Professional Documents
Culture Documents
PREPRINT
J. G. Berryman
H. F. Wang
January 1995
James G. Berryman
Lawrence Livermore National Laboratory
P. O. Box 808 L-202
Livermore, CA 94551-9900
and
Herbert F. Wang*
Department of Geology and Geophysics
University of ~Visconsin
1215 West Dayton Street
Madison. WI 53706
Phenomenological
equationsfortheporoelastic of a duaJporosity
behavior medium arefor°
mulatedandthecoefficients
in theselinearequationsareidentified.Thegeneralization from
thesingleporosity
c~.se
increases thenumber
of independent
coe~cientsfromthree to sixforan
isotropic
applied
stress.
Thephysical interpretations
arebaseduponconsiderationsof different
temporalandspatialscales.Forveryshorttimes,bothmatrixandfractures behavein an
undrainedfashion.
Forverylongtimes, thedouble porositymediumbehaves as an equivalent
singleporosity
medium.At themacroscopicspatial level,thepertinent parameters {suchas
thetotalcompressibility)
maybe determined by appropriatefieldtests.At themesoscopic
scalepertinent
parametersof therockmatrix canbe determined directlythrough laboratory
measurements
on core,andthe compressibility canbe measured for a singlefracture. All
sixcoefficients
aredeterminedfromthethreeporoelasticmatrixcoefficientsandthefracture
compressibility
fromthesingle assumption
thatthesolidgrainmodulus of thecomposite is
approximately
thesameas thatof thematrix fora smallfracture porosity.
Underthisassump-
tion,thetotalcompressibility
andthree-dimensional
storagecompressibility
of thecomposite
arethevolumeaverages
of thematrix andfracture contributions.
[. INTRODUCTION
The equations for deformation-coupled flow in a single porosity mediumare easily derived
from Biot’s equations of poroelasticity by taking the low frequency limit, assumingonly that in-
ertial effects and second derivatives with respect to time are negligible. If the solid displacement
is ui and the fluid pressure is p (positive in compression}, the resulting equations are
3K 3(1 - 2v}K
2(l+v) eJ+ 2(l+v) Ui’jj’-aP’i’ (1)
where the solid dilatation is e = ~tj.j, K and v are the drained bulk modulus and Poisson’s
ratio, respectively, and
(2)
¢tP.ii - Bh.,,[~ + ~.
where K~, is the undrained bulk modulus and B is Skempton’s coefficient [Skempton, 1954].
The dots indicate time derivatives. The remaining constants are the Biot-Willis parameter
c~ = 1 - K/K~ = (1 - K/K,,)/B. the permeability k. and the fluid viscosity #. [Another
constant we will need, but that does not appear explicitly here. is the undrained Poisson’s
ratio u~,.] A detailed discussion of all these constants will be given in the next section. These
equations are well knownand maybe found in Rice and Clcary [1976], Cleary [1977]. etc.
The generalization of (1) and (2} to double porosity media is straightforward. First,
assume that there are two types of porosity and corresponding pressures p{D and p{~) within
the fluids contained in each pore type. (See Figure 1.) Then, to generalize the equation for the
solid displacement, we merely change the forcing term on the right hand side to allow for more
components, so (1) becomes
3K 3(1 - 2v)K
a{~}.(~} (3)
where we have introduced phenomenological constants ~(l) and (~) whose p recise p hysical
significance must be determined. An important characteristic of this equation is that each
poiat in space nowhas two fluid pressures associated with it, and therefore these pressures are
{not the true microscopicpressures in the fluid, but actually) averages over sorae representative
volume element. Similarly, Khaled et al. [1984] write two equations in place of (2) having the
forms
IZ ’ Bi~)1)
h’(.,
and
B(.~_~(~i~t~ (,5
) + c~(-~)~+ ~,.(p(.a) _ ply)).
and we will show howto interpret their coefficients in terms of a more complete derivation.
The coefficients k(~) and k(~) are permeabilities associated with matrix and fracture poros-
ity, respectively. The remaining factors are straightforward generalizations of the constants
appearing in (2). Long time analysis to be presented later shows that k(1) + k(~) =
also provides other relations amongthe various constants. The terms proportional to pressure
differences have been introduced to drive the pressures in the two types of porosity towards a
single equilibrium pressure that will be approached at long times. Moredetailed derivations
of these equations may be found in the papers by Wilson and Aifantis [1983], Khaled et al.
[1984], and Beskos and Aifantis [1986]. In particular, note that setting K = a(~) = o~(~)
results in the classical double porosity model of Barenblatt et al. [1960]. The main point to be
emphasizedhere is that these equations are just linear relations between stresses and strains.
Except for questions about the neglect of possible cross-coupling terms (.which we showlater is
a real issue but fortunately a small one), the form of the equations is not in doubt: however,
the meanings and values of various coefficients may be nontrivial to deduce from the physics
and mechanics of the underlying microscopic problem.
Theremainder of the paper is devotedto a careful analysis of the precise physical significance
of the various parameters appearing in equations (3)-(5}.
In the absence of driving forces that can maintain pressure differentials over long time peri-
ods, double porosky models ~nust reduce to single porosity models in the long time limit when
the matrix pore pressure and crack pore pressure beco~ne equal. It is therefore necessary to
remind ourselves of the basic results for single porosity modelsin poroelasticity. So, one impor-
tant role these results play is to provide constraints for the long time behavior in the problems
of interest. A second significant use of these results (which weaddress later in this paper) arises
when we makelaboratory measurementson core samples having properties characteristic of the
matrix material. Then the results presented in this section apply specifically to the matrix
stiffnesses, porosity, etc.
For isotropic materials and hydrostatic pressure variations, the two independent variables in
linear mechanics of porous media are the contining (external) pressure Pc and the fluid (pore)
4
pressure pl. The differentiaJ pressure P,l = Pc - Pl is often used to eliminate the confining
pressure. The equations of the flmdamental dilatations are then
for the fluid volume VI. Equation {6) serves to define the various constants of the porous
solid, such as the drained frame bulk modulus g and the unjacketed bulk modulus K~ for the
composite frame. Equation (7) de~nes the jacketed pore modulus h’p and the unjacketed pore
modulus g~. Similarly, (8) de~nes the bulk modulus KI of the pore fluid.
Treating ~p~ and ~p/ as the independent v~riabhs in our poroelastic theory, we deiine
the dependent variables #e e ~V/V and ~ ~ (6¢g - ~~)/g, both of which ~re positive on
expansion, and which ~re respectively the total volume dilatation and the increment of ~uid
content. Then. it bllows directly from the de~nitions and from (6), (7), and (8)
( 11
6Pc 6(=0
and is therefore given by
L
B = 1 + g’p(l/K
s - tlK, )’ (12)
It follows immediately from this definition that the undrained modulus Ku is determined by
(also see Carroll [19S0])
h"
K,, -- l - c~B" (13)
where we introduced the combination of moduli knownas the Blot-Willis parameter a ---- 1 -
h’/h’~. This result was apparently first obtained by Gassmann[1951] (t hough not in this form~
for the case of microhomogeneousporous media (i.e.. h’~ = h’d = K,,, the bulk modulusof the
single mineral present.) anti by Brownand Korringa [1975] and Rice [1975] for general porous
media with multiple minerals as constituents.
Finally’, we condensethe general relations from (9) together with the reciprocity" relations
[Brown and Korringa. 1975] into symmetric form as
S -- = (15)
"[’his storage compressibility is the change in increment of fluid content per unit change in the
fluid pressure, defined for a condition of no changein ezternal pre, ssure. It has also been called
the three-dimensional storage compressibility by" Kfimpel [1991].
Wemayequivalently eliminate the Blot-Willis parameter ~. and write ( 14 } in terms of the
undrained modulus so that
Equ~ttion ( 16} has the advantage that all the parameters trove very well defined ph.vsical inter-
pretations, and are also easily’ generalized for ~ double porosity model. Finally, note t.hat (14)
shows that Kp = tK/a. which we generally refer to as the reciprocity relation.
The total strain energy functional (including shear) for tiffs problem maybe written in the
form
2E = br,:be 0 + (17)
where b’eij is the change in the average strain with 6eli =~ be being the dilatation. (rii being
the change in the average stress tensor for the saturated porous mediumwith ~brii = -~Pc. It
follows that
OE
and
OE
~ps= 0(~¢)’ (19)
bothof which arealsoconsistent withBetfi’s
reciprocal
theorem[Love,1927]sincethematrices
in (14)and (16)are symmetric. The shearmodulusG is relatedto the bulkmodulus
Poisson’s ratioby G - 3(I - 2v)K/2(l+ v).Then,it followsthatthestressequi]]brium
equation is
Fromthese equations, (1) and (2) maybe easily derived using the identity
( p e
K,, -K - ( BK.)2 + BK----~’ (22)
which follows easily from (16). Equation (22) is used to eliminate the confining pressure
from the equations.
~Ve ,low assume two distinct phases at the macroscopic level: a porous matrix phase with
tile effective properties K{U, G(I), h’~1, ¢{1} occupying volume fraction V(1)/V = v{Uof the
total volume and ¯ macroscopic crack or joint phase occupying the remaining fraction of the
volumeV(2}/I" = v{2) = 1 - o(x). The key feature distinguishing the two phases .-- and therefore
requiring this analysis -- is the very high fluid permeability k~221 of the crack or joint phase
and the relatively lower permeabi~ty k(tO of the matrix phase. Wecould also introduce a third
independent permeabifity k02) = k{2~) for fluid flow at the interface between the in~trix and
crack phases, but for simplicity we assume here that this third permeability is essentia~y ~he
same as that of the matrix phase, so k{~2) = {~x}.
k
Welmve three distinct pressures: confining pressure 8pc. pore-fluid pressure ~p~}. and
joint-fluki pressure ~p~}. Treating ~pc,~p}~}, and *p~2’ as the independent variables in our
double porosity theory, we define the dependent variables ~e ~ $V/V (as betbre), 6~(1}
(~)- ~v~lJ)/V, and ~({2) = (~,.~2) _ ~1..~21)/i,,, which are respectively the total volume
dilatation, the increment of fluid content in the mntrix phase, and the increment of fluid content
in the joints. Weassume that the fluid in the matrix is the sanle kind of fluid as that in the
cracks or joints, but that the two fluid regions maybe in different states of ~verage stress and
therefore need to be distinguished by their respective superscripts.
Linear relations amongstrain, fluid content, and pressure then take the general form
where unbazred quantities refer to one experhnent and barred to ~mothcr experiment to show
t hat
Hence. a~-3 = o3~. Similar arguments have often been used to establish the symmetry of the
other off-diagoual components. Thus, we have established that the matrix in (23) is completely
symmetric, so we need to determine only six independent coefficients. To do so, we consider
a series of gedankenexperiments, including tests in both the short time and long time limits.
The key idea here is that at long times the two pore pressures must come to equilibrium
(P~) = P~") = as t - ~ .~) as l ongas th e c rosspermeability k ivy) is fin ite . However, at ver y
short times, we may assume that the process of pressure equilibration has not yet begun, or
equivalently that k(12} = 0 at t = 0. Wenevertheless assume that the pressure in each of the
two componentshave individually equilibrated on the average, even at short times.
Weshould emphasize that lhese are thought experiraents, and as such maynot necessarily
be realizable in the laboratory in all cases.
Also, note that the e~stence of a second pore pressure and increment of fluid content
leads to the definitions of several Skempton-likeor Biot-Willis-like coefficients. :[’lie somewhat
complicated notation we introduce will attempt to emphasize the defining boundary conditions
for the various cases. Wewill clarify these differences in the Discussion Section.
0 = an~p~ + a.n6p~~ ~)
+ a~abp~
0 = al3¢~pc + ~/23~p~1) ~ a33~p "~).
Defining
8
we can solve (26) for the two Skempton’scoefficients and find the results
a22a33
](I} a23a13
3 -- _a12a33
a]
EB -- (2S)
and
/j(2) a23a12 -- a13a22
EB "-- a22a33 _ t~223
(29)
Similarly, the effective undrained modulusfor the matrix phase is found from (31) to be deter-
mined by
Notice that if a23 = 0 then (28) and (32) are the same.
6P~2)
I _ a31. (35)
Similarly, the effective undrained modulusfor the joint phase is found from (34) to be deter-
mined by
(36)
K[u(~)] = ~l=.;p’)"=o = at1 + a~3B[u(~)].
Wem;~y properly view Eqs. (32), (33), (35), and (36) as "defining" relations among
parameters.
Notice that if a23 = 0 then (29) and (35) are the s;~me.
1
all =
where K is the overall drained bulk modulus of the system including the fractures.
~; -- 6((~) + ,~(’(~)
(38)
assumingthe total mass of Ihfid is confined. ’.[’hen, it follows that
be = -all’P,:- (a12
0 = -(a~ ~ az~)~p¢ - (a2~ + a~3 + aa: + a33)~p],
(39)
showingthat the overall pore-pressure buildup coefficient is given by
I0
Similarly, the undrained bulk [nodulus is fouud to be given by
= all+Cat2+ (41)
1~K-IlK.
a12 + at3 = = -a/K, (43)
B
while (33) and (37) show
0}]
a12= -- 1/K - 1/K[u -_
(44}
S[?./.(1)]
and similarly (36) and (37) show
Relations such as (46) showing dependencies a,nong the various constants are useful because
they show that consta,lts potentially difficult to measure (such as the undrained joint modulus
K[u(2)]} can actually be determined from other more easily accessible data.
These results showthat "all the constants in the first row of the nlatrix in (23} have nowbeen
determined in tcrrns of quantities that could in principle be measured. Similar manipulations
give the remaining coI~stants as we will show.
11
¯ t.8 Summaryof reswlts
Combiningthese results, we obtain the following general relations
(47)
~(1)
¢c,~
=zT(’~ ~I+(~,.~
~(I )~(2))P~’~)
+~[,,(,~]A.[,,(~]~A----:---
(-19)
and
-~(’2) (’2)
+( a..,:{
~L)~(~) )P
a~e + ~[
c(") ~,c~l]A[,(~l]" ~--=-
A (50)
These ecLnations should be compared to (-l) and (5). A detailed comparison of differences
nume,ical predictio,ts is beyond the scope of the present paper, but will be pursued in future
work.
L2
LABORATORY MEASUREMENTS ON CORZ SAMPLES
I, I a{~)/B(,})
_~{1)
(~e(1)-(~(~’{1)/W{1))
-"h’O)l_[ -a( ( (53)
These constants cart all be estimated [Rice and Cleary, 1976; Detournay and Cheng, 1993] or
found in the laboratory by ~nalysis of core samples of matrix material [Hart and Wang,1995].
This fact is important because it suggests another way of identifying certain combinations of
the general double-porosity coefficients in (23}.
’lb obtain the connecting equations, first note that
~e = ~ ~p" = vO}~V0}V0---~
+ -~’~V(~)= v0)~e0)+ v(’~)~e(~}" (54)
and
v v v
The final relation follows from the identity/iV {’~) = ~V~
u}, since the fracture volume is void
space.
Thus. to obtain the desired relations in a form analogous to {53). we must rewrite (23}
where the third diagonal element has been modified to e~minate the fluid contribution such
that
Nowwe consider another thought experiment: Suppose the confining pressure is equal to
the fracture pressure so ~Pc = ~P~}. "[’his situation mimicsthat of the matrix core sample in a
laboratory experiment by completely surrounding the matrix material with a uniform pressure
13
field. (See Figure 2.) Then,by combining the appropriate rows and columns, we can telescope
the 3 × 3 system down to a 2 × 2 system of the form
) (59)
~,-~(t~)/v {1) = av~ + a2a a22
So, except for an overall factor of vO) = V(I}/V (which is generally very close to unity in the
cases of interest), equations (53) and (59} are of the same form. Weare therefore led to
identifications
-v(l)a(~)//((~) - ) 4,63)
-/f((2) -c~/h" - a12 -c(l)c¢(1)//t "(t) - ass _,Sp~)
where
a~ = (65)
Combining (60) and (61) then shows
~(~}
~22 -- (122 -- ~ (69)
1.1
6. FRACTURES IN A R, IG[DLY CONSTRAINED MATRIX
In thelastsection,
setting
p(~)= pc created
a situation
wherethematrix
was completely
surroundedby an effective
confining
pressure = 2).
pc pl Thistrickessentially isolated
the
matrix materi~l fromthefracture behavior andpermitted a simple identification of certain
combinations of thedouble-porosity parameters in termsof them~trix constants.
Nowwe wouldfiketo isol~te thethefractures justas we isolated thematrix in thelast
section. Unfortunately, thesametrickdoesnotapply. Setting p~l)= Pc doesnotgenera~y
isolate thefracture (although theremightbe p~rticular geometries whereit does), because
neither theconfining pressure northematrix fluidpressure arcapplied directly to thesurface
of thefractures. So instead we mustconsider somealternative thought experiment to achieve
thedesireddecoupfing, ff we consider an experiment on a doubleporosity mediumso that
the m~trix material remains rigid (i.e., ~e(~) = 0) while the fractures deform, this condition
specifies a p~rticular choice of the m~trix fluid pressure p~) = p* (but norm~Hyp" ~ pc)
we will show that the desired separation is then accomplished.
Consider equation (57). To determine whether ~ particular set of stresses results in the
m~trix remaining rigid or not. we need an equation isolating the nmtrix strain 6e(1} from the
overall strain be. Since ~e = v(1)~e{~} + v{~}~e{~), it follows directly from {57} that we have zero
matrix strain when
This final equatity (setting the right hand side equal to zero) provides the needed rigidity
condition. Solving the resulting equation for 8p~t) in terms of 8pc and ~p~2}provides the relation
required to determine the value of matrix fluid pressure p~l) = p. needed to guarantee overall
matrix rigidity (see Fig. 3). Then, weobtain the desired equation for t’{216e(2) as a function of
6p~ and
= (-opc) + , el ..
a~2 + a3~ ax2 + a32
The definitions of the constants for the fracture system in a rigid matrix (Bert) = 0)
given by
( = ~(’)/B(’))( ))"
[72)
It is important to recognize that, whereas B(~) and K(~) have their well defined (standard)
interpretations for the fracture phase as experimental observables, ~(~) is nowa parameter that
strictly speaking does not have the usual Biot-Wil~s [1957] micromechanical interpretation
in terms of frame and gr~in modulus since there is no grain modulus associated with the
fracture phase. Wecan nevertheless define ~n effective grain modulus according to
ff(~)/(1 - ~(:)), but care should be taken not to overinterpret this parameteL With
definitions, weare finally led to the identifications:
15
and
Note that. if a23 vanishes, the right hand sides of these equations reduce to -au~ and ~’3~,
respectively.
The analogous calculation of the fracture fluid increment 6~(2) shows that
Since reciprocity showsthat the oil-diagonal terms of (72) are equ,’fi, we obtain a condition
the coefficients by equating the numerators of these expressions, showing that
a(~)~)
h’!
a~- h’(~) K., (77)
7. DISCUSSION
16
Thetermsa22,a23, a32 = a23,anda33aregeneralized storage coefficients,
i.e.,a~jis the
volume of fluidthatflowsintoa control volume (normalizedby thecontrolvolume)of phase
i - I dueto a unitiucreasein fluidpressure in phasej - I. Theycanbe thought
of as forming
a tensor storagecoefficient,
linkingthevector composed
of theincrementsof fluidcontent
and~’(~)to thetwoporepressures p~)andp~2).In Biot’s single porosity
theory,thestorage
compressibility S = o~/BK.The diagonalcomponents a~ and s33 for the doubleporosity
theory are~(0/B[s(~)]I(for~ = 1: 2, whereagainthenotation emphasizes
thesimilaxityto
single porositycase,butalsoemphasizes th~.ttheyarenotnecessarily thevaluesof ph~ei as
a singleporositymaterial.’]’hedouble porositymaterial
is a composite,
andhencehaseffective
modulithatdependon theinteractions between thetwophases.
7. 2 Stress formulation
The six coefficients that describe completely the double porosity material can be obtained
from the three values for the matrix and the fracture compressibility with just one additional
assumption about the coupling between the matrix and fracture phases at constant stress. In
our stress-based formulation, the coefficient, a~a is a cross-storage coefficient for conditions of
constant confining pressure. The assumption a~a = 0 is equivalent to K~~} = v(1)K~, which
obtains for a matrix of a single constituent and v(~) -’, 1. The assumption a~a = 0 immediately
leads to equality between the various Skcmpton-like coefficients: B(~)= ~’E~
~(t) = B[u(~)]
-a~2/a2~ and B (~) (~}
B: EB= B[u (2)] = -a~a/a33. This assumption also leads from (43) to the
result that the overall compressibility a~ is the volumeaverage of the matrix compressibility
~/K(Dand the fracture compressibility o(~)/K(~) = 1/(k,~s}, where k,, is the fracture stiffness
(GPa/m}and s (m) is the fracture spacing (cf. Elsworth and Bai [1992]).
Numerical values for the matrix and fracture properties are given for Berea sandstone and
~Vesterly granite in Table 1. Expressions and numerical values for all the aljs are given in
Table 2. Note that if v(1) "-’ 1, then the results for a22 and a33 are the same as the normal
three-dimensional storage coefficient for the ~natrix and fracture phases separately. Also, the
long-time three-dimensional storage coefficient is the sum of the storage coefficients of the
individual phases. These results are intuitively reasonable, and provide meansof predicting the
deformation and fluid storage behavior of the fractured, porous mediumfrom knowledge of the
individual phases. In particular, note that all the coefficients in Table 2 were computedfrom the
values of K(1), K~~), KI, ¢5(~’}, v(~), and K(~} quoted in Table 1. Thus. these six measurements
(together with Poisson’s ratio) are sufficient to determine completely the behavior of the double-
porosity model.
In contrast to the preceding example, Table 3 presents data for Chelmsford granite and
~Veber sandstone taken from l~boratory measurements by Coyner [198,i]. The data available
in these experiments differs somewhat from the preceding case, since Coyner’s experiments
included a series of tests on several types of laboratory scale rock samples at different confining
pressures. The values quoted for K and K~ are those for a moderate confining pressure of 10
MPa(values at lower confining pressures were also measaured but we avoid using these values
because the rocks generally exhibit nonlinear behavior in thai region of the parameter space).
while the values quoted for h"(~) and K~~) are at 25 MPa. Thus, based on the idea that the
pressure behavior is associated with two kinds of porosity in the laboratory samples -- a crack
porosity, which is being closed between 10 and 25 MPa, and a residual matrix porosity above
17
25 MPa,we assume the a.vailable data ,~re K, h’.~, K(1), K~~), KI, O{1). and v(2). Wefind that
these data are sufficient to computeall the coefficients, and therefore no assumption need be
madeabout the value of a~a. In "_[’able 4, we find for both types of rock that this coefficient is
positive and small -- about an order of magnitude smaller than the other matrix elements. The
only other unusual feature of the results computedusing this laboratory data is the occurrence
of values larger than unity for B[u(0] in Chelmsfordgranite and for B[,~(~)] andOEB ~(~) in Weber
sandstone. Note also that o~{~) for both rocks is very close to unity. In this example, seven
measurements( together with Poisson’s ratio) are sufficient to determine completely the behavior
of the double-porosity model. The main difference between this example and the preceding one
is that having a direct measurementof A" eliminates the necessity of assu~ning a~3 = 0.
¯ -1~3
= bp~Z
~ ~ ~’=~v~
~((:~ (78)
~=0
The fractures will expand within the constraint of zero total strain because ~n increase in the
fluid pressure in the fractures will cause compression of the matrix. Therefore, fluid must be
withdrawnfrom the matrix in order ~o maintain 6p~t} = 0. Thus, the constant A.z3 is expected
to be negative, while in fact A2a = -ai2aia/a~ < 0 follows front the assumption that a23 = 0.
The diagonal storage coefficient in the strain formulation A33 = a:~3 - a~3/a~. As in the
discussion of .4~3, this coefficient is for the case of constant total strain. ~ndthe fractures are
to exp~ndin response to an increase in p~Z)’- because the matrix contracts. Therefore. the
fracture storage is not negfigible, as it would be for ~ rigid matrix. Elsworth ~nd Bai [1992]
calculated an unreasonably smaUvalue of n{~) of 2.3x l0 -4. because they used the rigid matrix
value for A.~a.
The strain formulation is as valid a formulation as the stress formulation we have used.
However. the ad hoc ~ssumption that the cross storage coefficient .4~ = 0 is not justified:
significant coupling occurs between ~he fracture and matrix for conditions of constant total
strain. The assumption that .-t.z3 = 0 or .4~ ~ 0 leads to significant underestimation of the
early pressure buildup in the m~trix at short times. On the other hand. the assumption that the
cross-storage coefficient a~3 = 0 is justified on the grounds that the overaU solid grain modulus
is ~kely to be close to that of the matrix grains (also compareTable 4).
18
7.4 Teraporal and spatial scales
The local mechanical and fluid pressure response of a double porosity material is time and
scale dependent, because fluid is exchanged between the two phases and because the two fluid
pressures can be m~intained independently {at least in a gedanken experiment). As demon-
strated in our preceding detailed analysis and that of Wilson and Aifantis [1983], consideration
of short, intermediate, and long time scales leads to theoretical relationships between measure-
ments and phenomenological coefficients. By making separate analyses at the mesoscopic scale
of a typical core sample and of a typical fracture, we showedthat all six constants are deter-
mined from the matrix poroelastic coefficients and the fracture compressibility ba~ed on the
single assumption that a23 --" 0. Then all the constants defined for different time scales can be
determined. At the long-time scale, the double-porosity mediumbehaves as an equivalent single
porosity mediumwith a single macroscopic compressibility: Skempton’s coefficient, and stor-
age compressibility. In general, a23 does not vanish identically, but is likely to be significantly
smaller in magnitude than the other matrix elements.
8. CONCLUSIONS
’[’he six coefficients in the double porosity theory can be broken downinto three categories
corresponding to the three coefficients in the single porosity theory. The effective medium
hulk modulus replaces the ordinary bulk modulus of a single porosity medium. Two
poroelastic expansion coefficients, one for the pore pressure in each phase, replace the
single poroelastic expansion coefficient in a single porosity medium.A symmetric two-
by-two poroelastic storage tensor consisting of three coefficients, two diagonal and one
off-diagonal, replaces the single storage coefficient in a single porosity medium.
The magnitude of the off-diagonal coefficient a2a in a stress-based formulation can be as-
sumedto be zero (or at least quite small) for a fractured, porous medium.The assumption
that the strain-based cross-coefficient A23- 0 is not justitied.
Consideration of very short, intermediate, and long time scales yields definitions of a
number of poroelastic moduli, manyof which are physically realizable in the laboratory
or field, and the interrelationships between these poroelastic moduli.
Finally, an important direction for future work is to deal with those situations where it
might be either difficult or impossible to makethe required measurements of the parameters
(but predictive capability is vital}. Then, we will want to introduce a new microscopic point
of view in order to relate the phenomenologicalcoefficients to quantities such as grain moduhls
and porosity at the microscale. The standard model of the microscale (and the one used in the
examples in the present paper) is that used by Gassmann[1951], which carries the restrictive
assumption that the solid frame is composed of only one type of elastic constituent. A more
general point of view has been introduced recently by Berrymanand Miltou [1991. 1992], Norris
[1992], aud Berryman[1992], whoshow howto relate macroscopic coefficients in poroelasticity
to quantities at the microscale whentwo types of solid constituents are present. Makinguse
of this approach will permit us to makethe microscopic identification of coefficients in more
complex and therefore more realistic geologic media.
19
ACKNOWLEDGMENTS
JGB thanks N. G. W. Cookand S. Ita for helpful conversations regarding reciprocity rela-
tions. The work of JGB was performed under the auspices of the U. S. Department of Energy
by the Lawrence Livermore National Laboratory under contract No. W-7405-ENG-.18and sup-
ported specifically by the Geosciences Research Program of the DOEOffice of Energy Research
within the Office of Basic Energy Sciences. Division of Engineering and Geosciences. The work
of HFWwas also supported by OBESunder grant no. DE-FG02-91ER14194.
Bai. M., D. Elsworth, and .].-C. Roegiers, Modeling of naturally fractured reservoirs using
deformation dependent flow mechanism, Int. J. Rock Mech. Min. Sci, ~ Geomech. Abstr.
30. 1185-1191, 1993.
Berryman. J. G.. and G. W. Milton. Exact results for generalized Gassmann’s equations in
composite porous media with two constituents, Geophysics 56. 1950-1960, 1991.
Beskos. D. E., and E. C. Aifantis, On the theory of consolidation with double porosity, -- II,
Int. J. Engng. Sci. 24, 1697.-1716. 1986.
Blot, .kI. A., General theory of three di~nensional consolidation, J..4ppl. Phys. 12. 155-164,
1941.
Blot, M. A., anti D. G. Willis, The elastic coefficients of the theory of consolidation, J. App.
Mech. 24, 594-601, 1957.
Brown,R. J. S., and J. Korringa, On the dependenceof the elastic properties of a porous rock
on the compressibility of a pore fluid, Geophysics 40. 608-616. 1975.
C.oyner, K. B.. Effects of Stress. Pore Pressure, and Pore Fluids on Bvlk Strain, Velocity, a~d
Permeability of Rocks. Ph. D. Thesis. Massachusetts Institute of Technology, 198.I.
2O
Detournay, E., and A. Ii.-D. Cheng, Fundamentals of poroelasticity, in Comprehensive Rock
Engineering, edited by J. A. Hudson, Vol. 2, Chapter 5, PergamenPress, Oxford, 1993.
Elsworth, D., and M. Bai, Flow-deformatioa response of dual-porosity media, ASCEJ. Geotech.
Engng., 118, 107--124, 1992.
Geertsma, J., The effect of fluid pressure decline on volumetric changes of porous rocks. Tans.
:tIME 210. 331-340, 1957.
Hart. D. J., and It. F. Wang,Laboratory measurementsof a complete set of poroelastic rnodull
for Berea sandstone and Indiana limestone, J. Geophys. Res., submitted. 1995.
Khaled. M. Y., D. E. Beskos, and E. C. Aifantis, On the theory of consolidation with double
porosity -- III A finite element formulation, Int. J. Num. Anal. Methods Geomech. 8,
101-123, 1984.
Kiimpel. H.-J.. Poroelasticity: parameters reviewed, Geophys. J. Int. 105, 783-799, 1991.
Love, A. E. H.. A Treatise on the Mathematical Theory of Elasticity. Dover, NewYork, 1927,
pp. 173-174.
XlcTigue. D. F.. Thermoelastic response of fluid-saturated porous rock. J. Geophys. Res. ~1.
9533-9542. 1986.
Rice. J. I~., On the stability of dilatant hardening for saturated rock masses. J. Geophys.Res.
80. 1531-1536. 1975.
Rice, J. R.. and M. P. Cleary, Somebasic stress diffusion solutions for fluid-saturated elastic
porous media with compressible constituents, Rev. Geophys. Space Phys. 14.227-241. 1976.
Rosen, B. W.. and Z. IIashin, Effective thermal expansion coefficients and specific heats of
composite materials, Int. J. Engng. Sci. 8, 157-173, 1970.
Touloukian. Y. S., W. R. Judd, R. F. Roy: Physical Properties of Rocks and Minerals, McGraw-
llill. NewYork. 1989, Chapter 11(2}.
Vart der Kamp. G.. and J. E. Gale. Theory of earth tide and barometric effects in porous
fortnations with compressible grains. Water Resources Res. 19, 538-544. 1983.
21
Wang,It. F., Quasi-static poroeIastic parameters in rock and their geophysical applications.
PAGEOPH141,269-286, 1993.
Warren,J. E., and P. J. R,oot, The behavior of nal;urally fractured reservoirs, Soc. Pet. Eng. J.
3, 2,15-255, 1963.
~Vilson, It. K., and E. C. Aifantis. On the theory of consolidation with double porosity, Int. J.
Engng. Sci. 20, 1009-1035, 198,.
.2;.)
List of Symbols
overall fluid permeability
matrix and fracture permeabilities
confining pressure
Pc - Pl. the differential pressure
fluid pressure
matrix and fracture fluid pressures
drained and undrained fluid pressures
volumefractions occupied by matrix and fractures
withV(1) "J- i)(2) 1
Skempton’scoefficient
shear modulus
bulk modulus of drained porous frame (jacketed)
fluid bulk modulus
material (or grain) bulk modulus
CK/o~. an effective pore bulk modulus(jacketed)
an effective solid bulk modulus(unjacketed)
bulk modulus of undrained (confined) porous frame
an effective pore bulk modulus(unjacketed)
a/BK, the storage compressibility
total volume
total matrix and fracture volumes
(1 - o)V, the solid volume
CV, the pore volume
Blot-Willis parameter
volume thermal expansion coefficient
pressure difference coefficient
increment of total fluid content
increments of matrix and fracture fluid content
temperature
fluid viscosity
Poisson’s ratio
total porosity = v(1)¢(t) (2
+v
matrix porosity (fracture porosity is unity)
’FABLEI. M,~terial Pro:)erties
Berea Westerly
Parameter Sandstone Granite
K(1) (GPa,) ~
8.0 25.0"
u(’) 0.204 0.25"
~)
A’~ (GPa} 36.0" ~
45.4
(~)
a 0.78 0.45
KI (GPa) ~
3,3 ~
3.3
¢)(1) b
0.064 b
0.00106
(1)
B 0,847 0.98’!
5’(1) -~
)(GPa 0,115 0.0183
c(2) c0.0064 0.0106:
K(’~)(GPa) c
0.00775 c
0.0876
~FromRice and Cleary [t976]
bFromTouloukian et al. [1989]
CFromElsworth and Bai [1992]
TABLe. 2. Double porosity parameters coznputed frommaterialproperties
in Tablei.
Berea Westerly
Parameter Formula Sandstone Granite
K (GPa) -~
[v(’)/K0)+ ~(~)/K(~)] 1.0520 6.25
I(~ (GPa) 36.2 45.9
1 - K/K, 0.971 0.86,1
-~)
a~t (GPa 0.951 0.1601
ar~ (GPa-’) (~)
-~(Da(D/K -0.0969 -0.0178
-1) (~)
_~(~)/K
ala (Gea -0.826 -0.121
a.~ -~)
(GPa (~)
~O)aO)/B(~)K 0.11,14 0.01817
-t)
a23 (GPa 0.0 0.0 0.0
-~)
a33 (GPa (:))
v(~)Cl/Kl + L/K 0.828 0.12-t
1.0 1.0 1.0
B -(al~ + a~:])/(a~ 0.979 0.977
B(~1, B[u(’)], -a12/a22 0.$47 0.984
B(’~-), B[u(~)],n(~) -a13/a33 0.998 0.974
K~(GPa) -1
Jail -- (a12 + ala)2/(a22 a33)] 21.5 39.6
K[u( 1)] (GPa) 1.151 7.01
K[u(~)] (GPa) [~. - .h/a~]-’ 7.92 23.4
-~)
8 (GPa a/BK 0.943 0.1415
S(~) (GPa-’) (~)
a(~)/B(~)K 129.5 11.67
TABLF;3..’vla.t(:rial Pro)erties
Chehnsford Weber
Parameter Granite Sandstone
K (GPa) a8.0 a4.0
K, (GPa) ~
5’1.5 ~
37.0
~
0.85 a0.89
I(I~) (GPa) ~
17.0 a
10.0
0.25 O.15
~1
h’.~ (GPa.) a, 55.5 ~
38.0
1)c~(
0.69" 0.7.V
It’] (GPa) 3.3 3,3
~
)o( 0.0011 0.095"
B(~) 0.992 0.355
S(x)(GPa-~) , (}.0409 0.208
v(~) ~
0.011 0.0095
"I:’rom Coyner[198.1]
TABLE4. Double porosity parameters computed from material properties in Table 3.
Chelmsford Weber
Parameter Formu|a Granite Sandstone
-~)
a~ (GPa 0.125 0.250
-~)
a~ (GPa -0.0413 -0.076
-~)
a~3 (GPa -0.0649 -0.1,17
-|)
a~ (GPa 0.0,105 0.206
-~)
a~ (GPa 0.00119 0.00270
a33 -1)
(GPa 0.0664 0.145
-1)
~.’~3 (Gea 0.0630 0.142
0.997 0.994
B 0.973 0.624
(1)
B 0.992 0.355
1.022 0.368
EB 0.993 0.355
B(~) 0.950 0.980
0.978 1.011
B(2}
EB 0.961 1.004
K~ (GPa) 46.3 8.99
K(2) (GPa) 0.179 0.0666
-1
S )(GPa 0.1092 0.357
-~
S(2)
) {GPa 5.87 15.24
Figure 1,: The elements of a double porosity model are: porous rock matrix intersected by
fractures. Three types of m~croscopicpressure arc pertinent in such a. modehexternal confining
pressure pc. ]nterna~l pressure of the matrix pore fluid p~}, and internal pressure of the fracture
pore fluid p~’}.
o
o ,o
o
Pc
p~l)
¯ Pc
Figure 3: For any given value of confining pressure Pc and fracture fluid pressure p~2), there is
a special choice of the matrix fluid pressure p~} = p" that guarantees no macroscopic strain
{$e(~) = 0} ia-.the matrix. This situation allows the behaviorof the fracture to be isolated from
that of the matrix essentially as if the matrix material were macroscopica]lyrigid.