roof and floor, not fromtransformation can divide a signal into different fre- coal, because the velocity of waves in the roof isquency bands and carry out inchoate signal decompo- higher than the velocity in the coal due to the differsition and extraction. Wavelet transformations have ence of density and the Lame constants between coaladvantages over traditional Fourier Transforms for roof and floor. In addition, these first arrivals ofrepresenting functions that have discontinuities and waves with high frequency signals are composed of asharp peaks. The mathematical expression for the number of subgroups, which is also evidence that thewavelet transform(WT) is presented by Eq (1)given newly arrived signals are from the roof and floorthe condition of Eq (2)when f(t) is a square inte- Therefore, it is necessary to filter the high frequencygrable function of time t.signals when analyzing the channel waves in the coalseam. These phenomena will be discussed in the fol-Wf(u,s)=(f, Yu)=COYdr(1) lowing section of time frequency analysiswhere u,s∈R,S≠0,* denotes a complex conjugation and y is the basis or mother wavelet functionThe factor 1/s is used to normalize the energy sothat it remains at the same level for different values ofgrue n凵中国煤化工6ado<+∞oCNMHGginally recorded for a transmission surveyWe have used the Gabor wavelet based on a gaus-(event #1276)at the testing site with the display windowssian function as a mother wavelet that is given asranging from 0-800 millisecondMining Science and TechnologyAfter the high frequency time period, signals withSeismic signals of event 1276, after using amuch lower frequencies began to appear in the300-800 Hz bandpass filter, are shown in Fig. 6 withnature Characteristic for these signals is that theythe display window ranging from 100 to 220 milli-very resilient with a long duration and very slow at- seconds. It is clearly seen that there are two grouptenuation, which are the characteristics of channel waves with low frequencies and high amplitudes inwaves in coal seams. Their frequencies are located each waveformbetween 400 and 600 Hz. In order to highlight theseismic signals with the lower frequencies in the coal3. 2 Case study for spectraseam,a 300-800 Hz bandpass filter was applied tov(t)is defined as the particle velocity of signalsthese original seismic signalsacquired by sensors installed in the coal seam andvOf)is their infinite Fourier Transform. The relationship is oftenenw()台(f)the transformation from one to the other in either direction. Using the inverse transformation, Eqs. (5)and体(6)are obtainedAi-wMwwwwini120140160180200v(t)= v(f)exp(rift )dyFig 6 Seismic signals after using a 300-800 Hv(f)=v(r)exp(-2nift )diEa81030Frequency(Hz)Frequency(Hz)(a)S7(b)s8菲離毛uency(Hz)ig. 7 Velocity spectra for event #1276 are shown for $7, S8 and S10 respectivelyThe velocity spectrum is calculated by using Eqs. (5) values in the three velocity spectrum graphs. One isand(6). The spectrum program is designed to visual- located in the range of 450-600 Hz, the other rangesize and analyze the velocity spectrum of a seismic from 1000 to 1200 Hz. That means two main factorssignal waveform. The velocity spectra of selected affectere captured by theeismic signals of event 1276 are listed in the se中国煤化 st arrival group ofquence of S7, S8 and S10 as shown in Fig. 7. Ac-signalCNMH Gnd floor. The sec-cording to their velocity spectrum analysis, it is reond grucontinued by the channelvealed that there are two regions with peak velocity waves in the coal seam.YANG Zhen et alCharacteristics of transmitting channel wave in a coal seam3.3 Time-frequency analysisThis group of signals travels through the roof andfloor to the sensors at high frequencies and attenuaWavelet transformations can easily convert signals tion in the field test site. Therefore, there are almostfrom the time domain to the time frequency domainThe advantage of using wavelet transformation is that no clear reflected waves with such high frequenciesdifferent resolutions can be obtained at different timesof 1500-3000 Hz in the following time domain. theand different frequencies. The data of signals col. second group of waves is at the bottom left of thelected in the coal seam are complicated. Some infor-time frequency graph. The characteristics of thismation will be lost when using velocity spectrungroup are a low speed and attenuation. Because theanalysis. The basic characteristics of those"trappedblasting holes are near the sensor section for trans-seismic signals in the coal seam will be discussed in mission(Fig. 2)and the boundary of coal seam is notthe time frequency domaininfinite, the second region is contributed by the firstIn Fig. 8, the first part is the original waveform. arrival channel wave, superimposed on the surfaceThe middle section is the Wt coefficients vs time atwave along the rugged rib(Fig. 3). The last part is onthe bottom right of the time frequency graph, which isfrequency of 600 Hz. The last section is the graph for the reflected channel wave from the upper boundarytime-frequency analysis of the S7 seismic signalof the coal seam. Fig 9 shows the 3D view of thetime frequency domain with the wavelet transformation coefficient for the seismic signal acquired in theS7 sensor hole by event #127610×10412×10414×10416×10418×1020×104Fig 9 3D view of time frequency domain for seismicsignals collected in the S7 sensor hole02×x1044×1046×1048×10410×1044 ConclusiorAccording to the analysis of seismic signals ac-quired in an extremely thin bituminous coal seamwith a sandstone roof and floor. the characteristics ofthe channel wave have been discussed. when the roofand floor are stronger than the coal seam, we can eas-ily find the "trapped"channel wave which can be10×1041210414×104161018×10°20×104 developed and utilized for void detectionTime series with range of 100-20In the velocity spectrum graph it has been provenFig8 Seismic signals collected in the S7 sensor holethat the"trapped"channel wave and the wave comingfrom the roof and floor can retain peaker values ofThe trend of channel waves is located easily after velocity than those in the range of other frequencythe wavelet transformation The data of event 1276 bands.recorded in sensor hole S7 are shown in Fig. 8. TheFrom the time frequency domain graph it is easy tosignals from the roof and floor and the signals from see that there are three peak value regions. The chanthe coal seam were superimposed separately on the nel wave and surface wave are located in the lowwaveform. The time frequency analysis provides an frequency region of 400-600 Hz. As well, the re-ivestigative insight into the entire seismic signal as fledshown in Fig. 8. There are three regions with high cles中国煤化工ce, the characteristivalues of wavelet transformation coefficients. One is ofCNMH Guencies and attenulocated on the top left of the time frequency analysis torve wave for detectinggraph with a time interval of 110-120 milliseconds. voids, boundaries and faults in coal seams with strongMining Science and TechnologyVol 19 No. 3roofs and floors[2] Dresen L, Ruter H. Seismic Coal Exploration Part BIn-Seam Seismic. New York: Pergamon, 1994Acknowledgements[3] Theodore C K. Channel waves as a tool of applied geo-physics in coal mining. Geophysics, 1963(28): 701-714J Thomas wK. Fourier Analysis. Cambridge: CambridgeWe would like to thank the Massey Energy and ElkUniversity Press, 1989Run Coal Co. for providing this unique testing site1 Ingrid D. Ten Lectures on Wavelets. Philadelphia: SIAM(very thin coal seam, 36 inches(0.9 m), with strong92.roof and floor) for the ISS project. The data from this1 Ziqin Z, Hojjat A, Wavelet energy spectrum for time-test will be an important component for assessing therequecny localization of earthquake energy. international Journal of Imaging System and Technology, 2003effectiveness of the ISS void detection technique for(13):133-140.bituminous mines. This work was funded by the U.S. [7] Suzuki H, Kinjo T, Hayashi Y, Takemoto M, Ono K,Ha-Mine Safety and Health Administration (MSHA)ashi Y. Wavelet transform of acoustic emission signals(B2532532)Journal of Acoustic Emission, 1996(14): 69-84[8] Stephone M. A Wavelet Tour of Signal Processing (2ReferencesEditon). San Diego: Academic Press, 1998[I] Evison F F. A coal seam as a guide for seismic energy.ature,1955(176:12241225.中国煤化工CNMHG