Pageoph, Vol. 115 (1977), BirkNiuser Verlag, Basel
Shear and Tension
Hydraulic
Permeability By P.
SOLBERG,1) D.
Fractures
in L o w
Rocks
LOCKNER1) and J.
BYERLEE1)
Abstract - Laboratory hydrofracture experiments were performed on triaxially stressed specimensof oil shale and low-permeabilitygranite. The results show that either shear or tension fractures could develop depending on the level of differential stress, even in specimenscontaining preexisting fractures. With 1 kb of confining pressure and differential stress greater than 2 kb, hydraulic fluid diffusion into the specimens reduced the effectiveconfining pressure until failure occurred by shear fracture. Below 2 kb of differential stress, tension fractures occurred. These results suggest that hydraulic fracturing in regions of significant tectonic stress may produce shear rather than tension fractures. In this case in situ stress determinations based on presumed tension fractures would lead to erroneous results. Key words: Stress hydrofracture; Hydroffacture technique.
1. I n t r o d u c t i o n
The use of hydraulic fracture data for in situ stress measurements has become increasingly popular because of their simplicity and applicability to a wide variety of situations. The hydraulic fracturing technique, which was.developed in 1947 for secondary oil recovery, consists of sealing off a section of a borehole and then pressurizing it with fluid until fracture occurs. HUBBERT and W m H s (1957) published the first theoretical study of the relations between hydraulically induced tension fractures and surrounding in situ stresses. SCrIEIDEGGER(1962), KEHLE (1964), and FAmnVRST (1964) presented further theoretical refinements. These workers recognized the effect of pore pressure in the surrounding rock formation on hydraulic fracturing, and noted that fracture fluid penetration into the rock away from the borehole could affect results. GRETENER (1965) criticized the use of hydraulic fracturing data for determining in situ stresses, noting that the distribution of fracture fluid around the borehole wall significantly affected breakdown pressure and that the effect was virtually impossible to assess accurately. HAIMSON(1968) presented the first theoretical analysis of the effect of fracture fluid penetration, which predicted breakdown pressures lower than those calculated from the simple elastic model of Hubbert and Willis. HAIMSON(1968), HAIMSONand FAIRHURST (1969), HAIMSONand STAHL (1970), and HAIMSONand EDL (1972) conducted laboratory hydraulic fracturing experiments 1) U.S. Geological Survey, Menlo Park, CA, USA.
192
P. Solberg, D. Lockner and J. Byerlee
(Pageoph,
to test the theoretical hypotheses. They conclude that all fractures produced were tension fractures and that experimental results were in accord with theoretical predictions. LAMONTand JESSEN (1963) conducted early experiments on the effects of existing fractures on the extension of hydraulic fractures. More recently ZOBACKet al. (1976) have found it possible to nullify the effects of existing fractures by the use of high-viscosity drilling mud but note that this may cause anomalously high breakdown pressures. The possibility of failure by shear, rather than tension, in hydraulic fracturing experiments has not received adequate attention. PAULDING (1968)presented a theoretical discussion of the shear failure possibility in hydraulic fracturing. LOCKER and BYERLEE(1977), in a series of laboratory experiments on the hydraulic fracturing of Weber Sandstone, found that either shear or tension fractures were produced depending on the differential stress, the rock permeability, and the viscosity and injection rate of the fracturing fluid. If shear fractures are formed during field hydraulic fracturing experiments designed to measure in situ stresses, serious errors could result if failure under tension is assumed. We have conducted the following additional laboratory hydraulic fracturing experiments with a low permeability granite and an oil shale to investigate this problem.
2. Experimental methods
The experimental method employed in this study is the same as that used by LOCKNERand BYERLEE(1977). All samples were machined into cylinders 2.54 cm in diameter and 6.35 cm in length. A 0.25 cm diameter hole was drilled along the axis to accept the hydraulic fracturing fluid. Each sample was subjected to a confining pressure of 1 kb and an additional axial load of from 1 to 7 kb to simulate hydrostatic loading and tectonic stresses. Fluid was then injected into the axial hole at 3.3 x 10- 6 cc/s to 3.3 x 10- 3 cc/s using a constant-flow-rate pump. Shell Tellus Oil No. 15 was used as the fracturing fluid for all experiments. Its viscosity ranges from 20 cp at atmospheric pressure to 90 cp at lkb of pressure. Failure was indicated by a drop in borehole pressure. Westerly Granite, with a permeability of 3.5 x 10 -s darcies at 1 kb confining pressure (BRACE et al. 1968), was used because of its physical homogeneity and relatively well known physical and mechanical properties. Eocene oil shale from the Mahogony Zone of the Green River Formation was supplied by the Rifle Oil Shale Facility of the Energy Resource and Development Agency, Rifle, Colorado. It is referred to as oil shale in this study. Oil shale was selected because it contains no measurable porosity or permeability at room temperature and pressure (J. WARD SMITH, personal communication). It is much weaker than Westerly Granite. Several oil shale specimens contained preexisting fractures that were oriented either parallel to the borehole axis or parallel to the ends of the specimen. It was decided to include
Vol. 115, 1977)
Shear and Tension Hydraulic Fractures in Low Permeability Rocks
193
these specimens in order to evaluate the effect of preexisting fractures on the hydraulic fracture mechanism.
3. Experimental results Five Westerly Granite and nine oil shale specimens were hydraulically fractured under triaxial conditions at differential stresses between 1 and 6.9 kb. At differential stresses greater than 2 kb specimens failed by shear fracture, whereas at lower differential stress they failed in tension. This relation held even in oil shale specimens that contained preexisting fractures.
1500 -
-
PUMP RATE : 3.3 xlO -5 CC/SEC
Crc=I KB - ~ ,
moo
500 g
o
~~ I
8KIB I KB 3 KB
~ 6.9 KB I
I
IO00
I
,
TIME(MINUTES)
,
,
I
2000
Figure 1 Breakdown curves for four experimentsconducted on Westerly Granite at 1 kb confining pressure and an injection rate of 3.3 x 10-s cc/s. Differential stress ranged from 1 to 6.9 kb. The experiments on Westerly Granite shown in Fig. 1 were conducted at 1 kb confining pressure and a constant hydraulic fluid injection rate of 3.3 x 10- 5 cc/s. The specimens loaded to 3, 5, and 6.9 kb differential stress failed by shear fracture, and increased levels of differential stress resulted in decreased breakdown pressures. The shear strength determined in these experiments is closely in accord with that determined by MOGI (1966) and BYZRLEE (1967) if confiningpressure is replaced with the confining pressure minus the borehole pressure. Thus, consistent with the effective stress concept, it seems that at failure the pore pressure equalled the borehole pressure in the region of fracture propagation. The Westerly samples which failed in shear were found to contain incipient fractures surrounding the borehole. HAIMSONand EDL
P. Solberg,D. Locknerand J. Byerlee
194
(Pageoph,
(1972) have shown experimentally that when the differential stress exceeds the unconfined compressive strength, localized shear fracture occurs at the borehole wall even before fluid injection. The specimen at 1 kb differential stress failed by tension fracture at a fluid pressure of 1110 bars. The hydraulic pressure-breakdown curves for the three oil shale samples run at 3.3 x 10-5 cc/s injection rate are plotted in Fig. 2. The specimen loaded to 2.5 kb differential stress failed by shear fracture, whereas those loaded to 1 and 2 kb failed in tension. Results from specimens run at injection rates of 3.3 x 10 . 4 and 3.3 x 10- 3 cc/s are consistent with those run at the slower injection rate. HAIMSONand FAIRHURST (1969) have shown experimentally that increased injection rates result in increased
ISO0~-- ~ RATE=3.3x10-SCC/SEC
_=
%-, I K B
moo--
" '
B
-
o3
Ip
""
o.~176500
2.5 KB
/i
, 0
2000 (MINUTES)
TIME
I 4000
Figure 2 Breakdowncurvesfor threeexperimentsconductedon oil shaleat 1kb confiningpressureand an injection rate of 3.3 x 10-s cc/s. Differentialstressrangedfrom 1 to 2.5 kb. breakdown pressures. ZOBACK et al. (1976) in related work suggest that anomalously high breakdown pressures at high injection rates may be due to pressure losses at the propagating crack tip. Apparently, at the slow injection rates used in our experiments such pressure losses and anomalous breakdown pressures do not occur. Both the oil shale specimens loaded to 2 and 2.5 kb differential stress contained preexisting fractures that intersected the borehole. The specimen run at 2 kb differential stress failed at a borehole pressure of 1020 bars along the axially oriented fracture. The crack reopened as hydraulic fluid pressure exceeded the 1 kb confining pressure. The specimen run at 2.5 kb differential stress contained a preexisting crack parallel to the sample end but failed at a borehole pressure of 680 bars by shear fracture.
Vol. 115, 1977)
Shear and Tension HydraulicFractures in Low PermeabilityRocks
195
The sample loaded to 1 kb differential stress failed in tension at 1217 bars. It did not contain any preexisting cracks. The specimen exhibited multiple breakdown pressures (Fig. 2) and was found to contain several small tension fractures. The opening of these may have caused transient pressure drops as the pump rate was temporarily exceeded by the ability of the fracture to accept fluid. Repetition of such a process could cause the apparent multiple breakdowns. 4. Discussion
Regardless of the fracture type (shear or tension), hydraulic fluid diffusion from the borehole into our samples apparently took place in all cases, even in seemingly impermeable oil shale. This could be due to increased permeability with increased effective differential stress. ZOBACKand BYERLEE(1975) studied the relations between permeability and differential stress for Westerly Granite and found that before failure the permeability increased by approximately a factor of 3, owing to dilatant microcrack opening. Presumably the same phenomenon occurs in our samples. Virgin oil shale may seem impermeable because pores and preexisting microcracks are filled with high-viscosity oil, but increased differential stress and resulting crack opening of even small magnitude could provide many new pathways for the hydraulic fluid and dramatically increase permeability. Samples that failed in tension did so at significantly less than twice the confining pressure, the theoretically predicted breakdown pressure for impermeable material using Kirsh's solution given by TIMOSHENKOand GOODrER (1951) and applied by HUBBERT and WILL~s (1957). GRETENER (1965) suggested that in permeable rocks hydraulic fluid diffusion away from the borehole increases the pore pressure, thus reducing the effective confining pressure and the breakdown pressure for tensile failure. HAIMSON(1968) has shown theoretically how the breakdown pressure could decrease due to fluid penetration, and HAIMSONand EDL (1972) published experimental results verifying the theoretical expectations. At the increased differential stresses at which shear fracture occurs samples become increasingly dilatant and hence permeable to the fracturing fluid. Fluid diffuses throughout the sample, which at some stage can no longer support the differential stress. Shear fracture occurs when the rock's shear strength is exceeded. It is well known that shear strength is a function of effective pressure (confining pressure minus pore pressure). As effective pressure increases, shear strength increases. Thus at a fixed confining pressure, as in our experiments, an increase in pore pressure (due to hydraulic fluid diffusion from the borehole) will cause a decrease in shear strength and therefore in breakdown pressure when failure occurs by shear. The oil shale sample loaded to 2 kb differential stress failed by the reopening of a preexisting macroscopic fracture at a fluid pressure of 1020 bars. As stated previously, Kirsh's solution predicts that the 1 kb confining pressure would cause a 2 kb circumferential concentration of stress at the borehole wall, and one might expect that 2 kb
196
P. Solberg,D. Locknerand J. Byerlee
(Pageoph,
fluid pressure would be necessary to reopen the crack. Once formed, however, a macroscopic fracture may be propped open by loosened grains on the fracture surface. Hydraulic fluid would need to flow into the crack only a short distance before the stress concentration would diminish and fluid pressure could overcome the normal stress, opening the crack. ZOBACKet al. (1976) have found that in some cases preexisting fractures become leaky at substantially lower fluid pressures than the normal stress holding the crack closed. In a similar manner the preexisting crack in the oil shale specimen loaded to 2.5 kb differential stress may have acted as an extended source of hydraulic fluid, Fluid could permeate out into the sample through the crack as well as through the borehole wall. Thus the presence of preexisting fractures tends to increase the overall permeability and favor the production of shear rather than tension fractures at sufficiently high levels of differential stress. In regions with high tectonic stresses, such as those simulated in our experiments, the existing pore pressure may be sufficient to cause shear fracturing even before fracture fluid injection. If not, but with differential stress levels greater than the unconfined compressive strength of the rock, fracture fluid diffusion away from the borehole may raise the pore pressure sufficiently to cause shear fracturing, Field hydraulic fracturing experiments designed to measure in situ stresses commonly use more rapid injection rates than those of our experiments to insure the production of tension rather than shear fractures. The experimental procedure consists of injecting the fracturing fluid into the borehole until a drop in pressure indicates fracture initiation. Pumping is continued to propagate the fracture and then stopped to record the instantaneous shut-in pressure. During shut-in, the reservoir of fracture fluid trapped in the fracture slowly diffuses out into the surrounding rock. Therefore, even if a tension fracture is produced initially, fluid diffusion out into the rock during shutin might raise the pore pressure enough to cause shear fracture in tectonically active areas. Unrecognized shear fracturing may have occurred in hydraulic fracturing experiments in the field. AAMODT(1974) discussed in situ stress determinations from the hydraulic fracturing of a 758 meter drillhole in northern New Mexico for the Los Alamos Dry-Hot-Rock Geothermal Energy Program. He concluded that two anomalous pressure drops after shut-in were the result of closing hydraulically produced vertical and horizontal tension fractures. An equally reasonable explanation however, is that only one large tension fracture was produced and that later pressure drops were the result of shear fracturing due to fluid diffusion from the tension fracture after shut-in. Recent seismic evidence from Los Alamos (AI.m~IGHT and HANOLD 1976) suggests that this may well be so. Although not a hydraulic fracturing experiment designed to measure in situ stresses, hydraulically induced fractures at the Rocky Mountain Arsenal disposal well were originally assumed to be tension fractures, but later seismic studies (HEALYet al. 1968) indicated failure in shear due to massive fluid diffusion.
Vol. 115, 1977)
Shear and Tension Hydraulic Fractures in Low Permeability Rocks
197
5. Conclusions
This study suggests that even in very low permeability rocks such as granite or oil shale hydraulic fracturing may produce shear rather than tension fractures if significant tectonic stresses exist, regardless of the presence of preexisting fractures. The measured breakdown pressures and orientations of hydraulically produced tension fractures in our experiments are in accord with the results of previous workers (HAIMSON 1968, HAIMSONand FAIRHURST1969, HAIMSONand STAHL 1970, HAIMSON and EDL 1972). New fractures are oriented in axial planes containing the borehole, as expected. If tension fractures are produced in the field, these results indicate that in situ stress determinations based on the principles developed by HUBBERTand WILLIS (1957) are useful. But under differential stresses greater than 2 kb all samples failed by shear rather than tension fracture. Unfortunately, the differentiation between shear and tension fractures in deep boreholes is difficult with present day technology. Misidentification of fracture type in field studies could lead to gross errors in the determination of both orientation and magnitude of in situ stresses.
REFERENCES AAMODT, R. L. (1974), An experimental measurement of in situ stress in granite by hydraulic fracturing, L.A.S.L. in house publication LA-5605-MS, 1~4. ALBR~GHT, J. N. and HANOLD, R. J., Seismic mapping of hydraulic fractures made in basement rocks in E.R.D.A. 2nd Annual Symp. on Enhanced Oil and Gas Recovery (Tulsa, Oklahoma 1976). BRACE, W. F., WALSH, J. B. and FRANGOS, W. T. (1968), Permeability o f granite under high pressure, J. Geophys. Res. 73, 2225-2236. BYERLEE, J. D. (1967), Frictional characteristics of granite under high confining pressures, J. Geophys. Res. 72, 3639 3648. FAIRHURST, C. (1964), Measurement of i n situ rock stresses, with particular reference to hydraulic fracturing, Rock, Mech. and Engineering Geol. 2, 129-147. GP,ETENER, P. E. (1965), Can the state of stress be determined from hydraulic fracturing data ? J. Geophys. Res. 70, 62054212. HAIMSON, B. C. (1968), Hydraulic fracturing in porous and nonporous rock and its potential for determining in situ stresses at great depth, PhD. Thesis, University of Minnesota. HAIMSON, B. C. and FAIRHURST, C. (1969), Hydraulic fracturing in porous-permeable materials, AIME Petrol. Trans. 811-817. HAIMSON, B. C. and FAmttURST, C. (1970), In-'situ stress determination at great depth by means of hydraulic )Tacturing in Rock Mech. - Theory and Practice, Proc. of l l th Symposium on Rock Mechanics (ed. Somerton, Soc. Mining Engineers of AIME), 559-584. HAIMSON, B. C. and STAHL, E. J. (1970), Hydraulic fracturing and the extraction o f minerals through wells in 3rd Symp. on Salt, Northern Ohio Geol. Soe. Cleveland, Ohio, 421432. HAIMSON, B. C. and EDL, J. N., JR (1972), Hydraulic fracturing o f deep wells, AIME, Petrol. Trans. 4061, 1-12. HEALY, J. H., RUBEY, W. W., GRIGGS, D. T. and RALEIGH, C. B. (1968), The Denver earthquakes, Science 161, 1301-1310. HUBBERT, M. K. and WILLIS, D. G. (1957), Mechanics of hydraulic fracturing, Trans. AIME 210, 153-163. KEHLE, R. O. (1964), The determination of tectonic stresses through analysis of hydraulic well fracturing, J. Geophys. Res. 69, 259-273.
198
P. Solberg, D. Lockner and J. Byerlee
LAMONT, N. and JESSEN, F. W. (1963), The effects of existing fractures in rocks on the extension of hydraulic fractures, AIME Petrol. Trans. 203-209. LOCKNER, D. and BYERLEE,J. (1977), Hydrofraeture in Weber Sandstone at high confining pressure and differential stress, J. Geophys. Res. in press. MOGI, K. (1966), Some precise measurements of fracture strength ofrocks under uniform compressive stress, Intl. J. Rock Mech. Geomech. 4, 41-55. PAULDING, B. W. (1968), Orientation of hydraulically induced fractures in Proe. 9th Syrup. on Rock Mechanics, 461-483. SCHEIDEGGER,A. E. (1962), Stresses in the earth's crust as determined frorn hydraulic fracturing data, Geologie und Bauwesen 7, 45-53. TIMOSHENKO, S. and GOODIER, J. N., Theory of Elasticity, 3rd ed. (McGraw-Hill, New York 1951). ZOBACK, M. D. and BYERLEE,J. D. (1975), The effect of microerack dilatancy on the permeability of Westerly Granite, J. Geophys Res. 80, 752 755. ZOBACK, M. D., RUMMEL,F., JUNG, R. and RALEIGH, C. B. (1976), Laboratory hydraulic fracturing experiments in intact andprefractured rock, Intl. J. Rock. Mech. Geomech., in press. (Received 28th October 1976)