Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Int. J. Rock Mech. Min. Sci. & Geomech.Abstr. Vol. 18, pp.

429 to 435, 1981 Printed in Great Britain

0148-9062,81/050429-07502.00/0 Pergamon Press Ltd

Effect of Pore Pressure and Confining Pressure on Fracture Permeability


J. B. WALSH*

The effective pressure pc for measurements of fluid permeability ( k ~ is shown to be (Pc - spp) where pc is confining pressure, pv is pore pressure, and s depends on the topography of the fracture surfaces and rock type. Measurements of flow through simulated fractures by Kranz et al. [,1-1 show that s can vary between 0.5 and 1.0. The analysis here suggests that ( k ) ~ should be linearly related to In pc. Data from studies by Kranz et al. I'1], Brat & Stesky [-2], and Jones 1,3]follow the theoretical relationship.

INTRODUCTION The permeability of a fracture to the flow of fluids decreases under increasing compressive stress because of two factors. In the first place, the aperture decreases under increasing compression, and the resistance to flow increases because of the smaller cross-sectional area. Also the number of points of contact between asperities on the fracture surfaces, and their area, increase under higher compressive stresses, and the resistance increases accordingly because of the longer and more tortuous fluid path. Fracture permeability may also vary with fluid pressure. Increasing the fluid pressure causes the fracture to open, and the permeability increases because of the larger aperture and the decreased area of contact between the surfaces. The permeability of fractures and its variation with fluid pressure and tectonic stress is a matter of considerable interest, and several studies, primarily experimental, have been made. A review of the literature on the subject is given by Iwai [4], and Gale 1,5], and experimental work carried out since then is described by Trimmer et al. 1,6], Kranz et al. 1-1-1, and Jaeger & Cook I-7]. A few field experiments on the effect of stress on fluid flow in fractures have been reported (for a survey, see Brace [,81). I have been studying theoretical aspects of fluid flow in fractures with the purpose of determining the fundamental joint parameters which control the process. I took as a starting point a recent analysis 1,9] of 'joint stiffness', i.e. the normal stress required for a unit change in separation between the surfaces. In this analysis, the fracture was assumed to be two rough surfaces, each with random topography. The stiffness of such a fracture model increases as normal compressive stresses are increased because more asperities come into contact. The theory showed that fracture stiffness * Department of Earth and PlanetarySciences, MassachusettsInstitute of Technology,Cambridge,MA 02139, U.S.A.

should increase linearly with increasing normal stress, and this relationship was found to hold to a good approximation in the one field experiment for which we could obtain data. Engelder & Scholz [,21] and Tanoli et al. 1,22] have since compared the theoretical relationship between stress and stiffness with data from experiments in the laboratory, and they find good agreement both for mechanically-prepared surfaces and for natural fractures. Goodman 1,10] on the basis of his examination of experimental data, proposed an empirical relationship which is equivalent to the one that we derived from theory. All available evidence suggests, therefore, that the deformation of fractures is adequately described by the simple model developed by Walsh & Grosenbaugh [-9]. The model is also suited for studying fluid permeability because it includes the essential elements of the process: the open area between the fracture surfaces is continuous and the effects of both the aperture of the flow channels and the obstacles formed by asperities in contact can be studied. Only the effect of applied normal stress on fracture deformation is considered in the simple theory discussed above, and the additional effect of pore pressure must also be considered in practical problems. Effective stress, that is, those combinations of applied stress and pore pressure which produce the same effect irrespective of the individual values of applied stress or pore pressure, was considered by Nur & Byerlee [11] and Robin 1,12]. Their discussion seems to have been neglected in recent descriptions of experimental results, and so I reconsider the theoretical development in the analysis below and apply the results to data published since their article. ANALYSIS

Fluid flow Consider a differential element of the flow channel between the fracture surfaces, as shown in Fig. I. Flow

429

430

J.B. Walsh
q,r+ ('B %/'b y) dy

tive 'flow conductance' ~,ik ), is (k) = - - k l+:~ (5)

where k is conductance of the fracture with no asperities and :( is the ratio of the contact area to the total area of the fracture. Fluid conductance k in (5) is, by analogy to heat flow,

k = q/(?,p/ax)
which, from (1) is
Fig. 1. Flow is two-dimensional in the x-y plane, and resistance is due to drag on the upper and lower surfaces. Flow q. and qy in the x and y-directions changes with distance because of changes in aperture.

(6)

k = (2aa/3#) Introducing (7) into (5), we find (k) = - + 11--~t

(7)

(2aa/3#).

(8)

resistance at lateral boundaries is neglected because the wetted area of these boundaries is assumed to be much less than the plan area of the flow channels. The equation governing one-dimensional laminar flow qx between surfaces separated by a small (compared with lateral dimensions) distance 2a can be found in elementary texts:

Changing the normal stress on the fracture changes both the aperture a and the contact area, as characterized by :~; differentiating (8) with respect to applied stress, p, and rearranging gives: dp = a ~ . (9)

qx = (2a3/3p)(aP/dx)

(1) Walsh & Grosenbaugh [9] show for a fracture surface having approximately random surface topography that

where (dp/dx) is the pressure gradient and/~ is viscosity. For an incompressible fluid and incompressible rock, we see (refer to Fig. 1) that conservation of mass requires that

da/dp = x,/2 hip

(10)

dqx
dx
or

aqy
+-~-y= 0

(cg/dx)(a 3 dp/dx) + (d/dy)(a 3 dp/Oy) = 0.

(2)

where h is the r.m.s, value of the height distribution. Similarly, following Greenwood & Williamson [14] and Whitehouse & Archard [15] we find that the contact area increases linearly with load for this model; that is (d~t/dp) = b = x/3rr (f/h)/E(1 - v2) (11)

For sufficiently gradual changes in aperture, that is, for O,a/dx, and da/dy sufficiently small, (2) becomes V2p = 0 (3)

Equation (3), Laplace's equation, also describes twodimensional heat flow, and so solutions derived using heat conduction theory can be applied to the problem here. In particular, one can adapt Maxwelrs technique for finding the effective conductivity of a three-dimensional model of a mixture of two conducting phases to the two-dimensional problem here. I find, by following Carslaw & Jaeger [13, p. 426] that the effective conductivity < c> of a sheet having conductivity c containing more or less circular cylindrical inclusions of material with conductivity ci is given by the expression <c) - c (c) -20t c - ci c + c~ - ~,(c - ci) (4)

where E and v are Young's modulus and Poisson's ratio for the country rock and f is the auto-correlation distance (i.e. the distance a curve must be indexed relative to itself to become uncorrelated). Substituting (10) and (11) into (9) and integrating gives

(k)/(ko) = [1 - x/2(h/ao)In (p/po)] 3

[li-

Sr b(p -~ ~o).J

b(P-P0) 1

(12)
where ( k o ) and ao are the fluid conductance and the half-aperture at some reference stress Po- The only parameters in (12) other than the primary variables ( k ) and p are h, b, and a0, which depend on the configuration of the fracture surface and the elastic moduli of the country rock. Once these parameters have been measured for a specific fracture, (12) can be used to show how flow through the fracture responds to changes in the applied stress. Note that both of the factors which affect flow are represented in (12): the first term in brackets is aperture and the second term in brackets is tortuosity.

where 7 is the ratio of the area of the inclusions to the total area. Here, we are interested in the effect of contacting asperities on flow, and so the 'conductivity' c~ in (4) is zero. As shown above, two-dimensional heat flow in a medium with constant conductivity is entirely analogous to laminar flow of an incompressible fluid. Making use of this analogy, we find from (4) that the effec-

Effect of Pore Pressure and Confining Pressure on Fracture Permeability


Effective stress

431

The effect of changes in the pressure of the fluid in the fracture, or the effect of changing both the applied stress and the pore pressure simultaneously, may be of interest in some cases. In such cases, the concept of dvj, = (Ovp/~p)(dpc - dpp) + vpfl, dpp (15) effective stress provides a convenient means for treating For no change in pore volume, dvp = 0, and so the many different combinations of pore pressure and confining stress. An 'effective stress law' defines all the 0 = (c~v,/Op)dp, - [(Ov,/~p) - v, f l , ] d p , (16) combinations of confining stress and pore pressure The change in pore volume when the pore pressure is which produce the same effect. As Nur & Byerlee [11] and Robin [12] point out, the expression for effective zero provides one boundary condition for the effective stress depends upon the process which is being studied. pressure dp : Here, where fluid flow is being considered, we are interdp~(Ov/Op) = (~v/~p)dp, - (~v/~p - v, f l , ) d p , ested in deriving an expression giving the combinations of pore pressure and confining stress for which the flow o r rate is the same. For the flow rate to be the same, the dp = dpc - [1 - vpfl,/(Ovp/~p)]dpp (17) configuration of the flow passages must be the same. Refer to Fig. 2, where, in (a), the hydrostatic confining The term in brackets in (17) is constant for materials stress on the fracture is changed by dpc and the pore which are linearly elastic, and for this case integration pressure is changed by dpp. In (b), the differential gives changes in pore pressure and confining stress are deP~ = Pc - spp (18) composed into the sum of two elementary stress states where s = 1 - vpfls/(~vp/Sp). which is the equivalent of (a). Equation 18 was previously derived by Robin [12], Because of the symmetry of the applied stresses, and it is a fundamental parameter in the poro-elastic changes in the shape of the flow passages need not be considered, and flow rates are equal for stress states for theory developed by Biot [16] and by Rice & Cleary which the volume of the flow passages are equal. The [17]. Robin [12] rightly pointed out that the term in change dv t of the pore volume for the first stress state brackets is not necessarily constant for rocks. Nevertheless, we make the approximation here that the brackin Fig. 2(b) is eted term is nearly constant; the discussion below comdvt = vpfls dpp (13) paring predicted behavior with data from several exwhere vp is the pore volume, fls is compressibility of the periments shows that the approximation is adequate, at rock surrounding the fracture. The change dv2 in pore least over the range of pressures that were used. volume for the second stress state in Fig. 2(b) can be written for a linear material in the form: COMPARISON WITH DATA
dr2 = (Ovp/~p)(dpc - dpp)
(a)

where (Ovp/Op) is the rate of change of pore volume with applied hydrostatic pressure for a joint with no pore fluid. The net change in pore volume from (13) and (14) is

(14)

cipc

,/"
dp

A crude analysis which I have not included here suggests that the second term in brackets in (12) is smaller than the first term except at very high pressures. Accordingly, I make the assumption, subject to experimental verification, that (12) can be written in the form
(( k >/( ko >)~ ~- 1 - ( \ / 2 h / a o ) l n ( p / p o ) e

(19)

t
(b)

ld(Pg%)

I dp"

-.--+ d (p~" p . )

,al,.....

Fig. 2. In the first diagram, the external stress p acting on a fracture and the pressure pp of fluid and changed by differential amounts. These changes can be considered to be the superposifion of the two stress states shown in the lower part of the figure.

where subscript e refers to effective pressure. Note that the pressures in (10) are effective pressures in accordance with the discussion in the previous section. We see in (19) that the cube-root of flow rate should vary linearly with lnpe. Experiments on artificial fractures carried out by Kranz et al. [1], Brar & Stesky [2], and Jones [3] provide appropriate data for testing these predictions. Kranz et al. [1] made experiments in which both pore pressure and confining pressure were varied, and so these experiments can be used to test both the effective stress law in (18) and the pressure dependence given by (19). A three-stage procedure was used. First, I plotted k -~ as a function of pore pressure at fixed values of confining pressure (Fig. 3(a) is such a plot for a joint made by a tension fracture in Barre granite). Next, I cross-plotted these data, producing curves of Ps vs p, for constant values of k (see Fig. 3(b)). If the

432

J.B. Walsh

(a)

k" i/3

_ 124

\..

Iv~

\.
A

- - 0

I I I

200

50

I00

200

Pore pressure, MPo

(b)
160

140

(c)
120

~00

8o

k~3

40

% "X

20

\
I
200

I I00

I IZO

I 140

I ~

I 180

L 30

I00

Gonfininq pressure, MPa

Effective pressure, MPa

Fig. 3. (a) Data from experiments on a tension fracture in Barre granite [1] are plotted as k-~ in order to simplify interpolation. Dashed lines refer to cross-plots in Fig. 3(b). (b] Data from Fig. 3(a] is cross-plotted as shown to find the expression for effective pressure for this fracture. All curves have the same slope (0.56], as suggested by the theory, giving an effective stress P, = Pc - 0.56pl. (c) Data at all confining pressures and pore pressures are plotted as k ~'vs In p, = In(pc - 0.56 p~). All data points fall on a straight line, in agreement with the theoretical relationship (19).

effective stress law g i v e n b y (18) is valid, the c u r v e s in this plot s h o u l d b e a f a m i l y of p a r a l l e l s t r a i g h t lines; further, the s l o p e of these c u r v e s is the p a r a m e t e r s. W e see in Fig. 3(b) t h a t the c u r v e s are i n d e e d parallel s t r a i g h t lines with a s l o p e s = 0.56. T h e effective stress

law for this t e n s i o n f r a c t u r e is therefore Pe = Pc - 0.56 P l I n Fig. 3(c), I h a v e r e p l o t t e d o r i g i n a l d a t a as k ~ v s l n p e . W e see f r o m (18) a n d (19) that d a t a at all

Effect of Pore Pressure and Confining Pressure on Fracture Permeability

433

pore pressures and confining pressures should fall on a straight line. The data in Fig. 3(c) are linear as predicted. The slope of the curve in Fig. 3(c) can be used to find the ratio of the r.m.s, height h to the aperture ao at it~s some reference pressure P0. I find from Fig. 3(c) that h/ao "" 1 where ao is the aperture at an effective confining pressure of 100 MPa. Kranz et al. [1] also carried out flow measurements "'--..,,,..,,,. on artificial joints made by joining two machined and polished surfaces of Barre granite. I analyzed data from two of these experiments and found that the results I I I I I I I | I I I I I I I ! 5 IO 20 50 followed the effective stress law given by (18) with a Confining pressure, MPa coefficient s equal to approximately 0.9 for both joints. Flow rate plotted as k~ is shown as a function of In pc Fig. 5. Flow data from experiments by Brat & Stesky [2~]. Pore in Fig. 4 for joint surfaces polished with 600 grit pressure was approximately atmospheric pressure, and so effective powder. As in the previous case, the data are found to pressure and confining pressure are nearly the same. Data fall on a straight line, in agreement with the theoretical relationship (191. follow a linear trend, in agreement with the theoretical relationship (19). Brar & Stesky [2] studied joints which were made by between effective pressure and confining pressure was joining ground surface samples of a variety of rocks. negligible. Jones found that data for all fractures could Permeability, fracture aperture, and electrical conducti- be represented by an empirical equation having the vity were measured as a function of confining pressure. form k ~ = A - Blnpc; note that this is the same as The pore pressure was negligible in these experiments, (19) which was derived analytically for the fracture and so confining pressure and effective pressure had model under consideration in this analysis. very nearly the same value. Brar & Stesky [2] found that the aperture closed linearly with increasing log Pc, DISCUSSION in agreement with the response predicted theoretically by Walsh & Grosenbaugh [9]. Flow measurements for I derive in the above analysis an expression (12) a typical case (a joint in a pyroxene granulite sample) which describes how fluid flow through a fracture are shown in Fig. 5 plotted as k ~ vs In Pc. Again, the varies as a function of pore fluid pressure and external observed behavior follows the linear relationship pre- confining pressure. A fracture is assumed to be two dicted theoretically. I find from Fig. 5 that h/ao is ap- rough surfaces in contact. Resistance to flow through proximately 0.4, where a0 is the aperture at 370 bars. the fracture is caused by viscous drag of the fluid in the Brar & Stesky's I2] measurements show that the aper- narrow openings between the surfaces and by the torture at 370 bars is almost 10/tin, and so the r.m.s. tuosity of the flow path as fluid is diverted around height of the asperities on these surfaces is about 4/~m. asperities in contact. The analysis shows that flow rate Finally, consider Jones' 1'3] study of flow through in (12) can be expressed as the product of two factors, simulated fractures in carbonate rocks and materials one giving the effect of aperture and the other giving such as cement and plaster of Paris. Pore pressure was the effect of tortuosity. approximately atmospheric and confining pressure The effect on flow rate of independent changes in ranged as high as 140MPa, and so the difference pore pressure and confining pressure can be described in terms of changes in 'effective pressure'. Robin [12] showed that effective pressure Pe in flow problems is related to confining pressure Pc and pore pressure pp by the expression pe = Pc - spp where s is given by (18). F o The value of s is taken to be 1.0 in most presentations of data that are found in the recent literature. An imkh portant conclusion to be drawn from the work that I present here is that the value of s is not 1.0: s is about F 0.9 for joints with polished surfaces in the study by Kranz et al. ['1] and s is only 0.56 for a joint made from a tension fracture. The expression for s in (18) depends on how the I I I I i 1 ! i t volume (and shape) of the flow passages changes with )o 20 50 lOg confining pressure when pore pressure is zero. To get Effective pressure, MPo an idea of how s depends on the shape of the flow passages, consider, as a crude approximation to a chanFig. 4. Data from experiments by Kranz et al. [1] on flow between nel in a real fracture, flow through a cylindrical channel granite surfaces polished with 600 grit compound. The effective stress, which is found following the procedure outlined in Fig. 3, is in an elastic continuum; take, as extreme cases, a Pe=Pc-0.91pp. cylinder with a circular cross-section and one with a

-.<

434

J . B . Walsh with effective pressure according to the law k ~' ~ A - B I n p e where A and B are constants which depend on certain geometric parameters describing the microcracks and their surface topography. Jones & Owens [19] find empirically that k '~ is linearly related to lnpe for the low permeability gas sands that they studied, and so apparently the complications which arise when one extends the analysis from two-dimensional flow to three-dimensional flow do not affect the fundamental aspects of the process. Several other experimental studies of the fluid permeability of various rock types have been made (e.g. [20], and I plan to apply the results of the present analysis to these data.

very elongated (crack-like) elliptical cross-section. The expression for s in (18) can be evaluated by following s t a n d a r d texts on the theory of elasticity. I find for a channel with a circular cross-section that s = 2(1 + 2v)/(5 - 2v) where v is Poisson's ratio for the country rock. The value of s therefore must fall in the range 0.4 ~< s ~< 1.0 for possible values of Poisson's ratio 0 ~< v ~< 0.5 Poisson's ratio for granites is typically near 0.15 at low stress, and so s in the experiments by K r a n z et al. [1] on joints in Barre granite is 0.5-0.6. O n the other hand, s for channels that are crack-like is found (see [18]) to differ from unity by a term which is the order of ~, where ~, the aspect ratio of the elliptical cross-section, is a very small number. I conclude from this simple analysis that values of s near 1.0 are indicative of narrow, crack-like flow passages, and lower values of s indicate passages with more equant dimensions. The measurements by K r a n z et al. [1] appear to agree with this interpretation: the value of s is low (0.56) when the flow passages are the interstices between the jagged surfaces formed by a tensile fracture, and the value is near unity (0.9) for joints with smooth, g r o u n d surfaces. Values of s greater than unity have been measured (A. Nur, personal c o m m u n i c a t i o n , 1980) in experiments on sandstone samples, and values as high as 5.0 were observed in some cases. N u r suggests that the presence of clays in the pore spaces cause these high values and this explanation seems reasonable. The model used in the analysis here does not apply to rocks having pores partially filled with a low m o d u l u s material like clay, and clearly the results here should not be applied to these m o r e complicated cases. As mentioned above, the expression (12) describing the relationship between flow rate and effective pressure is the p r o d u c t of two terms, one giving the effect of aperture and the other the effect of tortuosity. All the flow data that I examined show that the effect of tortuosity can be neglected, i.e. k ~ is linearly related to In pc. The reason that tortuosity has a relatively small effect on flow rate is evident in (12), where we see that the expression for aperture is raised to the third power, whereas tortuosity is not. O n the other hand, I expect that tortuosity will be found to have a significant effect on the electrical conductivity of fractures when these data become available. One can show following the derivation that I present above that the expression for the electrical conductivity of fluid-saturated fractures is the p r o d u c t of the same two factors in (12), except that both factors appear raised to the first power. Both factors have the same importance, and so tortuosity p r o b a b l y c a n n o t be neglected as it is in the analysis of fluid flow. Walsh & G r o s e n b a u g h [9] found that their analysis of fracture deformation could be used to study the compressibility of country rock. They showed that each microcrack in a rock acted, in effect, like a miniature fracture, and the overall deformation of a sample could be found by s u m m i n g the contribution of each of them. If we apply this principle to the analysis here, we conclude that the permeability k of a rock sample changes

Acknowicdgements---N. S. Brat and R. M. Stesky kindly offered me the use of their unpublished data on joint permeability and deformation, and R. Kranz provided information on his experiments which I needed in the analysis. Discussions with W. F. Brace on experimental results were especially helpful. This research was supported by the Army Research Office, Durham, under contract DAAG29-79-C-0032.

Received 12 November 1980; in revised form 12 January 1981.

REFERENCES I. Kranz R. L., Frankel S. D., Engelder T. & Scholz C. H. The permeability of whole and jointed Barre granite. Int. J. Rock Mech. Min. Sci. & Geomech. Abstr. 16, 225-234 (1979). 2. Brar N. S. & Stesky R. M. Permeability of intact and jointed rock. Eos 61(46), I 112 (1980). 3. Jones F. O. A laboratory study of the effects of confining pressure on fracture fiow and storage capacity in carbonate rocks. J. Petrol. Technol., pp. 21-27 (January, 1975). 4. lwai L. Fundamental studies of fluid flow through a single fracture. Ph.D. Thesis, Univ. of California, Berkeley, 208 pp (1976). 5. Gale J. E. A numerical, field and laboratory study of flow in rocks with deformable fractures. Ph.D. Thesis, Univ. of California, Berkeley, 225 pp (1975). 6. Trimmer D., Bonner B., Heard H. C. & Duba A. Effect of pressure and stress on water transport in intact and fractured gabbro and granite. UCRL Preprint 83932, Lawrence Livermore Laboratory (February, 1980). 7. Jaeger J. C. & Cook N. G. W. Fundamentals of Rock Mechanics, 513 pp, Methuen, London (1969). 8. Brace W. F. Note on permeability changes in geologic material due to stress. PAGEOPH 116, 627-633 (1978). 9. Walsh J. B. & Grosenbaugh M. A. A new model for analyzing the effect of fractures on compressibility. J. geophys. Res. 84(B7), 3532-3536 (1979). 10. Goodman R. E. Methods of Geological Engineering in Discontinuous Rocks, 472 pp, West, New York (1976). 11. Nur A. & Byerlee J. D. An exact effective stress law for elastic deformation of rock with fluids. J. geophys. Res. 72126), 6414-6419 (1971). 12. Robin P.-Y. F. Note on effective pressure. J. geophys. Res. 78(14), 2434-2437 (1973). 13. Carslaw H. S. & Jaeger J. C. Conduction of Heat in Solids, 510 pp, Oxford Univ. Press, London (1959). 14. Greenwood J. A. & Williamson J. B. P. Contact of nominally flat surfaces. Proc. R. Soc. Lond. A 295, 300 (1966). 15. Whitehouse D. J. & Archard J. F. The properties of random surface of significance in their contact. Proc. R. Soc. Lond. A 316, 97 (1970). 16. Biot M. A. General theory of three-dimensional consolidation. J. appL Phys. 12, 155-164 (1941). 17. Rice J. R. & Cleary M. P. Some basic stress diffusion solutions for fluid-saturated elastic porous media with compressible constituents. Rev. Geophys. Space Phys. 14(2), 227-241 (1976). 18. Walsh J. B. The effect of cracks on the uniaxial elastic compression of rocks. J. geoph.vs. Res. 70(21, 399---411(1965).

Effect of Pore Pressure and Confining Pressure on Fracture Permeability


19. Jones F. O. & Owens W. W. A laboratory study of low per, meability gas sands. Soc. Pet. Eng., A.IM.E., Paper SPE 7551, 10 pp (1979). 20. Brace W. F., Walsh J. B. & Frangos W. T. Permeability of granite under high pressure. J. geoghys. Res. 73(6), 2225--2236 (1968).

435

21. Engelder T. & Scholz C. H. Fluid flow along dry smooth joints at effective pressures up to 200 MPa. Monograph 25, Am. Geophys. Union (1981). In press. 22. Tanoli S. K., Stesky R. M. & Brar N. S. Fracture closure under uniaial stress. Eos 61(46), 1113 (1980).

You might also like