ISSN 10637834, Physics of the Solid State, 2011, Vol. 53, No. 9, pp. 1876–1881. © Pleiades Publishing, Ltd., 2011. Original Russian Text © A.G. Kadomtsev, E.E. Damaskinskaya, V.S. Kuksenko, 2011, published in Fizika Tverdogo Tela, 2011, Vol. 53, No. 9, pp. 1777–1782.
MECHANICAL PROPERTIES, PHYSICS OF STRENGTH, AND PLASTICITY
Fracture Features of Granite under Various Deformation Conditions A. G. Kadomtsev*, E. E. Damaskinskaya, and V. S. Kuksenko Ioffe PhysicalTechnical Institute, Russian Academy of Sciences, Politekhnicheskaya ul. 26, St. Petersburg, 194021 Russia * email:
[email protected]. Received February 21, 2011
Abstract—Fracture mechanisms of dry and watersaturated granite samples and slip (displacement) along a ready fault have been studied by measuring acoustic emission signals. It has been found that disperse defect formation is observed in dry samples under mechanical load, then localization occurs, and a fracture source is formed, whose development results in macrofault formation. In water saturated samples, chaotic defect formation occurs in the entire volume, which leads to a high degree of material damage. At the closing defor mation stage, several fault source zones are formed in which main cracks develop. In the case of slip along a ready fault, stoppers at crack edges are broken. DOI: 10.1134/S1063783411090150
1. INTRODUCTION This study is devoted to the effect of deformation conditions on rock fracture. Interest in this problem is caused by the fact that many factors affect rocks under natural conditions, such as uniform compression, porewater pressure, and upper bed pressure. Further more, the Earth’s crust in many regions is cut by numerous crust faults, along which displacements can occur, and by cracks which are stress concentrators. Laboratory experiments make it possible to differenti ate the effect of each factor on fracture development. In this work, the defect accumulation was analyzed in experiments of three types: (i) deformation of dry samples under controlled loading conditions (using a feedback device, i.e., the load depends on the defect generation rate), (ii) deformation of watersaturated samples, and (iii) deformation of samples, allowing stickslip modeling. Fracture features under controlled loading condi tions were studied in detail in [1, 2]. In this paper, we present some results of these experiments which are required for direct comparison with fracture mecha nisms under other deformation conditions. It is known that liquid penetration into a material causes changes in its physicomechanical properties: the Pwave propagation velocity changes, the volume increases, and the electrical resistivity and strength change [3–6]. To explain these phenomena, several models were proposed [3, 7]. In [5], it was shown that the technological impact associated with liquid pump ing or extraction from a rock massif has a significant effect on the seismic conditions. However, the mecha nism of the influence of a liquid on the strength has not been adequately studied. Therefore, we have per
formed laboratory studies of watersaturated granite fracture, compared the fracture nature of dry and watersaturated samples, and proposed possible mechanisms of the influence of water on changes in material mechanical properties. Rock massifs under natural bedding conditions always contain faults caused by natural (seismic, vol canic, etc.) or anthropogenic activity. As a massif stressed state changes, a displacement along these faults can occur. Many strong earthquakes in the sub duction zone1 developed according to such a mecha nism [8]. In modeling under laboratory conditions, it is conventional to call the slip along a ready fault as the stickslip [9]. In this study, we performed sample deformation experiments modeling stickslip, which will then make it possible to compare the experimental results with the earthquake source development. 2. MATERIALS AND EXPERIMENTAL TECHNIQUE In all three types of experiments using a setup with controlled deformation and water pressure (see [10] for more details), cylindrical Westerly granite samples (h = 190.5 mm, d = 76.2 mm) were loaded. Samples were deformed under conditions of constant uniform compression (pressure is Pc = 50.0 ± 0.2 MPa) and uniaxial loading (Pax). During the experiment, the axial load, longitudinal strain, and transverse strain were measured. Cracking was observed by means of measuring acoustic emission (AE) signals. It is known 1 The
1876
subduction zone is a linear zone along which Earth’s crust blocks sink under others.
FRACTURE FEATURES OF GRANITE
160
(b)
500
160 400 120
80
Pax, MPa
Z, mm
120
80
40
I stage of loading
100
80
300
60
200
40
100
20
40 0
0
II stage of loading
AE activity, imp/s
(a)
1877
20
40 60 X, mm
0
20
40 60 X, mm
that the main AE source in rocks is cracking [11], and there exists a relation between crack and AE signal parameters [12]. To record AE signals generated dur ing loading, a system of six piezoelectric transducers (with a resonant frequency of 0.6 MHz) was attached to the sample. The determination accuracy of hypo center coordinates of AE sources was ~3 mm in the entire sample volume for more than 104 signals. The experimental data were entered into a database, i.e., a set of parameters of chronologically sequential AE sig nals. Each signal was characterized by the emission time, three hypocenter coordinates, and amplitude A reduced to a reference sphere of radius Rf = 10 mm. The reduced amplitude is a signal energy parameter. 3. SAMPLE DEFORMATION UNDER CONTROLLED LOADING CONDITIONS The feature of typeI experiments is that the axial load variation was defined by fracture parameters. The load was varied so that the activity of AE signals with abovethreshold amplitude did not exceed a preset level. In [10], it was shown that such a mode made it possible to expand in time the usually fast faultsource stage and to study it in detail. Vol. 53
No. 9
2 t, 105 s
3
4
0
Fig. 2. Time dependences of the axial stress and the AE activity during watersaturated sample deformation.
Fig. 1. Projections of the hypocenter coordinates of AE sources for (a) dispersed fracture (AE signals recorded for the time from the loading start to 0.9 of the sample dura bility and (b) localized fracture, i.e., the formation of a fault source and macrofault propagation (AE signals recorded for the time from 0.9 of the sample durability to the sample fracture).
PHYSICS OF THE SOLID STATE
1
Figure 1 shows the XZ projections of hypocenter coordinates of AE sources (in essence, it is the spatial distribution of formed cracks). We can see that defects are initially formed dispersedly (Fig. 1a). Then local ization is observed, and a fracture site is formed, whose development results in macrofault formation (Fig. 1b). It should be noted that such distributions of hypocenter projections are observed when considering AE signals of large enough amplitudes, i.e., when ana lyzing large defects [2]. The fracture development in typeI experiments was previously analyzed in [1, 2]. It was found that the fracture occurs by the twostage mechanism [13] with the formation of, as a rule, a single main crack. 4. DEFORMATION OF WATERSATURATED SAMPLES To study the role of water in the fracture develop ment, we used an identical sample as in the experi ments described above, but completely saturated with water. The sample was deformed under uniform com pression conditions (50 MPa), water was pumped under a pressure of 1 MPa. Axial deformation was set by discrete steps, and the axial load was measured [6]. As seen in Fig. 2, AE activity bursts were observed at the instant of a stepwise increase in the sample load. It was interesting to trace the defect accumulation pat tern in time intervals when deformation and load remained almost unchanged. As in [6], we chose two stages (Fig. 2), stage I (87400–258500 s) and stage II (258500–437300 s), where the loads were 86% and 95% of the fracture load, respectively. Figure 3 shows
2011
1878
KADOMTSEV et al. (a) 160
120
120
Z, mm
160
80
80
40
40
0
20
40 60 X, mm
ized fracture nature was observed in these experiments for all AEsignal amplitudes, i.e., for all defects sizes. Only after the last stepwise increase in the load, the formation of several regions with increased defect concentrations is observed for ~100 s (which is 2 × 10 ⎯4 of the sample lifetime), which can be considered as fracture sites. Such regions are most clearly seen when analyzing not all AE signals, but only signals with suf ficiently high amplitudes (large gray circles in Fig. 3c). This can be interpreted as the formation of several main cracks whose development resulted in sample fracture. Based on typeII experiments, we can assume that the role of water role is as follows. (i) Water fills all defects such as cracks, pores, and capillaries already existed in a material before loading. These defects are dispersedly distributed over the sam ple volume. (ii) As is known [14], the hydrolytic mechanism significantly decreases the fracture activation energy. In [14], it was shown that water molecules play an active role in hydrolytic cleavage of interatomic bonds:
(b)
0
20
40 60 X, mm (c)
Si–O–Si + H2O
160
Z, mm
120
80
40
0
20
40 60 X, mm
Fig. 3. Defect accumulation in the watersaturated sample (a) at deformation stage I, (b) at deformation stage II, and (c) after the last increase in deformation. Loads are the same as in Fig. 2.
the XZ projections of AE signal hypocenters, corre sponding to these temporal stages. We can see that localization was not observed dur ing these stages; defects were formed dispersedly (cha otically) throughout the sample volume. A similar spa tial distribution was observed when analyzing both AE signals of all amplitudes and signals with above threshold amplitudes. It is important that the delocal
Si–OH + HO–Si.
In this case, the hydrolysis reaction under external stress occurs even at room temperature. The bond breaking activation energy in such a process is about fourfold lower and in the purely force process, which promotes the development (coarsening, coalescence) of existing defects. (iii) We can see that only a sharp deformation strengthening results in cracking, which is accompa nied by a significant but short increase in the AE activ ity (Fig. 2). (iv) Then, under constant load conditions, these cracks are filled with water, and their further growth is also accompanied by elastic energy release; however, its value is below the detection threshold of the used AE system. Most likely, new cracks are not formed at this time. (v) Finiteelement calculation of the stress field showed that the overstress (localtoaverage stress) ratio on the surface of the waterfilled defect is smaller than that on the defect surface in a dry material. Hence, the interaction of cracks, which is controlled by the stress field, is different for dry and waterfilled cracks. (vi) As a result, we obtain a severely damaged mate rial which is fractured under a smaller load than in experiments with the dry sample (Fig. 4). (As shown in [6], the volume of new defects was no less than 5% of the sample volume.) Thus, the fracture mechanism changes. When the fracture develops in watersaturated samples, the role of defects existed before loading appears significant. Waterfilled cracks develop under smaller stresses.
PHYSICS OF THE SOLID STATE
Vol. 53
No. 9
2011
FRACTURE FEATURES OF GRANITE
1879
600
400
Pax, MPa
160
1
2
200
0
1
2
3 Δl, mm
4
5
Z, mm
120
6
80 Fig. 4. Dependences of the axial stress on the strain for (1) watersaturated and (2) dry samples.
40
5. DEFORMATION OF SAMPLES ALLOWING STICKSLIP MODELING In traditional stickslip experiments, the sample is prepared in two parts tightly pressed to each other [9]. The displacement occurs along the contact plane placed at an angle to the axial load. In this study, dis placement conditions were not created in advance, but naturally appeared when loading an initially integer sample. Such situation is characteristic of natural condi tions of rock fracture (e.g., in subduction zones). In typeIII experiments, samples were loaded under hydrostatic pressure (50 MPa) conditions of uniaxial compression with constant deformation rate. In the first experimental stage, a fault was formed under applied load. This can be suggested by the abrupt (almost fourfold, from 375 to 99 MPa) drop in load and the AE activity burst. Figure 5 shows the XZ projections of hypocenter coordinates of AE sources detected at that time. We can see that the fault plane makes an angle of ~30° with the vertical. The fault geometry is schematically shown in Fig. 6a. If loading would be continued under the same conditions, the sample would be soon split into two parts. To prevent this, the hydrostatic pressure was twofold increased (to 100 MPa). In the second stage, under the increasing axial load, defects were accumulated and the lower sample part adjoining the loading device support was broken off (Fig. 6b). As a result, the fault took a con figuration (Fig. 6c) which allowed stickslip, i.e., the displacement shear of one sample part with respect to another, modeling. Let us analyze in detail the fracture development before the displacement along the fault. Figure 7a shows the variations of the axial stress and amplitude (A) of AE signals immediately before the displace ment. The spatial distribution of the same signals is shown in Fig. 7b. We can see that defects are formed PHYSICS OF THE SOLID STATE
Vol. 53
No. 9
0
20
40 X, mm
60
Fig. 5. XZ projections of the hypocenter coordinates of AE sources recorded from the loading start to the fault forma tion instant.
(a)
(b)
(c)
Fig. 6. Schematic representation of the fault in the sample: (a) fault formation, (b) breakoff of the sample part adjoin ing the loading device support, and (c) formation of the fault allowing stickslip modeling.
near the fault zone. The AE signal with an amplitude of 42 arb. units (which corresponds to large defect for mation) was detected at the central sample region, i.e., in the region where displacement is possible. This sig
2011
1880
KADOMTSEV et al. (a)
500
70
450
2
3
4
60 50
1 400
40 30
350
A, arb, units
Pax, MPa
configuration and the stress required for breaking pin ning centers.
80
20 300
1.524
1.525 1.526 t, 105 s (b)
1.527
10 1.528
(i) When deforming dry initially integer samples, the fracture develops in two stages: dispersed defect accumulation, localization, and the development of, as a rule, a single fault source.
The development of defects existed in the sample before loading is most significant. This causes random fracture in the entire sample volume and significant material damage. As a result, several fault source zones are formed, in which main cracks arise, resulting in fracture of the entire sample.
120 Z, mm
The present study allows the following conclusions.
(ii) When deforming watersaturated samples, the fracture mechanism changes.
160
(iii) In the case of fracture by the stickslip mecha nism, pinning centers at fault edges are broken. It can be assumed that an often observed seismic quiescence is explained by the time taken by Earth’s crust stresses to reach a value sufficient break the last strong pinning center constraining a displacement which is an earth quake. Therefore, to understand the development of such processes in the Earth’s crust, information about geometrical characteristics of active faults is required.
80
40
0
6. CONCLUSIONS
20 40 60 X, mm
ACKNOWLEDGMENTS
Fig. 7. (a) Time dependences of the axial stress and AE sig nal amplitude and (b) the distribution of hypocenter coor dinates of the sources of the same AE signals. Each dot corresponds to one AE signal. Signals with large ampli tudes are shown by gray circles.
This study was supported by the Russian Founda tion for Basic Research, project no. 090500639, and the Federal target program, state contract no. 02.740.11.0315.
nal is indicated by a gray circle in Figs. 7a and 7b. After that, multiple formation of relatively small (judging by signal amplitudes) defects was observed for 4 min. The next three AE signals with large amplitudes also appeared exactly in the displacement region (gray cir cles 2–4 in Fig. 7a). The described largeamplitude AE signals are most likely caused by the fracture of the last pinning center constraining the displacement. For the time between these events, sample deformation continued, the elas tic energy was accumulated, and stresses increased. When the stress reached a value sufficient for sample fracture, the stickslip occurred, which is indicated by an abrupt drop in load (Fig. 7a). In this case, an AE activity burst occurred, where all signals were small amplitude, i.e., the material was intensely fractured in the displacement region. Thus, deformation of a sample containing a fault results in the fracture of pinning centers at fault edges. The sequence of their fracture is controlled by the fault
REFERENCES 1. N. G. Tomilin, E. E. Damaskinskaya, and V. S. Kuk senko, Phys. Solid State 36 (10), 1649 (1994). 2. N. G. Tomilin, E. E. Damaskinskaya, and P. I. Pavlov, Izv., Phys. Solid Earth 41 (8), 660 (2005). 3. C. H. Scholz, L. R. Syke, and Y. P. Aggarwal, Science (Washington) 181, 803 (1973). 4. G. A. Sobolev, Fundamental of Earthquake Prediction (Nauka, Moscow, 1993) [in Russian]. 5. V. V. Adushkin and S. B. Turuntaev, Anthropogenic Pro cesses in the Earth’s Crust (Risks and Catastrophes) (INEK, Moscow, 2005) [in Russian]. 6. S. A. Stanchits, D. A. Lockner, and A. V. Ponomarev, Bull. Seismol. Soc. Am. 93, 1803 (2003). 7. P. M. Shearer, Introduction to Seismology (Cambridge University Press, Cambridge, 1999). 8. C. J. Ammon, H. Kanamori, and T. Lay, Nature (Lon don) 451, 561 (2008).
PHYSICS OF THE SOLID STATE
Vol. 53
No. 9
2011
FRACTURE FEATURES OF GRANITE 9. G. A. Sobolev, Kh. Shpettsler, A. V. Kol’tsov, and T. L. Chelidze, in Construction of Models of Seismic Pro cesses and Earthquake Precursors, Ed. by G. A. Sobolev (Institute of Physics of the Earth, Russian Academy of Sciences, Moscow, 1993), p. 38 [in Russian]. 10. D. A. Lockner, J. D. Byerlee, V. Kuksenko, A. Pono marev, and A. Sidorin, in Fault Mechanics and Trans port Properties of Rocks, Ed. by B. Evans and T.F. Wong (Academic, London, 1992), p. 3. 11. V. I. Myachkin, B. V. Kostrov, G. A. Sobolev, and O. G. Shamina, Izv. Akad. Nauk SSSR, Fiz. Zemli, No. 10, 2526 (1974).
PHYSICS OF THE SOLID STATE
Vol. 53
No. 9
1881
12. V. S. Kuksenko, A. I. Lyashkov, K. M. Mirzoev, S. Kh. Negmatulliev, S. A. Stanchits, and D. I. Frolov, Dokl. Akad. Nauk SSSR 264, 846 (1982). 13. V. Kuksenko, N. Tomilin, E. Damaskinskaya, and D. Lockner, Pure Appl. Geophys. 146, 253 (1996). 14. V. A. Bershtein, V. V. Nikitin, V. A. Stepanov, and L. M. Shamrei, Phys. Solid State 15 (11), 2173 (1973).
2011
Translated by A. Kazantsev