International Journal of Rock Mechanics & Mining Sciences 86 (2016) 5–15 Contents lists available at ScienceDirect International Journal of Rock Mechanics & Mining Sciences journal homepage: www.elsevier.com/locate/ijrmms Atypical evolution of seismicity patterns resulting from the coupled natural, human-induced and coseismic stresses in a longwall coal mining environment Maria Kozłowska a,n, Beata Orlecka-Sikora a, Łukasz Rudziński a, Szymon Cielesta a, Grzegorz Mutke b a b Institute of Geophysics, Polish Academy of Sciences, Warszawa, Poland Central Mining Institute, Katowice, Poland art ic l e i nf o a b s t r a c t Article history: Received 23 October 2015 Received in revised form 12 February 2016 Accepted 31 March 2016 A very unusual time-space distribution of seismicity accompanying coal exploitation has been observed in a Polish mine. The earthquakes did not follow the depth of mining but exhibited changing depths from far below to close to the mined seam. One of the deep events which occurred at close epicentral distance to the active mining front was ML3.7 event. This paper presents the study of possible coupling of natural, human-induced and coseismic stresses in a longwall coal mining environment. The full moment tensor solution showed that the strong earthquake occurred on an almost vertical plane consistent with approximate strike of local tectonic structure. The signal correlation analysis revealed some highly correlated pairs within both deep and shallow event groups. To evaluate inducing factor of ongoing and past exploitation, geomechanical modelling of its influence on strain and stress at the target depth has been performed. The results exhibited a changing vertical stress regime, which might have promoted failure on preexisting, almost vertical planes of weakness. The earthquakes' rate variation in time showed no increase in activity right after the occurrence of ML3.7 event. The P-S-wave spectra corner frequency ratio had an average of 1.0, suggesting slow rupture. However, the Coulomb stress change analysis showed that the magnitude of stress changes due to coseismic slip of ML3.7 event at the hypocentral depth is of the same order as the stress changes caused by mining. Thus, the distribution of seismicity at this level could have been driven by both exploitation and coseismic stresses. Moreover, the seismicity which occurred within first few weeks after ML3.7 event, followed positive stress changes. All the obtained results let us prove that the ML3.7 event was a tectonic earthquake triggered by ongoing exploitation and that the distribution of following seismicity was affected by coupled natural, exploitation-induced and coseismic stresses. & 2016 Published by Elsevier Ltd. Keywords: Mining induced seismicity Coulomb stress transfer Tectonic stress 1. Introduction Seismicity accompanies different methods of mining and occurs as failures of the strata, collapses or rock bursts (e.g. Ref. 1, and the reference therein). The distribution of mining seismicity is strictly related to the exploitation progress of mining works and, consequently, to the local stress field (e.g. Ref. 2, and the reference therein). The longwall mining system changes the stress field across the panel, ahead of the face, but also along the sides of the panels within headgate and tailgate.3–7 All these areas are highly probable for seismic events occurrence. According to many analyses, the maximum number of tremors occurs at a distance of n Corresponding author. E-mail address: mkozlow@igf.edu.pl (M. Kozłowska). http://dx.doi.org/10.1016/j.ijrmms.2016.03.024 1365-1609/& 2016 Published by Elsevier Ltd. tens of meters ahead of the face, above and below the panel.7–10 There are also local extremes in a goaf at a longer distance from the face.11 However, analyzing the seismicity from one of the coal mines of Upper Silesia Coal Basin (USCB), Poland, we observed very interesting and unusual time-space behavior of seismicity pattern – the seismicity typically accompanying exploitation exhibited changing depths from great below to close to mined seam. What is more, one of the events which occurred at great depth below, but at close epicentral distance to active mining front, was a strong event of ML ¼3.7. Such rather unusual seismicity pattern made this case worth studying in the context of possible natural, induced and coseismic stresses governing seismicity distribution in mining environment. In this paper we present the results of investigation of possible factors, both natural and induced by mining, causing this strong seismic event. Using various methods for studying the event, we 6 M. Kozłowska et al. / International Journal of Rock Mechanics & Mining Sciences 86 (2016) 5–15 try to find out whether mining exploitation in this region can activate underlying tectonic structures and trigger earthquakes. Also, we investigate whether strong events in mining and geological settings of Upper Silesia can influence the number and distribution of following weaker seismicity. This study is the continuation and extension of earlier work done by Marcak and Mutke,12 where the seismicity of the same mine was analyzed. Authors were investigating the tectonic activation of rock masses as a result of mining. We start our analysis from relocation of strong events to verify results obtained by the mine's routine. In the second step, we perform signal correlation to investigate whether deep and more shallow events have significantly different recorded signals. In order to carefully investigate the ML3.7 event, we determined its moment tensor and source parameters. To evaluate inducing factor of ongoing and past exploitation we performed geomechanical modelling of its influence on strain and stress in the rock mass at the target depth. The last part of the analysis is focused on investigating the effect of the coseismic stress changes due to strong event on following seismicity. Static stress transfer phenomenon has been studied in natural earthquakes and was proven to influence the distribution of aftershocks.13 Recently, few studies have analyzed the influence of static stress transfer on mining seismicity,14–16 showing that strong mining events are capable of producing stress changes big enough to influence the spatial and temporal distribution of following seismicity. In our study, we want to investigate whether similar effect was observed after ML 3.7 event. 2. Seismogenesis conditions of USCB The USCB in southern Poland is the place of intense coal exploitation carried out since 18th century. Coal seams deposited within Carboniferous sandstones, conglomerates and mudstones are now being exploited in more than thirty mines using the longwall method. Mining induced seismicity in the area has been monitored since 1929 when the first mining-dedicated seismological station was opened in Racibórz city in southern USCB. Since then, with only three years break after World War II, seismic monitoring was constantly improved leading to the present day Upper Silesian Regional Seismological Network (GRSS) conducted by Central Mining Institute (CMI) and a number of local networks conducted by mines themselves. Locations of most seismic mining areas correlate with tectonically complicated parts of the USCB, i.e. the Kłodnicki fault, Main anticline and Bytom syncline regions (Fig. 1). Seismicity in these regions' mines is not only the most intense in the area but also characterized by the highest magnitudes.17,18 The strongest events recorded in the USCB in the last twenty years had local magnitudes ranging from 3.8 to 4.0 and occurred in mines in mentioned areas. In numerous studies of mining seismicity authors have distinguished several mechanisms of seismic events which correspond to diverse stress state around mining face and exploited voids.3,19,20,21 Bimodal frequency-magnitude distribution indicates two main types of events and has been observed in several mines: e.g. Polish coal mines,22 Polish copper ore mines23 and South African gold mines.24 Mentioned studies suggest that smaller earthquakes - named type A after Richardson and Jordan,24 are directly connected with mining operations whereas bigger earthquakes, type B, are associated with local and regional tectonic structures which create weak zones and promote shear failure. The type A seismicity occurs as a result of interaction between mining induced and local tectonic stresses and cluster mainly in close vicinity to mining stope. Such events are characterized by low and medium magnitudes and can have significant non double-couple 18˚24' 18˚48' 50˚24' 19˚12' 50˚24' A 50˚00' 50˚00' 18˚24' 18˚48' 19˚12' Paralic Series Mudstone Series Upper Silesian Sandstone Series Krakow Sandstones Fig. 1. Map of USCB with seismicity recorded by GRSS in 2014, only ML 4 2.5-black points. Note the location of Bobrek mine. Main seismotectonic units mentioned in the paper are marked with letters: A-Bytom syncline, B-Kłodnicki fault, C-Main anticline. component associated with volume change caused by stope closure.19,20,25 Mining tremors associated with local and regional tectonic structures, type B, have usually larger magnitudes, their mechanisms are coupled to pre-existing faults or other zones of weakness and thus are rather of double-couple type. Such events can also occur at greater distances from an active mining front.3,18,26 Such events in USCB are usually associated with wellknown geological units intersecting the area mostly in its north part, including well mapped Kłodnicki fault and Main anticline17 (Fig. 1). Many of these big events have also been assigned to Bytom syncline intersecting numerous mines and striking WNW-ESE,12 however, no significant faults have been mapped there. One of the mines currently inducing intense seismicity and passing Bytom syncline axis is Bobrek Coal Mine in Bytom city (Fig. 1). In December 2009 strong seismic event of ML ¼3.7 occurred during exploitation of longwall panel (LP) 3 of coal seam no. 503. The routine mine location showed that the event occurred when mining front approached the syncline axis. The lack of damage at mining level and the fact that the event was felt on the surface at great distance from its epicenter indicated that its hypocenter was much below mined seam suggesting tectonic character of the event.12 Its great depth was later confirmed by 3D locations (Fig. 2). Interestingly, distribution of smaller events, typically accompanying exploitation, also exhibited great depths. What is more, the seismicity became more intense when the mining front passed approximately half of planned panel's length. 3. Data The strong seismic event of ML ¼3.7 occurred on December 16th, 2009 during mining of panel 3/503 in Bobrek mine. The mine, located in one of the most seismically active parts of USCB (Fig. 1), has been extracting coal from twelve seams using multilevel method and longwall system. Local magnitudes have been calculated from energy measurements using formula developed for USCB and based on local mine network recordings.27 A local network was designed to monitor and locate seismicity accompanying exploitation. At the time of the event occurrence it consisted of seven vertical and five three-component short period (1– 200 Hz) DLM3D seismometers manufactured by CMI, with sampling rate of 500 sps. In the second half of January 2010 the network was extended with 6 additional vertical sensors with the 7 M. Kozłowska et al. / International Journal of Rock Mechanics & Mining Sciences 86 (2016) 5–15 a) 4800 0 b) 5200 400 6000 6400 800 depth [m] coal seams X [m] 5600 1200 6800 1600 7200 7200 80 0 400 0 −400 5200 5600 6000 Y [m] 6400 6800 X [m] Fig. 2. Distribution of seismicity associated with LP 3/503 exploitation – black points, showing the ML 3.7 event – red star, and seismic stations closest to the panel 3 – triangles and black rectangle, respectively. Arrows show exploitation direction. a) Map view including seismicity recorded during exploitation of other mining panels of seam 503 – grey points, b) N-S cross section showing exploitation level – black horizontal line. X and Y coordinates are given in local coordinate system “Sucha Gora”. Note that not all stations presented in cross section are shown in a map view because of their large horizontal distance to the panel. Events behind mining front Events ahead of mining front 1000 800 N 600 400 200 500 400 300 200 100 0 -100 -200 -300 -400 0 -500 same parameters. The mining routine location and determination of events' energy was based on manual picking of P and S waves arrivals and performed only in 2D. Exploitation of seam 503 was held in four longwall panels, advancing from north to south. Since the start of exploitation of panel 1 in 2005, three strong events with local magnitude above 3.0 were observed, two of which are supposed to be associated with underlying Bytom syncline axis. The seismicity accompanying the exploitation of panel 3 became the subject of Laboratory for Monitoring Mining Induced Seismicity project (LMMIS) held by Central Mining Institute, a part of ISEPOS project (http://is-epos.eu/). It has been relocated by CMI using the same software as the mining geophysical survey, but in 3D mode, using repicked signals. The final catalog comprises of 2996 events with magnitude of completeness MC ¼ 1.5. CMI located ML3.7 event at depth of E700 m below exploited seam. The depths of smaller events accompanying exploitation exhibit untypical distribution - the majority of seismic events have also been located much below the seam with only few events typically surrounding the stope level (Fig. 2b). Note that no prior exploitation has been performed at such depth. The depth distribution changed at some point when, according to Marcak and Mutke 12, the mining front reached the Bytom syncline axis. At this point the seismicity climbed toward stope level, probably along southern limb of the syncline, reaching it at approximately 700 m of exploitation front advance. Rapid increase in a number of recorded events visible in Fig. 2b at approximately 6400 m could be linked to network extension, however the Modified Goodness-of-fit test used to determine Mc28 did not reveal decrease of Mc for catalog after network extension. What is more, such an increase in a number of events was observed not only in panel 3, but also in all other panels of coal seam no. 503 which were monitored by stable seismic network (Fig. 2a). This may indicate that such distribution of events is mainly the result of local geology and tectonics, rather than only the expansion of network. Unfortunately, we could not compare our data with the vertical distribution of seismicity from mentioned panels since we had an access only to mining routine 2D locations. However, according to Ref. 12, seismic events which occurred in other panels were also located at great depths below the seam. Interestingly, only vertical distribution of seismicity seems to be unusual; the horizontal distribution follows the typical pattern where mining events occur at close epicentral distances from active mining front (Fig. 3). To verify the unusual seismicity depths, epicentral distance to the front [m] Fig. 3. Distribution of number of events relative to the epicentral distance from the mining front, for seismicity of LP 3/503. Positive values of distance stand for events occurring ahead of mining front, within intact rock; negative values stand for events occurring behind active front. we relocated the strongest events using the single event relocation method based on probabilistic inverse theory.29 In the second step of depths verification the signals cross-correlation was performed. These two techniques are described below. 4. Events relocation Seismic events location procedure carried out by mining geophysical surveys usually has to deal with some specific issues. Two of them are crucial for our analysis, namely lack of origin time and, consequently, poor depth estimation. Since network timing is not controlled by any standardized time, it is also difficult to use some relocation approaches based on differences of recorded first arrivals times e.g. Double Difference.29 Here we used seismic signals recorded on vertical seismometers to relocate induced seismic events of ML 42.3 associated with LP. The location procedure worked with probabilistic inverse approaches and was already tested for Polish local mining seismic networks. Details of the 8 M. Kozłowska et al. / International Journal of Rock Mechanics & Mining Sciences 86 (2016) 5–15 procedure are described in Rudziński and Dębski.29 The locations were obtained using mean P-wave velocity of 3850 m/s, based on average velocities estimated for seismic stations at the time of longwall excavation (mining geophysical survey – personal information). Our procedure gave similar results as CMI relocation, with horizontal location differences not greater than 60 m. For five out of nine relocated strong events, new depths were smaller than those reported by CMI, however they were still much below exploited seam, e.g. E500 m below in case of the ML3.7 event (red star in Fig. 2b). The location errors differ among coordinates. Mean latitude error was 38 m (min: 33 m; max: 46 m) while mean longitude error was 42 m (min: 37 m; max: 50 m). Depths of hypocenters were much more influenced by location inaccuracy with mean errorE130 m (max: 171 m). Even though depths are the weakest point of events location, it did not affect observed seismic pattern. To investigate whether different velocity model could change events' depths, we performed tests using different velocities by adding and subtracting up to 500 m/s. Final results did not change the general observation that events' depths changed in time. Thus, the relocation step helped us confirm great depths of observed events and their untypical time pattern. In subsequent steps of the analysis, we applied new locations for the strongest events and CMI locations for the rest of the data bearing in mind small differences between these two relocation results. check whether their signals differ significantly between and within these groups. We used signals from 1 s before to 4 s after P-wave onsets and their maximum cross-correlation value as an indicator of signals similarities. Fig. 4 shows a matrix of correlation coefficients calculated for all event-event pairs. We used signals from a station with epicentral distance exceeding 900 m to be able to assume similar wave paths for all events. Also, this station was the least noisy from all distant ones. The mean cross-correlation coefficient within shallow events group – quarter II in Fig. 4-is 0.53. The same value within deep events group – quadrant III - is 0.52. By cross-correlating events from these two groups we get mean coefficient value of 0.48 – quadrants I and IV. The mean values of correlation coefficients are not significantly higher within groups than between them, however there are some eventevent pairs with very high cross-correlation close to 1 which occurred within both groups (Fig. 5). What is more, there is highly correlated group of about twenty-five events within quadrant II – shallow events. The cross-correlation analysis did not give us clear evidence of the changing events' depth, however it revealed some highly correlated pairs within both deep and shallow events groups. Such high values were not observed when correlating events between groups. It may indicate a changing depth of events. 4.1. Signals cross-correlation analysis 5. Characterisation of seismic source of ML3.7 event According to CMI relocation, events depths changed at the time when mining front approached the axis of Bytom syncline (Fig. 2b). Since inaccuracy in depth estimation could affect our interpretation, we decided to provide some further analysis to verify this unusual distribution of seismicity. Based on the idea that signals of events with close hypocenters and similar focal mechanisms recorded on the same station are very similar, we carried out additional signals processing, namely signals crosscorrelation. Assuming close epicentral distance among analyzed events, possible differences in recorded signals are either due to events' different depths or focal mechanisms. Here, we analyzed only tremors from central part of mining panel 3 to exclude those events which occurred close to panel's edges where local stress is accumulated (Fig. 4). Thus we could assume that focal mechanisms of analyzed events were similar. If so, only the vertical location could be the factor affecting recorded signals shapes. In our analysis we relied on 50 tremors before and 50 after depth change to 5.1. Moment tensor solution We manually picked P-wave onsets on all twelve available inmine stations and performed moment tensor inversion in the time domain.30 The inversion was performed using FOCI 3.4 software,31 applying the relocated origins and constant velocity and density of 3850 m/s and 2700 kg/m3, respectively. In the inversion procedure the amplitudes and polarities of the P-wave displacement onsets are used. Fig. 5 shows the full moment tensor solution for the ML 3.7 event. The full moment tensor solution shows that the earthquake occurred as reverse faulting on a northwest-striking plane which is consistent with approximate strike of Bytom syncline. Nodal plane 1 is almost vertical and plane 2 almost horizontal. Very high double-couple component of the moment tensor solution – almost 95%, can be well explained by the great depth of the event and, thus, supports the idea of its tectonic, rather than purely mining, Fig. 4. Cross – correlation coefficients matrix for selected seismic station (triangle in 3D view). Note that first 118 seismic events (blue points in 3D view) are stronger correlated among each other in comparison with the rest 118 tremors (green points in 3D view). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) 9 M. Kozłowska et al. / International Journal of Rock Mechanics & Mining Sciences 86 (2016) 5–15 0 100 200 [m] approach proposed by Andrews32 to use J and K velocity and displacement integrals33 as two independent parameters in frequency domain and based on them to calculate the low-frequency spectral level Ω0 and corner frequency f0. The source radius r0, slip u and stress drop ∆s were calculated using Brune's source model34 and Aki and Richards35 seismic moment definition. S wave velocity Vs and density ρ for GRSS network are 2660 m/s and 2500 kg/m3, respectively, giving shear modulus G value of 1.77
1010 Pa. Table 1 presents values of source parameters of ML3.7 event. Based on obtained seismic moment we calculated a moment magnitude of MW3.8,36 which is close to local magnitude. strike dip slip plane 1 plane 2 131° 309° 80° 90° 10° 87° Tensor components: ISO: 1.9 % CLVD: -3.5 % DC: 94.7 % 6. Numerical modelling of the effect of exploitation on analyzed depth rock mass Fig. 5. Focal mechanism of ML 3.7 event (star) obtained by moment tensor inversion of P-wave onsets. The event is shown on corresponding mining layout: dashed areas indicate mining panels: no. 2 (right hand panel – mined-out) and no. 3 (left hand panel - under exploitation at the time of the event). An arrow shows exploitation direction. character. The obtained mechanism is similar to the one described in Ref. 12, where authors also present reverse fault mechanism with similar strikes of both nodal planes, however, with different dips. The possible reason for such difference may be the fact that in 2009, when the analyzed event occurred, the mine network was being replaced and extended. At the time of the event occurrence, sensors from both networks, the old and new, modernized one, were in operation. Marcak and Mutke12 presented focal mechanism calculated based on old network recordings, whereas we present new network results. Unfortunately, data from the old network are not available anymore, so we could not compare both solutions. 5.2. Spectral analysis Spectral analysis was performed using MULTILOK: Mining Tremors Analysis Software developed by CMI. Since data from mine network were mostly clipped, in the analysis of ML 3.7 event we used S waves recorded at the one available at that time GRSS station with epicentral distance of 4.4 km. MULTILOK applies the Table 1 Mean source parameters of ML3.7 event calculated from six underground mine stations. Seismic moment M0 [Nm] Source radius r0 [m] Slip u [m] Stress drop ∆s [MPa] 4.4
1014 195 0.21 26 A numerical modelling approach of past and ongoing exploitation of Bobrek mine coal seams was proposed to study the stress and strain distribution in the rock mass. Particular attention was being paid to the depth range of mine exploitation influence. Our aim was to estimate the magnitude of stress and strain changes at the level of studied earthquake, i.eE500 m below exploited coal seam. To model the described case FLAC2D (Itasca Consulting Group, FLAC – Fast Lagrangian Analysis of Continua, Ver. 7.0. Minneapolis: Itasca) software was used. It is two-dimensional Fast Lagrangian Analysis of Continua software based on Finite Differences Method (FDM), commonly used to analyze static and dynamic problems. In our study we performed static analysis where modelled stress and strain state resulted from the appearance of the excavation at the particular depth.37 We applied Coulomb-Mohr constitutive model and built a model of rock mass as a 1200  1200 m continuum in plane strain conditions. The top edge of the model corresponds with sea level. The roof of analyzed coal seam 503 lies at the depth of 455 m bsl, however we are mostly focused on depth of ML3.7 event, namely around 1000 m. The model consists of 135 geological layers, represented by four different lithological units: sandstone, coal, sandy shale and clay shale, according to local geology.38 Geomechanical parameters of geological layers used in the model are presented in Table 2. The order of the deposition of the layers and their thickness were precisely read from the three borehole lithological profiles to the depth of 600 m bsl (Mining Geophysical Survey – personal information). Deeper layers were scattered randomly, based on general USCB geology.39 The grid consists of 257 856 elements (474 elements horizontally and 544 vertically). Boundary conditions were assigned to the edges of the model: zero horizontal displacements on both sides' edges and zero vertical displacement were applied to the bottom of the model. The additional loading of 7.24 MPa equal to the weight of the overburden was added to the top edge of the model. LP 3 in the coal seam 503 is located within the operating edges of twelve other coal seams. In the model, the order of exploitation has been preserved. The primary stress term, which is used later in the text, refers to the stabilized stress state Table 2 Geomechanical parameters of geological layers used in model.38,40 Coal Sandy shale Clay shale Sandstone Volume density ρ [kg/m3] Frictional angle ϕ [deg] Cohesion c [MPa] Young's modulus E [GPa] Poisson's ratio ν [dimensionless] Splitting tensile strength Rr [MPa] Tension Rc [MPa] 1290 2690 2600 2510 24 25 24 25.3 3.3 7.2 6 11.3 3.6 11.2 10 18 0.24 0.24 0.25 0.27 0.5 2.1 1.9 3.3 15.8 22.5 19.2 35.7 10 M. Kozłowska et al. / International Journal of Rock Mechanics & Mining Sciences 86 (2016) 5–15 Fig. 6. Profiles at the depth of 945 m bsl presenting: a) vertical displacement, and primary and secondary b) vertical stress Syy, c) horizontal stress Sxx, d) shear stress Sxy. Dashed line represents west edge of mined panel. resulting from exploitation of mentioned coal seams. The carried out numerical modelling of LP 3 exploitation resulted in the 2D distributions of secondary displacement and stress in the rock mass on the plane perpendicular to the mined panel. To carefully investigate the stress and displacement values at the depth of the studied earthquake, horizontal profiles were created (Fig. 6a–d). The location of the west edge of panel no. 3 was marked by a dashed line. Note that studied event's hypocentre was located E13 m east from panel's west edge. The biggest changes of all studied parameters can be observed near panel's edge, close to the hypocentre. The exact values read for hypocenter location are presented in Table 3. Presented results clearly show that past and ongoing mining exploitation has influenced the stress state at the depth of ML3.7 event. The observed stress changes are close to 1 MPa. An increase of compressive vertical stress around panel's edge resulted in vertical displacement pointed downwards. The area at the greater distance from panel's edge axis, especially on its west side, experienced opposite effect – relaxation of compressive stress caused vertical displacement pointed upwards. Such changing vertical stress regime can promote failure on pre-existing almost vertical planes of weakness. The obtained focal mechanism of analyzed event – almost vertical reverse fault – seems to support the presented idea. Also, the amplitude of vertical displacement along the Table 3 Stress changes in the ML 3.7 event's hypocentre. Negative values of stress indicate compression, positive – tension. Primary stress Vertical SYY Horizontal SXX Shear SXY 31.93 MPa 10.09 MPa 0.15 MPa Secondary stress 33.17 MPa 8.87 MPa 0.50 MPa Absolute difference 1.24 MPa (increase) 1.22 MPa (decrease) 0.65 MPa (caution- sign change) 11 M. Kozłowska et al. / International Journal of Rock Mechanics & Mining Sciences 86 (2016) 5–15 profile at earthquake's depth is of similar order as the slip on the fault – 0.21 m. Table 4 Time, energy and depth of eight strong seismic events which occurred within two months after ML3.7 event. No. Date ML Energy [J] Depth [m] 1 2 3 4 5 6 7 8 19.12.2009 31.12.2009 6.01.2010 13.01.2010 4.02.2010 5.02.2010 12.02.2010 19.02.2010 2.4 2.4 2.3 2.3 2.5 3.0 2.4 2.3 2
106 2
106 1.4
106 1.6
106 3.2
106 3.2
107 2.5
106 1.7
106 1179 1224 899 1159 1302 1290 1224 1093 7. Static stress transfer analysis In this chapter we want to investigate whether the strong mining event which occurred in the Bobrek mine affected the distribution of following seismicity. For this purpose we performed static stress transfer analysis, using Coulomb failure function 41: ∆CFF =∆τ +μ′∆σn, (1) where ∆τ and ∆σn are the shear and normal stress changes along receiving fault, respectively and μ′ is the effective coefficient of friction.42,43 A positive value of ∆CFF moves the fault towards failure, which increases the likelihood of fault rupture. The detailed analysis of ∆CFF influence on seismicity distribution has been already performed for copper ore mines in Poland 15. The results indicated clear influence of Coulomb stress changes on rate and location of following seismicity. In the case of the Bobrek mine, the seismicity rate variation in time shows no increase in activity right after the occurrence of ML3.7 event (Fig. 7). The mean daily rate in the week preceding the event was 10 events/day, whereas in the week following the event it was eight events/day. However, looking only at the number of strong events of energy 4106 J ( 4ML2.2) before and after the event, an increase is observed. Within two months after studied earthquake, eight events of energy 4106 J ( 4ML2.2) have occurred, including one of energy 4 107 J (4 ML2.7). For comparison, in the same period prior to the studied earthquake, only three events of energy 4106 J had occurred (none 4107 J). It may indicate that ML3.7 event caused an increase in number of strong events. One of possible explanations could be either static or dynamic stress change. However, eight mentioned events did not happen immediately after the main one, which excludes the dynamic character of the triggering effect44 (Table 4). Thus, to study the influence of ML3.7 event on following ones, we performed the static stress transfer analysis using Coulomb 3.0 software.45,46 Since mentioned strong events could also change the stress distribution in studied area, we included their effect in the performed modelling. The stress changes caused by exploitation promote vertical rather than horizontal fault rupture. Thus, in case of ML3.7 event, nodal plane 1 was accepted as the correct fault plane. Since data from mine network were clipped, we could not investigate the rupture's directivity. Hence, we assumed the event's hypocentre in the middle of the fault plane. We relocated all eight strong events and determined their focal mechanisms using methods described in previous sections (Fig. 9). To choose one of the nodal planes we relied on the results of Ref. 21. The author analyzed the focal mechanisms of mine tremors occurring in USCB and identified three types of seismic sources coupled to exploitation. In all three types the fault planes dips were greater than 45°. We performed our analysis accepting nodal planes which fit this pattern. The ML3.7 event and two strong following ones were close both in time and location – they occurred within eleven days and at hypocentral distance not exceeding 65 m. They had also similar mechanisms. Thus, we assumed that they represent continuing rupture and set hypocenters of strong events 1 and 2 in the lower edges of fault planes (Fig. 8). For the rest of strong events 3–8 we assumed the hypocenter in the middle of nodal planes. In the modelling we used the shear modulus and Poisson ratio values of 2.2 GPa and 0.24, respectively, which correspond to the mean values of those parameters for four floor strata.38 Fig. 9 presents the cumulative static stress changes due to ML3.7 and following events on receiver planes and at depths of target events 1-8. To investigate the influence of ML3.7 event on distribution of following weaker seismicity we divided the study area in three regions A-C, representing different parts of the panel (Fig. 10). For each region 70 Number of events per day 60 50 40 30 16.12.2009 ML=3.7 20 10 0 1.11.2009 30.11.2009 29.12.2009 28.01.2010 Time of seismc activity Fig. 7. Seismic activity in LP 3 in time. 27.02.2010 Fig. 8. Map of fault projections of the ML3.7 (star) and 1–8 strong events (circles). All planes are presented as projections at the hypocentres depths. 12 M. Kozłowska et al. / International Journal of Rock Mechanics & Mining Sciences 86 (2016) 5–15 Fig. 9. Cumulative static stress changes due to ML3.7 and following events on receiver planes and at depths of target events 1–8 ((a–h)). ML3.7 event is marked by a black star, subsequent strong events' epicentres are marked by black points. The rectangles represent mined-out areas, the position of front line of LP 3 corresponds to its actual position on events' occurrence day. M. Kozłowska et al. / International Journal of Rock Mechanics & Mining Sciences 86 (2016) 5–15 13 Fig. 10. Static stress changes due to ML3.7 event shown in three cross sections A-C. Seismicity following the event is shown as points coloured according to occurrence time: black – events which occurred before strong event no. 2, green – before event no. 3, pink – before event no. 4, grey – before events no. 5 and 6, red – before event no. 7, orange – before event no. 8, and points with no fill – after event no. 8. Strong events 1-8 are marked with stars coloured in the same manner as smaller events. An arrow indicates exploitation direction. The dashed line in section A marks the limit between ∆CFF calculated for receiver planes defined for upper and lower part of the region. Note that cross section A intersects the ML3.7 event's hypocentre. we determined the representative focal mechanism based on observed events, separately for upper and lower part of the region. In case of regions B and C, seismic events both in the upper and lower part had similar mechanisms. It is consistent with the assumption made in paragraph 4.1. In region A, seismic events closer to exploited seam had different mechanisms from the deeper seismicity. This part of LP is most affected by the exploitation geometry and stress changes are the greatest (Fig. 6). It may explain changing focal mechanisms of observed events. The fixed representative fault planes are as follows: upper region A: Strike¼ 222°, Dip¼ 67°, Rake¼ 52°, lower region A: Strike¼ 130°, Dip¼ 72°, Rake¼121°; region B: Strike¼ 265°, Dip¼ 76°, Rake¼ 108°; and region C: Strike¼129°, Dip¼62°, Rake¼102°. We performed the Coulomb stress change modelling accepting representative fault planes as receiver planes in each region separately. Fig. 10 shows the results together with seismicity observed in each region coloured according to the occurrence time. The distribution of weaker seismic events shows distinct correlation with CFF stress changes, especially within the range of lower depths of events. The seismicity which occurred within first three weeks after ML3.7 event (black and green points in Fig. 10) occurred mostly in two regions: below event's hypocenter and much above it, close to mined coal seam. Both areas were influenced by positive stress changes. Also four strong events 3, 5, 6 and 7 occurred within positive stress changes caused by the main and following strong events (Fig. 9). Both performed analyzes showed that two strong events, no. 1 and 2, occurred within negative stress changes areas caused by ML3.7 event. However, their close distance to event's fault plane lets us suppose that they were the continuation of its rupture. The actual stress change close to causative fault plane may be different from the one presented here due to complex slip distribution in the fault zone which cannot be well modelled without detailed knowledge of source geometry.47 8. Discussion and conclusions The presented results of ML3.7 event relocation, moment tensor and spectral analysis show that it was strong and deep earthquake with almost vertical fault plane and reverse faulting mechanism. It clearly shows that the event could not be directly induced by mining exploitation – the great depth of the event below exploited seam, E500 m, and almost vertical fault plane, with strike at an angle relative to panel's edges, do not match typical mining tremors observed in USCB and described in Ref. 21. The author identified three types of seismic sources coupled to exploitation based on seismicity of one of USCB mines with similar geology and exploitation depth to the one analyzed in present study. The first type are high energy tremors occurring in sandstone beds overlying the coal seam, with dominant normal faulting and fault 14 M. Kozłowska et al. / International Journal of Rock Mechanics & Mining Sciences 86 (2016) 5–15 planes at an angle of E 45° to the face line of LP. The second type events are mostly of normal faulting type, parallel to the coal face, within roof strata. The third type events have high explosive component and are coupled to the failure of coal seam. High energy seismic events occur mostly in roof strata, whereas low energy events are coupled to coal seam. The analyzed event clearly does not fit into any of these types. The calculated stress drop of analyzed event – 26 MPa, is in the range of both tectonic and induced earthquakes' values. There are studies which present low stress drop values for mining induced events,48,49 however, other studies present values similar to those of natural earthquakes.50,51 Dubiński and Mutke52 estimated the stress drops of earthquakes observed in USCB within limits of 0.028–32 MPa. Kwiatek et al.53 analyzed stress drops of microearthquakes with MW range from 4.1 to 0.8, recorded in a South African mine. Their study revealed changing stress drops between events close to the stope area and those at greater distances, within undamaged diorite dikes. The values varied from E06.–1.5 MPa close to the stope area to E2–8 MPa within undamaged dike. It suggests that low values of stress drops are linked to damaged zone surrounding the exploited area. The event analyzed in this study occurred at a depth that was great enough to assure undamaged rock and, as a result, stress drop comparable to natural earthquakes. Surprising is that after such strong deep event no clear aftershock sequence has been observed. This could be related to velocity of rupture and heterogeneity of the rock environment. The general approximation of rupture velocity can be based on the average relation of corner frequency for P-and S-wave spectra [Ref. 54 and the reference there]. Walter and Brune55 suggest that for seismic events with very rapid rupture the corner frequency ratio is around 1.4–1.5. For seismic events characterized by slow rupture, the discussed ratio is close to 1.0. In the case of the analyzed strong event the corner frequency ratio, estimated from the records of regional network GRSS, has an average of 1.0, suggesting slow rupture case. This would agree with other studies of mining induced seismic events56,57 implying slower rupture velocities than natural earthquakes of similar magnitudes. Although we assumed that mining exploitation did not cause fractures at the depth of E 500 m below exploited seam, we analyzed its impact on the stress state at this depth. The numerical modelling showed that past and ongoing exploitation influenced the rock 500 m below. The magnitude of stress change at this depth, close to the hypocentre, is around 1 MPa. For comparison, modelled vertical stress change only 20 m below exploited seam, in the event's epicenter, is close to 10 MPa, whereas below pillars or intact rock it can even reach 40 MPa. It shows that mining tremors which occur in close vicinity to the mine shaft are induced by stress change much higher than the value observed at analyzed target depth. Also, the event's stress drop is much higher than stress change caused by mining. It means that the event could not be directly induced by exploitation, but rather triggered. The Coulomb stress change analysis was performed to investigate if the ML3.7 and subsequent strong events could influence the distribution of following seismicity. The obtained results, combined with numerical modelling of exploitation effect, showed that the magnitude of stress changes caused by mining E500 m below its level are of the same order as Coulomb stress changes due to ML3.7 event, i.e. around 1 MPa. Thus, the distribution of seismicity at this level could have been driven by both exploitation and coseismic stress changes. Fig. 10 supports this idea – the seismicity which occurred within few weeks after ML3.7 event, at the same level, followed positive stress changes. Two strong events which occurred soon after the main one occurred within negative stress changes, however, their hypocenters where very close, mechanisms similar, so we assumed that they represent the rupture on the same, continuing fault zone. The majority of other strong events occurred within positive areas. The sudden change of depth distribution of seismicity which occurred few weeks after ML3.7 event was, according to Ref. 12, caused by mining front crossing the axis of Bytom syncline, which at this point continued as steeply dipping limb. The Coulomb stress change at this level had magnitudes many times smaller than the exploitation influence. Thus we can conclude that the seismicity which occurred after depths change, closer to mined seam, was induced by ongoing exploitation. However, as visible in Fig. 10 A, the ML3.7 event could have influenced the tectonically vulnerable area close to mined seam and promote its seismic activation. The tectonic stress analysis presented in Ref. 12 supports the hypothesis that ML3.7 event was the result of tectonic deformations caused by local geological structures. The complex analysis presented in this study let us prove that the event was the tectonic earthquake triggered by ongoing exploitation and that the distribution of following seismicity was affected by coupled natural, exploitation-induced and coseismic stresses. Acknowledgments This work was done in the framework of the project IS-EPOS: Digital Research Space of Induced Seismicity for EPOS Purposes (POIG.02.03.00-14-090/13-00), funded by the National Centre for Research and Development in the Operational Program Innovative Economy in the years 2013–2015. This work was financed by the Ministry of Science and Higher Education of Poland under contracts no. 3841/E-41/M/2013 and 3841/E-41/S/2015. The authors like to thank KWK Bobrek and Coal Company for sharing the data and support. Special thanks to K. Stec and A. Pierzyna for consultations. References 1. Gibowicz SJ, Lasocki S. Seismicity induced by mining: ten years later. Adv Geophys. 2001;44:39–181. 2. Kozłowska M. Analysis of Spatial Distribution of Mining Tremors Occurring in Rudna Copper Mine (Poland). Acta Geophys. 2013;61(5), http://dx.doi.org/ 10.2478/s11600-013-0137-x. 3. Gibowicz SJ, Kijko A. An Introduction to Mining Seismology. San Diego: Academic Press; 1994. 4. Holub K. Analysis of b-value in the frequency-energy distributions. Publ Inst Geophys Pol Acad Sci. 1995;M-19(281):153–161. 5. Fritschen R. Mining - induced seismicity in the Saarland, Germany. Pure Appl Geophys. 2009;167:77–89. 6. Holub K. A study of mining-induced seismicity in Czech mines with longwall coal exploitation. J Min Sci. 2007;42:32–39. 7. Boltz MS, Pankow KL, McCarter MK. Fine details of mining-induced seismicity at the trail mountain coal mine using modified hypocentral relocation techniques. Bull Seism Soc Am. 2014;104:193–203http://dx.doi.org/10.1785/ 0120130011. 8. Senfaute G, Chambon C, Bigarré P, Guise Y, Josien JP. Spatial distribution of mining tremors and the relationship to rockburst hazard. Pure Appl Geophys. 1997;150:451–459http://dx.doi.org/10.1007/s000240050087. 9. Gale WJ, Heasley KA, Iannacchione AT, Swanson PL, Heatherly P, King A. Rock damage characterization from microseismic monitoring. In: Proceedings of the 38th U.S. Rock Mechanics Symposium. Washington, D.C.; 2001. 10. Ellenberger JL, Heasley KA, Swanson PL, Mercier J. 2001 Three-Dimensional Microseismic Monitoring of a Utah Longwall. In: Proceedings of the 38th U.S. Symposium of Rock Mechanics. Washington, DC; 2001: 1321–1326. 11. Drzęźla B, Białek J, Jaworski A, Bańka P, Słapa W. Distributions of the number and energy of mining shocks nearby active longwall excavations. Publ Inst Geophys Pol Acad Sci. 1991;235:197–202 (in Polish). 12. Marcak H, Mutke G. Seismic activation of tectonic stresses by mining. J Seism. 2013;17:1139–1148http://dx.doi.org/10.1007/s10950-013-9382-3. 13. King GCP, Stein RS, Lin J. Static stress changes and the triggering of earthquakes. Bull Seism Soc Am. 1994;84:935–953. 14. Orlecka-Sikora B. The role of static stress transfer in mining induced seismic events occurrence, a case study of the Rudna mine in the Legnica-Glogow Copper District in Poland. Geophys J Int. 2010;182:1087–1095http://dx.doi.org/ 10.1111/j.1365-246X.2010.04672.x. M. Kozłowska et al. / International Journal of Rock Mechanics & Mining Sciences 86 (2016) 5–15 15. Orlecka-Sikora B, Lasocki S, Lizurek G, Rudziński Ł. Response of seismic activity in mines to the stress changes due to mining induced strong seismic events. Int J Rock Mech Min Sci. 2012;53:151–158http://dx.doi.org/10.1016/j. ijrmms.2012.05.010. 16. Kozłowska M, Orlecka-Sikora B, Kwiatek G, Boettcher MS, Dresen G. Nanoseismicity and picoseismicity rate changes from static stress triggering caused by a Mw 2.2 earthquake in Mponeng gold mine, South Africa. J Geophys Res Solid Earth. 2015:120http://dx.doi.org/10.1002/2014JB011410. 17. Idziak AF, Teper L, Zuberek WM. Seismic activity and tectonics of the Upper Silesian Coal Basin Katowice: Publ of Siles Univ; 1999 (in Polish). 18. Mutke G., Stec K. Seismicity in the Upper Silesian Coal Basin. Poland: Strong regional seismic events. In: Gibowicz SJ (Ed.). In: Proceedings of the 4th International Symposium: Rockbursts and Seismicity in Mines. Rotterdam: Balkema; 1997: 213–217. 19. McGarr A. Moment tensors of ten Witwatersrand mine tremors. Pure Appl Geophys. 1992;139(3/4). 20. McGarr A. Energy budgets of mining-induced earthquakes and their interactions with nearby stopes. Int J R Mech Min Sc. 2000;37:437–443. 21. Stec K. Characteristics of seismic activity of the Upper Silesian Coal Basin in Poland. Geophys J Int. 2007;168:757–768http://dx.doi.org/10.1111/ j.1365-246X.2006.03227.x. 22. Kijko A, Drzezla B, Stankiewicz T. Bimodal character of the distribution of extreme seismic events in Polish mines. Acta Geophys Pol. 1987;35:157–166. 23. Lasocki S. Quantitative evidences of complexity of magnitude distribution in mining-induced seismicity: Implications for hazard evaluation. In: Proceedings of the 5th International Symposium: Rockbursts and Seismicity in Mines, South African Institute of Mining and Metallurgy; 2001: 543–550. 24. Richardson E, Jordan TH. Seismicity in deep gold mines of South Africa: implications for tectonic earthquakes. Bull Seism Soc Am. 2002;92(5), http://dx. doi.org/10.1785/0120000226. 25. Gibowicz SJ. Seismicity induced by mining. Adv Geophys. 1990;32:1–74. 26. Lizurek G, Rudziński Ł, Plesiewicz B. Mining induced seismic event on an inactive fault. Acta Geophys. 2014:63http://dx.doi.org/10.2478/ s11600-014-0249-y. 27. Dubiński J. Calculation of seismic energy. In: Drzęźla B, Dubiński J, Fajklewicz Z, Goszcz A, Marcak H, Pilecki z, et al., eds. Mining Geophysicist Guide. Krakow: Publ of CPPGSMiE PAN; 1995. 28. Leptokaropoulos KM, Karakostas VG, Papadimitriou EE, Adamaki AK, Tan O, Inan S. A homogeneous earthquake catalogue for western Turkey and magnitude of completeness determination. Bull Seism Soc Am. 2013;103(5), http://dx. doi.org/10.1785/0120120174. 29. Rudziński Ł, Dębski W. Extending the double-difference location technique to mining applications Part I: numerical study. Acta Geophys. 2011;59(4):785–814 http://dx.doi.org/10.2478/s11600-011-0021-5. 30. Wiejacz P. Calculation of seismic moment tensor for mine tremors from the Legnica–Głogów Copper Basin. Acta Geophys Pol. 1992;40:103–122. 31. Kwiatek G. 〈http://www.induced.pl/projekty/foci〉 (In Polish); 2014. 32. Andrews DJ. Objective determination of source parameters and similarity of earthquakes of different size. In: Das S, Boatwright J, Scholz CH, eds. Earthquake Source Mechanics. American Geophysical Monograph, 37; 1986:259–267. 33. Snoke JA. Stable determination of (Brune) stress drops. Bull Seism Soc Am. 1987;77:530–538. 34. Brune JN. Tectonic stress and the spectra of seismic shear waves from earthquakes. J Geophys Res. 1970;75:4997–5009. 35. Aki K, Richards PG. Quantative Seismology: Theory and Methods. San Francisco: W.H. Freeman; 1980. 36. Hanks TC, Kanamori H. A moment magnitude scale. J Geophys Res. 1979;84:2348–2350. 37. Pilecki Z. Dynamic analysis of mining tremor impact on excavation in coal mine. In: Detournay C, Hart R (Eds.). In: Proceedings of the International FLAC 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 15 Symposium on FLAC and Numerical Modeling in Geomechanics. Rotterdam: Balkema; 1999: 397–400. Bukowska M. Laboratory studies of the mechanical properties of coal seam 503 and surrounding rocks in the “Bobrek-Center mine. In: Stec K, ed. Characteristics of Stress-deformation State in Seismic Hazard Based on the Parameters of Seismic Events. Katowice: Central Mining Institute; 2013:19–37 (in Polish). Gabzdyl W. Geology of Coal Deposits – Deposits in the World Warszawa: Polska Agencja Ekonomiczna; 1994 (in Polish). Stec K, Masny W. Numerical analysis of the dynamic influence of seismic tremors on mine workings depending to the location of focal plane. In: 2012 WorkshopNatural Hazards in Mining. Katowice: Central Mining Institute; 2012: 387–400. Scholz CH. The Mechanics of Earthquakes and Faulting. Cambridge: Cambridge University Press; 1990. Dieterich JH, Cayol V, Okubo P. The use of earthquake rate changes as a stress meter at Kilauea volcano. Nature. 2000;408:457–460. Catalli F, Cocco M, Console R, Chiaraluce L. Modeling seismicity rate changes during the 1997 μmbria-Marche sequence (central Italy) through rate- and state-dependent model. J Geophys Res. 2008, http://dx.doi.org/10.1029/ 2007JB005356. Belardinelli ME, Bizzarri A, Cocco M. Earthquake triggering by static stress and dynamic stress changes. J Geophys Res. 2003;108:2135http://dx.doi.org/ 10.1029/2002JB001779. Lin J, Stein RS. Stress triggering in thrust and subduction earthquakes, and stress interaction between the southern San Andreas and nearby thrust and strike-slip faults. J Geophys Res. 2004;109:B02303. Toda S, Stein RS, Richards-Dinger K, Bozkurt S. Forecasting the evolution of seismicity in southern California: animations built on earthquake stress transfer. J Geophys Res. 2005, http://dx.doi.org/10.1029/2004JB003415. Hainzl S, Marsan D. Dependence of the Omori-Utsu law parameters on main shock magnitude: Observations and modeling. J Geophys Res. 2008;113:B10309 http://dx.doi.org/10.1029/2007JB005492. Gibowicz SJ, Youn RP, Talebi S, Rawlence DJ. Source parameters of seismic events at the underound research laboratory in Manitoba, Canada: scaling relations for events with moment magnitude smaller than 2. Bull Seism Soc Am. 1991;81:1157–1182. Domański B, Gibowicz SJ. Comparison of source parameters estimated in the frequency domains for seismic events at the Rudna copper mine, Poland. Acta Geophys. 2008;56:324–343http://dx.doi.org/10.2478/s11600-008-0014-1. McGarr A. Some comparisons between mining-induced and laboratory earthquakes. Pure Appl Geophys. 1994;142:467–489. Yamada T, Mori JJ, Ide S, et al. Stress drops and radiated seismic energies of microearthquakes in a South African gold mine. J Geophys Res. 2007;112: B03305http://dx.doi.org/10.1029/2006JB004553. Dubiński J, Mutke G. Selected relations between earthquakes and mining seismic events recorded in Polish coal mines. In: Eskikaya S, (Eds.). In: Proceedings of 22nd World Mining Congress & Expo, 11–16 September, Istanbul. Turkish National Committee; 2011: Volume-I. 350–357. Kwiatek G, Plenkers K, Dresen G, JAGUARS Research Group. Source parameters of picoseismicity pecorded at Mponeng deep gold mine, South Africa: implications for scaling relations. Bull Seismol Soc Am. 2011;101:2592–2608http: //dx.doi.org/10.1785/0120110094. Collins DS, Young RP. Lithological controls on seismicity in granitic rocks. Bull Seism Soc Am. 2000;90:709–723. Walter WR, Brune JN. Spectra of seismic radiation from a tensile crack. J Geophys Res. 1993;98:4449–4459. Gibowicz SJ. Scaling relations for seismic events induced by mining. Pure Appl Geophys. 1995;144:191–209. Urbancic TI, Trifu C-I. Effects of rupture complexity and stress regime on scaling relations of induced microseismic events. Pure Appl Geophys. 1996;147:319–343.