Investigations on Gas Flow in Cracked Granite Samples

In the case of rock-fluids interaction, numerous studies have mainly focused on field and numerical simulations regarding the existence of toxic gases and investigations about the possibility of gasburst. This paper is related to the development of experimental investigations for granite formations on samples obtained from Creighton mine, Sudbury, Canada. An experimental methodology was developed for intact samples submitted to uniaxial compression for a certain level of temperature, with the monitoring of gas releasing from the rock. A detailed description of the rock-gas interaction apparatus is made. The rock cores were obtained at 2,400 m deep from the mine and results from a sample are presented, as well as theoretical assumptions are discussed. Finally, some conclusions about the investigations on pre-existing fluids in fractured rocks are presented.


Introduction
High in situ stress in a hard rock tunnel enables to trigger various types of failure such as spalling and rockburst during underground excavation at great depth (Diederichs et al., 2004;He et al., 2009).These events may take such proportions with serious consequences in the excavation process and the people involved in the work (Kaiser, 2009;He et al., 2012).Particularly under the disturbance of mining activity, deformation of surrounding rock enables to redistribute the pore pressure in the fluidrock systems to some extent, which provides conduits or closes fluids pathway, and in turn abnormal pressure can occur in regions where stress concentration appears around the cavities.
Many efforts have been made to investigate scientific issues concerning gas diffusion, swelling induced by gas sorption particularly in coal formations by laboratory tests (Karacan, 2007;Wang et al., 2007;Karacan et al., 2008;Yi et al., 2009;He et al., 2010;Vandamme et al., 2010).However, it is also needed to investigate gas transport to other types of rocks because of the existence of gases, sometimes toxic, and to investigate the possibility of gasburst.
Therefore, laboratory methods were performed at State Key Laboratory for GeoMechanics and Deep Underground Engineering (SKLGDUE) of Beijing to investigate the permeability of different types of rock samples sub-jected to a particular stress condition.Although the pore fluid pressure applied to the rock system can relieve the rock matrix from part of the higher in situ stress, failure of the rock is probably controlled by effective stress rather than total stress (Fjaer et al., 2008).Nevertheless, the experimental data necessary for the investigation of fluid flow are insufficient, because the pressurized fluid is more likely to suffer the influence of the changing porous structure on the stressed fluids.Fluid flow is strongly controlled by changes of porous structure of rock (Molli et al., 2010), which can act as localized conduits, barriers or combined conduit-barrier (Storti et al., 2003;Micarelli et al., 2006).
It is noted that deep rock core, such as granite and peridotite, commonly show low permeability.Fluid inclusions will be localized in micro-cracks of the rock or within the infilling of microstructures (Lespinasse et al., 2005).By taking advantage of the presence of fluid inclusions, it is possible to treat fluids inclusion as a preexisting gas reservoir stored in the rock.
In this article, instead of injecting selective stressed fluids into the tested rock core, intact granite samples obtained from Creighton mine in Canada were subjected to uniaxial compression at a particular temperature, being monitored the gas released from the rock samples.It is shown from one of a series of laboratory experiments that movement of original gas is similar to those observed during gasburst, and the relevant principle for evolution of pore fluid pressure during rock deformation is also discussed.
Apart from this brief Introduction, Section 2 presents the description of the apparatus developed at SKLGDUE and the experimental method used.Samples obtained from Creigton mine near Sudbury in Canada, are analysed in detail in Section 3 with a short description of the mine.Section 4 presents the results obtained in one sample with a detailed analysis.Section 5 discusses the assumptions that were made and conclusions are presented in Section 6. Acknowledgments and used references are also indicated.

Apparatus Description and Test Procedure
A new rock-gas interaction apparatus was specially developed to test original gas existing in rocks subjected to the external loading (Fig. 1).
The test system mainly comprises the following: 1) A servo-controlled hydraulic device that consists of an axial load cell and a pressure vessel containing oil to apply axial and confining pressure to the samples.2) A heating unit is used to measure and control the temperatures in the inner chamber of the steel pressure vessel and in the surface of sample.3) A gas monitoring unit for detecting, recording and analyzing the gas composition.
The specific requirements for the experiment were implemented in terms of rock deformation, gas transport and acoustic emission (AE) activity monitoring.A cantilever system was used to measure the relative displacements of four arms with strain gauges in order to accurately obtain circumferential deformation of the specimen.The four arms through ports on sleeve touch the surface of sample.
The ports between the arms and the heat-shrink tube are sealed using silicone gel.The axial deformation of sample is measured by relative displacement of the two platens against specimen by an axial strain jig.
The volumetric strain of sample is obtained from the following expression: where e V , e a and e r are, respectively, the volumetric, axial and circumferential strains.
The specimen is firstly assembled outside the vessel and then placed on the pedestal of the vessel.The size of specimen in the experiments is approximately 80 mm length, and 36 mm in diameter, which permits to have enough space for the installation of the AE sensor.The AE system is equipped with two-channel digital monitoring system of full waveform (sampling frequency 20 MHz, resolution 12 bit).Both thermal detector and AE sensor are attached to the surface of specimen connected by two rods with pore fluid outlet, which is followed by the wrap using a heat-shrink tube.
The gas flow pipe is connected to the gas vent at the top platen.In the test the gas outlet through the piston is open to the gas monitoring unit.Before testing, high pressure air is piped into the steel pressure vessel to check the gastight performance of the gas vent in the loading piston.Hydraulic oil is used as the confined fluid and the heating medium.
The gas monitoring unit was equipped with a highresolution gas pressure transducer (full scale ± 1,000 Pa, with a precision of 0.1%) manufactured by HELM Corporation, Germany, and with two different scale gas flowmeters by AALBORG Corporation, USA.The flowmeters can be alternatively selected according to the change of gas flux.Released gas from specimen is piped through the loading piston and enters in three-ways.One way is connected to a pressure transducer; the other ways are used as piping gases through a flowmeter and a Gas Chromatography (GC), successively.To monitor gas emission, pressure and flux data are recorded per second.The gas compositions are determined by a series of connected detectors that are equipped with a Thermal Conductivity Detector (TCD) and a Flame Ionization Detector (FID).The GC is calibrated repeatedly by the Universal Gas Calibration Standard before test.In order to obtain the relatively accurate concentration of gas component in mixed gases, FID is used as a dominant detector for hydrocarbon and carbon dioxide.The gas outlet condition is atmospheric.
The tests are performed by a computer-controlled system and monitored by data-acquisition software to record axial pressure, confining pressure, axial and circumferential strains, gas pressure, and gas flux.
In the experimental method used, the specimen set-up was assembled as described above.For the initial heating process, the temperature was programmed to heat the sample at a rate of 0.04 °C/min up to 47 °C, representative of the geothermal conditions of approximately 2 km beneath the ground surface.After reaching the set temperature, the system was left to thermal equilibrium for 40 min.The specimen was then loaded at a constant strain rate of 12.5 x 10 -6 s -1 until failure was seen in the stress-strain curve.Thereafter, the temperature of the fractured sample was linearly increased up to 63 °C again, which enabled the sample to release the existing gas.In the last stage, the confining pressure was applied to the deformed sample, which can drive out a large amount of gas.
The tests were all performed under uniaxial compression conditions, which are indeed different from the stress state of field granite masses at Creighton mine, where the in situ granite masses are structures under 3D stress and with high pore pressure actions.After excavating, 3D stress equilibrium of the granite masses can be changed into 2D stress one, like for example, for roadways and pillars.During failing process, shear failure of rock masses under 3D compressive conditions could be changed into flaking, spalling and possibly bursting of wall rock.Considering the limit of testing conditions, our investigation was focused on simulating the releasing process of gas from cracking of the granite samples under uniaxial compressive conditions.
The gas outlet is connected to atmosphere; the air in the laboratory was checked using GC to obtain the compo-nent concentration in atmosphere, so that the concentration of gas from the granite could be modified by subtracting the air concentration from the checked value.

Granite Samples From Creighton Mine
Creighton mine is more than 100 years old and it is located on southern part of the Sudbury Basin, Canada (Camiro, 1996).Creighton's sulphide orebodies are present in the lower sublayer of the hanging wall norites.The footwall rocks are mainly granite.The ore has been mined from surface to the deepest present level.A cross section of the mine is shown in Fig. 2.
Mine geometry has a critical effect on the observed seismicity.Footwall areas below 6600 and 7000 levels at granite formations experienced high seismicity.At these levels the granite exhibits toxic gases.
The overall geotechnical parameters corresponding to the depth of about 2,100 m for the rock mass are indicated in Table 1.In the table, E means the Young's modulus; n the Poisson's ratio; s 1 , s 2 and s 3 are, respectively, the in situ principal stresses, s t the tensile strength, c the cohesion and j the friction angle.The in situ state of stress was evaluated using overcoring tests.
The rock cores were obtained from the 2,400 m deep coring platform.With the greater production and increasingly deeper excavation, there exists an increasing variety of mining-induced accidents occurring in Creighton mine (Marisett, 2001;Mercer & Bawden, 2005).
The granite samples were cut to the measurement dimensions (diameter 36 mm, length 80 mm) in laboratory.
Investigations on Gas Flow in Cracked Granite Samples Figure 2 -Cross section of Creighton mine (Camiro, 1996).
The mineralogical composition of the granite is reported in Table 2.
The samples were bored from about the depth of 2,347 m at Creighton mine.After taking out from the drill pipe, the cores were immediately put into a bag that was sealed by wax before sampling.It is assumed that the gases in the granite pores are not lost during coring, transporting, and sampling because granite is compact and its porosity is generally low.
Several modelling approaches were developed to examine occurred failure mechanisms observed at deep levels.A 3D view of the mine of the examined areas is shown in Fig. 3.
As said before, three tests were carried out.It was found from the results of all other tests that they had the same characteristics of releasing gases, basically connected with the cracking of the cores.Thus in the paper only the results of one test are presented.

Results
The tests on the Creighton mine granite samples were performed under temperatures from 47 to 63 °C and strain rates at 12.5 x 10 -6 s -1 . In the paper, results from sample No. 123-127-660-2 are discussed in detail.
Figure 4 presents some results obtained.In a) the loading path and temperature are indicated; in b) the gas pressure at outlet; in c) the cumulative volume of releasing gas; in d) the releasing gas flux; and in e) the gas concentra-tion with time under loading at the temperature varying to 63 °C.
During the initial heating without the presence of other external loading, a slight increase in gas pressure was monitored due to the gas heat-expansion.But, as a result of the low porosity of intact granite, it can be assumed that some thermal expansion of the gas could increase pore pressure in the preexisting closed pore.
After the isothermal condition (47 °C) was achieved, the granite sample was subjected to uniaxial compression.Taking into account the presence of fissures generated in the igneous rock (the aperture is rather small), it is therefore accepted that stress concentration at the two ends during the deformation of rock matrix, leads to the change of deformation of the fissures.
More results of the test are presented in Figs. 5 to 8. Figure 5 illustrates the change of releasing gas pressure, flux and number hits of AE; Fig. 6 indicates frequency spectrum of AE activities for points A and B that are shown at Fig. 5; Figs.7 and 8 are referred to the change in volumetric strain and of AE hits, respectively.In Fig. 7a, the zero strain point means that the evolution of volumetric strain is first shrinking and then dilatant, particularly the initial state of volumetric strain can be restored when dilatant strain is zero.In Fig. 7b, normalized pore pressure in unipore element is defined by P n = (P -P min )/(P max -P min ), where P min is the lower limit of pore pressure, P max indicates the upper limit of pore pressure, and P represents the current pore pressure.
According to the figures, the changes of volume of the specimen can evolve into the following two phases: 1) Closure of pre-existing pore and cracks and compression-induced volumetric shrinkage, characterized by the gradual increase in cumulative hit number of acoustic emission.2) With the propagation and coalescence of crack, the volume of sample is changing from the compression to dilatancy, where the AE activity is dramatically increased in agreement with the previous conclusion on AE activity in rock salt (Alkana et al., 2007).(Camiro, 1996).
Investigations on Gas Flow in Cracked Granite Samples   The crack generated in the stressed rock sample is featured by tensional failure because of the no confinement of the lateral surface of cylindrical specimen.
In order to interpret the development of crack during the loading, a short-time window Fourier transform was used from Matlab software to filter the AE data obtained, which is shown in form of frequency-time.In this paper, the two spectrograms were used to respectively exemplify the characteristics for AE activities generated in the elastic deformation and near-failure stage (Fig. 6).
During the elastic deformation of the sample, AE activity, corresponding to point A, appeared on the high-frequency region of 350-400 kHz.According to the timefrequency images, a series of AE activities intermittently occurred, which indicated the occurrence of brittle fracture.It is generally accepted that high-frequency AE is featured by the abrupt cracking while the propagation of cracks can also lead to the low-frequency AE activities.With increasing loading of axial stress, the gradual increase in amount of AE activity resulted in the coalescence between microcracks where the localized dilatancy was occurring and began to drive the pre-existing gas within rock matrix.As a result of the increasing axial stress, the frequency ranges of AE activities are characterized by 140 and 380 kHz, respectively.Its duration was becoming longer, which indicated that considerable cracks were developing at the less tension position of sample.At the moment of failure of specimen the releasing gas pressure firstly plummeted to 1,000 Pa, and then gas pressure returned to the atmosphere after three seconds, which was followed by shortly increase in gas flux up to 0.09 mL/min indicating a small amount of gas emission.
Thereafter, the heating process was performed again, which readily drove the original gas out due to the permeability improvement of post-test sample.Apart from the CO, CO 2 and CH 4 , it is also clear that sulfureted hydrogen (H 2 S) started to release from the fractured granite rock, and its concentration tended to gradually fall down with increasing temperature.According to a series of SEM images (Fig. 9) of post-test sample, there are a number of intergranular and intragranular microcracks generated in rock matrix, which is more likely to split the fluids inclusion and in turn releases fluids included.The results for gas components are in agreement with the previous recorded data (Molnár et al., 1999).Therefore it is postulated that gaseous mixture obtained, including CO, CO 2 , CH 4 , and H 2 S, probably come from fluid-inclusion trapped in fractured surfaces of sample.
Figure 8 shows that most of the gas in the rock was gradually exhausted by an increasing confining pressure, which is programmed to apply on the fractured sample at 63 °C.The concentration of gaseous components increased from 17.5 ppm to 26.4 ppm for H 2 S, from 410 to 480 ppm for CH 4 , and from 95110 to 103460 ppm for CO 2 , respectively.This period is referred to as the closure of crack, which is characterized by the occurrence of some AE activities, and in turn compressed the gas distributed on the tortuous regions.When both aperture and crack closed to some extent, gas exhaust was accordingly ended.

Discussion About the Results
The presence of the fluids inclusion can provide numerous gas reservoirs, particularly for CO 2 (Elsworth et al., 2011;He et al., 2011;Qu et al., 2011).Reservoirs with igneous rock can also be considered (Sousa, 2011).Provided that the stress redistribution within the surrounding rock is as a function of the mining activities, fractures at rock matrix are generally initiated by stress concentrations at the tips or at the ends of fluids inclusion.The fluids included may release and ultimately contribute to macroscopic failure of rock due to the coalescence of the cracks.Fluids transport in rock mass strongly depends on pressure gradient, which is often ascribed to the changes of the structure of the stressed rock (Vargas et al., 2011).
During the rock deformation stage, brittle processes caused several main-fractures along the direction of loading from meso-scale CT images, and further generated a number of intergranular and intragranular micro-cracks from micro-scale SEM (Scanning Electronic Machine) images (Figs. 9 and 10).It can be therefore postulated that the microstructure of granite might be simply considered as a particular complex combination of crystal particle and pore, thereby applying the approximate dilute density method to interpret the pore pressure of the unipore subjected to the even stress field (Chen et al., 2009).Two assumptions are employed: 1) Low porosity of granite is proposed by considering the specimen consisting of equal elements, namely each element contains the one pore.
2) The interaction of fissures could be neglected, and the boundary condition of each element is similar to that of specimen.It is assumed that the volume of solid particle of granite, V S remains constant, then that is Eq. 2 can be rewritten in the form: (3) If it is assumed that the gas pressure in closed pore is P, then gas statue equation yields r is gas density, T temperature, R represents ideal gas constant.Eq. ( 4) can be given as The conservation equation of gas mass is given by (Zhao, 1994): It is assumed that there is no mass transfer between elements due to low porosity of granite at shortly loading process.We can therefore consider divergence of the element is null at the deformation stage, then with the help of Eqs. 4 and 5, we can rewrite Eq. 6 in the form: It is postulated that the variation of void volume is approximately consistent with volumetric strain obtained from Eq. 1: The normalized pore pressure curve in Fig. 7 suggests that the compressed sample at elastic deformation may initially increase the pore pressure of closed pore without fluids mass transport, and then level out a certain value.After the volumetric strain overtook the dilatancy boundary (Alkana et al., 2007), the volume of specimen underwent swelling.It is noted that the normalized pore pressure experienced a reverse process, just as breathing circulation.In other words, it firstly dropped, and then steep increased at the zero strain point, namely the volume of the sample tested returned to the initial volume without effects of stress.
It is clear from the CT images of post-test sample in Fig. 10 that the extensible cracks fully developed parallel to the loading direction.These observations imply that the dominant fluid flow is initially driven into the major dilatation zones by the fluid pore pressure gradient between new tortuous damaged zones and preexisting cracks.Subsequently, the rapid flow of the pore fluid fill the broken and damaged zones, so that the pore pressure distribution becomes in equilibrium within and around the gas reservoir, expelling the redundant gas.In contrast with the permeabil-Investigations on Gas Flow in Cracked Granite Samples ity enhanced by tensile cracks at uniaxial compression, the increasing confining pressure obviously reduced the aperture of fractures and momently expelled large amounts of gas.
The sequential seismic activities and gasburst are closely related to the fault movement.
The experiment performed by the same procedure with various confining pressure are given in Fig. 8.The gas emission is characterized by the abrupt release of high concentration and large volume of mixed gas as the increasing confining pressure.During the closure of aperture of crack network, the gas flux decay rapidly with loading-induced crack closure.Brittle processes, from millimeter-scale microcracking to kilometer-scale earthquake rupture, change differentially the permeability of fluid filled rock strata, which frequently alters the direction of preferred fluid movement driven by the fluid pore pressure variation.In most of the previous publications (Etiope and Klusman, 2002;Yang et al., 2003;Weinlich et al., 2006;Li et al., 2007;Shen et al., 2008), at the later stage of the mining induced failure of tunnel, the local cracks within surrounding rocks began to coalesce and develop large fracture or failure structure planes, where stresses and displacements change significantly.
It seems reasonable to suggest, from these field observation together with our test results, that the transient change of porous structures under stressed conditions definitely contribute to the unbalanced distribution of pore pressure in gas reservoirs, which in turn controls the transport of gases within the voids and fractures, as also reported field observation of gas concentration drop.

Conclusions
An investigation about preexisting fluid flow in granite samples from Creighton mine in Canada in a closed system under controlled physical conditions was performed.The samples under 47 °C were subjected to uniaxial compression, as well as with monitoring the AE activities.
Results obtained indicate that the released gas not only probably comes from the fluid inclusion trapped in granite matrix, but also depends on the volumetric change of stressed rock.
The pre-existing gas can be driven out from the failure of stressed granite sample in laboratory.The results show that the clustered AE activities that increasingly grow in stressed rock indicate the coalescence between cracks, and also provide the fluids with ample reservoirs.Such instant increase in gas reservoir may plummet down the gas pressure, correspondingly resulting in the gas pressure gradient among different gas reservoirs and subsequently equilibrating the distribution of pore pressure.It is believed from the relation between the deformation of sample and gas pressure in our test results that the instantaneous variation of pore structure can contribute to the suck-emission of fluids within the deforming rock.
The approximate dilute density method as a qualitative method describes the change of pore pressure in closed micro-pore, which is based on simple analysis of the relationship between pore pressure and volumetric strain of the entire sample.The change in pore pressure of matrix is determined by the volumetric strain of rock bulk.When the volumetric strain is changing from the compression to dilatancy, namely over the zero dilatancy boundary, pore gas pressure is experiencing firstly drop and then rise at shortly time.We can therefore consider the zero dilatancy boundaries as the threshold of pore pressure in closed pore.
A vast of pre-existing gas emission is controlled by the closure of stratified structure of rock mass in terms of amounts of gas releasing from the specimen.Among the gaseous mixture, CO 2 is the largest component, while H 2 S accounts for the least one.Although the experimental results obtained can be interpreted as gas transport within the stressed rock, it cannot be quantitatively analogized by the data obtained from in situ observation.Further study should be focused on the coupling effect of high-stressed fluids and stressed fluids taking the complication of the failure of deep rock stratum influenced by geophysics and geochemistry process.Also further study should include the scale effect in the investigations on gas flow, like for large field gas operations and for the occurrence of rockbursts, taking into consideration the existence of faults and other discontinuities

Figure 1 -
Figure 1 -Schematic diagram of the internal component of the apparatus.

Figure 4 -
Figure 4 -Test results for a granite sample from Creighton mine.

Figure 5 -
Figure5-Change of releasing gas pressure, flux and number hits of AE under uniaxial compression.

Figure 6 -
Figure 6 -Frequency spectrum of AE activities at point A and point B during the uniaxial compression.
Figure 7 -(a) Change in volumetric strain from the comulative number of AE hit number; b) normalized pore pressure in unipore element.

Figure 8 -
Figure 8 -(a) Changing of the AE hits and the confining pressure; (b) gas pressure and flux.

Figure 9 -
Figure 9 -Photos of fractured sample and SEM images of rock fragments on different position of post-test sample.

Figure 10 -
Figure 10 -CT images of the deformed specimen (red dot line represents the position of scanning slice of post-test).

Table 2 -
Mineral content of the granite sample.

Table 1 -
Geomechanical parameters for the rock mass