You are on page 1of 15

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 107, NO. B3, 2052, 10.

1029/2000JB000108, 2002

Models for computing geomechanical constants of


double-porosity materials from the constituents’ properties
James G. Berryman
Lawrence Livermore National Laboratory, University of California, Livermore, California, USA

Steven R. Pride
Geosciences, Université de Rennes 1, Rennes, France
Received 14 December 2000; revised 10 October 2001; accepted 16 October 2001; published 28 March 2002.
[1] In previous work, phenomenological equations for the poroelastic behavior of a double-
porosity dual-permeability medium (i.e., low-permeability storage porosity and high-permeability
transport porosity both present in the same three-dimensional (3-D) volume) were formulated, and
the coefficients in these equations were identified. In the present work, models are given instead
that allow all of these coefficients to be determined from the underlying constituents’ properties.
Two different models are provided for the six geomechanical constants in the isotropic theory. In
one model the low-permeability storage porosity is assumed to be a mixture of two porous
materials in nonwelded or partially welded contact, and the high-permeability joint porosity
develops as a misfit porosity at the interface between the two partially welded constituents. In this
model the joint phase is 100% fluid. In the other model the low-permeability storage porosity is
modeled as a single uniform porous material, while the high-permeability joint porosity is modeled
as a second distinct porous material in welded contact with the first. In this model the joint phase is
a porous continuum possessing a skeletal framework representing the asperities and gouge material
that are present in real joints. The complete set of formulas needed for forward modeling the
geomechanical constants is obtained for both scenarios. Examples are evaluated with these
formulas, and comparisons are made to previous results on double-porosity systems
analysis. INDEX TERMS: 3909 Mineral Physics: Elasticity and anelasticity; 5112 Physical
Properties of Rocks: Microstructure; 7260 Seismology: Theory and modeling; 8020 Structural
Geology: Mechanics; KEYWORDS: Geomechanics, reservoir dynamics, double-porosity,
microstructure, effective medium theory, seismology

1. Introduction [4] Berryman and Wang’s [1995] main thrust was phenomeno-
logical in nature, the point being to arrive at a series of possible
[2] In problems of fluid flow in hydrocarbon reservoirs and laboratory experiments that could be used to measure the six
aquifers the simplest and most frequent idealization is the double- independent elastic moduli of an isotropic double-porosity sys-
porosity dual-permeability model in which porous matrix blocks tem. However, what is still lacking is a set of constructive models
are intersected by a fracture network [Barenblatt and Zheltov, that enable quantitative estimates of all the coefficients to be
1960; Warren and Root, 1963]. It is generally assumed that the made in the geomechanical theory when the constituents are
fracture permeability is much greater than the matrix permeability, known and some information is available concerning the micro-
while the fracture porosity is much smaller than the matrix structure. The purpose of this paper is to provide such construc-
porosity. Therefore fluid flow occurs primarily through the fracture tive models, as well as the required analysis leading to the
network, but fluid storage occurs mostly in the porous matrix. desired formulas.
Changes in fluid pressure also affect the geomechanical response [5] To create models for the geomechanical constants, we use
of the system, either enhancing or degrading the reservoir perform- ideas developed by Berryman and Pride [1998]. Both volume
ance during pumping [Wilson and Aifantis, 1982; Khaled et al., averaging and phenomenological arguments that treat porous
1984; Beskos and Aifantis, 1986; Elsworth and Bai, 1990, 1992; materials considered to be mixtures of two porous components
Bai et al., 1993]. with each component being constructed from a different mineral
[3] Berryman and Wang [1995] showed how the quasi-static and having, in principle, a different type and magnitude of porosity
limit of Biot’s theory of poroelasticity [Biot, 1941, 1956a, 1956b, were considered. Berryman and Pride [1998] limited the treatment
1962] can be generalized to permit inclusion of both matrix to the case where the fluid pressure was uniform throughout both
porosity and joint/fracture porosity, having significantly different porous phases. In this work, we generalize the arguments to the
fluid permeabilities associated with each porosity type. The result- case where the components may have different fluid pressures, in
ing theory describes the geomechanics of a double-porosity dual- keeping with double-porosity concepts, but communication
permeability system. Tuncay and Corapciaglu [1995] used a between pores is generally permitted.
volume-averaging approach to arrive at very similar results. The [6] We begin by reviewing the results of single-porosity poroe-
correspondence between these two approaches was discussed by lasticity that are needed in the subsequent analysis. Section 3
Wang and Berryman [1996]. presents the macroscopic compressibility laws for the double-
porosity system. Sections 4 and 5 present two distinct approaches
Copyright 2002 by the American Geophysical Union. to determining the geomechanical constants aij for the double-
0148-0227/02/2000JB000108$09.00 porosity system: (1) the misfit model and (2) the two-phase porous

ECV 2-1
ECV 2 -2 BERRYMAN AND PRIDE: DOUBLE-POROSITY MODELS IN GEOMECHANICS

continuum model. The misfit model assumes that the joints arise [10] Now we consider two well-known gedanken experi-
from imperfect welding between different parts of the two-mineral- ments: the drained test and the undrained test [Gassmann,
phase medium. This approach is limited to very low volume 1951; Biot and Willis, 1957; Geertsma, 1957]. The drained test
fractions for the joint phase but is also very close in spirit to the assumes that the porous material is surrounded by an imper-
results of Berryman and Wang [1995]. The porous continuum meable jacket and the fluid is allowed to enter or escape
model assumes instead that the joints constitute a second distinct through a tube that penetrates the jacket (i.e., the fluid mass
phase of the medium. This second phase is not limited in volume is allowed to increase or decrease). Then, in a long-duration
fraction and therefore provides a more versatile model for some experiment the fluid pressure remains in equilibrium with the
applications. Examples of evaluating these coefficients for both external fluid pressure (e.g., atmospheric) and so dpf = 0. Hence
types of models are presented in section 6, which is then followed dpc = dpd. So the changes of total volume and pore volume are
by our conclusions. Some of the more tedious calculations are determined exactly by the drained constants 1/K and 1/Kp as
presented in Appendix A, and some corrections to Berryman and defined in (1) and (2). On the other hand, the undrained test
Wang [1995] are given in Appendix B. assumes that the jacketed sample has no tubes to the outside
world, so pore pressure responds only to the confining pressure
changes. With no means of escape the total fluid content cannot
2. Single-Porosity Geomechanics change, so the increment dz = 0. Then, the second equation in
(4) shows that
[7] In the absence of external driving forces that can main-
tain fluid pressure differentials over long time periods, double-    
porosity models must reduce to single-porosity models. This 0 ¼ f Kp dpc  dpf B ; ð5Þ
reduction occurs in the long-time limit when the matrix fluid
pressure and joint fluid pressure become equal. It is therefore
necessary to remind ourselves of the basic results for single- where Skempton’s [1954] pore pressure buildup coefficient B is
porosity models in poroelasticity, as the long-time behavior may defined by
be viewed as providing boundary conditions (for t ! 1) on the 
analysis of double-porosity coefficients. Further, in the specific dpf  1
B ¼    : ð6Þ
models we adopt for the geomechanical constants in the double- dpc dz¼0 1 þ Kp 1 Kf  1 Kf
porosity theory, extensive use of the single-porosity results will
be made.
[8] The volume changes of any isothermal, isotropic material It follows immediately from this definition that the undrained
can only be caused by pressure changes. The two fundamental modulus Ku is determined by [also see Carroll, 1980]
pressures of single-porosity poroelasticity are the confining (exter-
nal) pressure pc and the fluid (pore) pressure pf. The differential K
pressure pd  pc  pf is often used instead of the confining Ku ¼ ; ð7Þ
pressure. The volumetric response of a sample due to small 1  aB
changes in pd and pf takes the form [e.g., Brown and Korringa,
1975; Wang, 2000] where a is the combination of moduli known as the Biot-Willis
parameter, a = 1  K/Ks. The result (7) was apparently first
obtained by Gassmann [1951] (though not in this form) for the
dV dpd dpf case of microhomogeneous porous media (i.e., Ks = Kf = Km, the
 ¼ þ ð1Þ
V K Ks bulk modulus of the single mineral present) and by Brown and
Korringa [1975] and Rice [1975] for general porous media with
for the total volume V, multiple minerals as constituents.
[11] Next, to further clarify the structure of (4), we use Betti’s
dVf dpd dpf reciprocal theorem [Love, 1927], showing that the drained and
 ¼ þ ð2Þ undrained (denoted by (d) or (u) superscripts, respectively) pres-
Vf Kp Kf
sures and strains are related by the reciprocal relation
for the pore volume Vf, and   ðdÞ    ðuÞ 
ðuÞ de ðd Þ de
dpðcuÞ  dpf ¼ dpðd Þ
 dp :
dVf dpf dzðdÞ c f
dzðuÞ
 ¼ ð3Þ ð8Þ
Vf Kf

for the fluid volume Vf. Equation (1) serves to define the drained Upon substitution, we find directly that
(or ‘‘jacketed’’) frame bulk modulus K and the unjacketed bulk
modulus Ks for the composite frame. Equation (2) defines the 1 1 fB
jacketed pore modulus Kp and the unjacketed pore modulus Kf. ¼  : ð9Þ
Ku K Kp
Similarly, (3) defines the bulk modulus Kf of the pore fluid.
[9] Treating dpc and dpf as the independent variables, we define Comparing (7) with (9), we obtain the general reciprocity relation
the dependent variables to be de  dV/V and dz  (dVf  dVf)/V, [Brown and Korringa, 1975]
which are termed the total volume dilatation (positive when a
sample expands) and the increment of fluid content (positive when
the net fluid mass flow is into the sample during deformation), f a
¼ : ð10Þ
respectively. Then it follows directly from the definitions and from Kp K
(1), (2), and (3) that
     We note that this reciprocity relation and the form of the
de 1=K  1=Ks  1=K  dpc compressibility laws (4) also follow directly from general
¼ : ð4Þ
dz f=Kp f 1=Kp þ 1=Kf  1=Kf dpf thermodynamic arguments [e.g., Pride and Berryman, 1998]. It
BERRYMAN AND PRIDE: DOUBLE-POROSITY MODELS IN GEOMECHANICS ECV 2-3

follows that Skempton’s pore pressure buildup coefficient may be


written alternatively as

1=K  1=Ks
B¼  : ð11Þ
1=K  1=Ks þ f 1=Kf  1=Kf

[12] Finally, we condense the general relations from (4) together


with the reciprocity relations (10) into the symmetric form
    
de 1 1 a dpc
¼ ; ð12Þ
dz K a a=B dpf

where the Biot-Willis parameter a can now be expressed as

a ¼ ð1  K=Ku Þ=B: ð13Þ

This form of the compressibility laws is especially convenient


because all the coefficients are simply related to the three moduli
K, Ku, and B that have the clearest physical interpretation. A
storage coefficient R, which is a central concept in describing
poroelastic aquifer behavior in hydrogeology, can also be
introduced as in Biot’s [1941] original paper:
Figure 1. Elements of a double-porosity model are porous rock
 matrix intersected by fractures. Three types of macroscopic pressure
dz  a
R ¼ : ð14Þ are pertinent in such a model: external confining pressure pc, internal
dpf dpc ¼0 BK pressure of the matrix pore fluid p(1)
f , and internal pressure of the
fracture pore fluid p(2)
f .
This storage coefficient is the change in increment of fluid content
per unit change in the fluid pressure, defined for a condition of no
change in external pressure. It has been called the three-
dimensional storage coefficient [Kümpel, 1991]. This now crack or joint phase occupying the remaining fraction of the
completes our review of the standard results concerning the volume V (2)/V = v(2) = 1  v(1). In section 4 the joint phase will
single-porosity compressibility laws. be modeled as being a pure fluid residing entirely in the preexisting
misfit porosity, while in section 5 the joint phase will be modeled
as a porous continuum having the effective properties K(2), f(2).
3. Double-Porosity Geomechanics The general laws presented in this section are independent of such
[13] In this section we present the macroscopic compressibility modeling assumptions.
laws controlling the response of a double-porosity system. [15] The main difference between the single-porosity and
Formally, these laws represent the high-frequency limit that holds double-porosity formulations is that we allow the average fluid
before significant fluid pressure equilibration between the two pressure in the matrix phase to be different from that in the joint
phases has a chance to occur. How the high-frequency laws phase (thus the term ‘‘double porosity’’). Furthermore, we allow
treated here relax to the standard low-frequency single-porosity these two porous phases to experience different macroscopic
results is the subject of S. R. Pride and J. G. Berryman (manu- gradients in their average fluid pressures (thus the term ‘‘dual
script in preparation, 2002). The relaxation frequency separating permeability’’); however, the associated fluid transport is not
the high-frequency (double-porosity) and low-frequency (single- treated as part of this article since we analyze only the high-
porosity) response is principally a function of the permeability of frequency compressibility laws here. Altogether we have three
the storage porosity, the storage coefficient of the storage poros- distinct pressures: confining (external) pressure dpc, pore fluid
ity, and the linear dimensions of the mesoscopic extent of the pressure dp(1)
f , and joint fluid pressure dpf
(2)
(see Figure 1).
high-permeability phase. An exact expression for this relaxation Treating dpc, dp(1)
f , and dpf
(2)
as the independent variables in
frequency detailing such complicated dependence has been our double-porosity theory, we define the dependent variables to
obtained by S. R. Pride and J. G. Berryman (manuscript in be de  dV/V, dz(1) = (dV f(1)  dVf(1))/V, and dz(2) = (dV f(2) 
preparation, 2002) and can range anywhere from, say, 101 to dVf(2))/V, which are the total volume dilatation, the increment of
104 Hz, depending on the circumstances. Berryman and Wang fluid content in the matrix phase, and the increment of fluid
[1995] provide an extensive discussion of the various thought content in the joints, respectively. Finally, we assume that the
experiments and real laboratory experiments needed to identify all fluid in the matrix is the same kind of fluid as that in the cracks
the geomechanical constants present in these high-frequency or joints.
double-porosity compressibility laws. We will not repeat that [16] Linear relations among strain, fluid content, and pressure
discussion here. The interested reader should see Berryman and then take the general form
Wang [1995] for details not included in the following brief
presentation. 0 1 0 1 0 dp 1
de a11 a12 a13 c
@ dzð1Þ A ¼ @ a21 A B dpð1Þ C
3.1. Macroscopic Compressibility Laws a22 a23 @ f A: ð15Þ
dzð2Þ a31 a32 a33 ð2Þ
dpf
[14] In the double-porosity formulation, two distinct phases are
assumed to exist at the macroscopic level: (1) a porous matrix
phase with the effective properties K (1), f(1) occupying volume In this article, the aij are called ‘‘geomechanical constants.’’ By
fraction V (1)/V = v(1) of the total volume and (2) a macroscopic analogy with (12), it is easy to see that a12 = a21 and a13 = a31. The
ECV 2 -4 BERRYMAN AND PRIDE: DOUBLE-POROSITY MODELS IN GEOMECHANICS

symmetry of the new off-diagonal coefficients may again be from which follow
demonstrated by using Betti’s reciprocal theorem in the form
0 1 de ¼ a11 dpc  ða12 þ a13 Þdpf ð22Þ
  0
@ dp A ð 1Þ
de  dzð1Þ  dzð2Þ f 0 ¼ ða21 þ a31 Þdpc  ða22 þ 2a23 þ a33 Þdpf : ð23Þ
0
0 1
  0 These require that the overall pore pressure buildup coefficient is
¼ de  dzð1Þ  dzð 2Þ @ 0 A; ð16Þ given by
ð2Þ
dpf 
@pf  a21 þ a31
B ¼ ð24Þ
where quantities without overbars refer to one experiment and @pc dz¼0 a22 þ 2a23 þ a33
quantities with overbars refer to another experiment (both of which
are discussed in more detail in section 4.1) to show that and that the undrained bulk modulus is given by
ð1Þ ð2Þ ð1Þ ð1Þ ð2Þ ð2Þ 
dzð1Þ dpf ¼ a23 dpf dpf ¼ a32 dpf dpf ¼ dzð2Þ dpf : ð17Þ 1 de 
 ¼ a11 þ ða12 þ a13 ÞB: ð25Þ
Ku dpc dz¼0
Hence a23 = a32. Thus we have established that the matrix in (15) is
completely symmetric, so we need to determine only six 3.2.3. Fluid injection test, long time. [21] The conditions
independent coefficients. required to measure the three-dimensional storage coefficient R in
the long-time limit are that dp(1)
f = dp(2)
f = dpf, while dpc = 0. It
follows therefore from (4) and (14) that
3.2. Constraints on the aij Coming
From the Long-Time Limit   
@z  a 1 1
R ¼ a þ 2a þ a ¼ þ f  : ð26Þ
[17] Before discussing the specific models for the various @pf dpc¼0
22 23 33
K Kf Kf
coefficients, we state several general constraints (independent of
any modeling assumptions) on the geomechanical constants aij.
Note that in order to measure the aij in the laboratory we need to Recall that Kf is the unjacketed pore bulk modulus for the entire
only consider an isolated sample immersed in a ‘‘reservoir’’ sample.
characterized by three control parameters: pc, p(1) (2)
f , and pf ; that
3.2.4. Generalized Biot-Willis parameters. [22] Equation
is, gradients in these quantities and the subsequent flow do not (19) has already determined the coefficient a11. Thus (25) shows
enter the definition of the aij. (Issues surrounding both macroscopic that
and mesoscopic fluid pressure gradients are not pertinent to the
high-frequency results of the present paper but are discussed in 1=K1=Ku
detail by S. R. Pride and J. G. Berryman (manuscript in prepara- a12 þ a13 ¼  ¼ a=K: ð27Þ
B
tion, 2002).)
[18] The constraints are obtained from the limiting case in This relation provides a constraint on the sum of the two
which the rate at which pc, pf(1), and pf(2) are all changing is much generalized Biot-Willis parameters for the double-porosity
slower than the rate at which internal fluid equilibration can take problem.
place. In this ‘‘long-time limit’’ we are always in the quasi-static [23] Not all of these long-time results are independent. In fact,
state where there are only three independent equations among the five given
above expressing the aij in terms of the single-porosity (long-time)
ð1Þ ð2Þ moduli.
pf ¼ pf : ð18Þ

Left to itself, any system having finite, cross-connected perme- 4. Modeling the Constants aij by Treating
abilities will achieve this state as t ! 1. the Joint Phase as Misfit Porosity
3.2.1. Drained test, long time. [19] The long-time drained
(or ‘‘jacketed’’) test for a double-porosity system should thus [24] We now present a model that allows the aij to be
correspond to the condition dp(1) f = dp(2)
f = 0 so that the total determined in terms of more easily measured and interpreted
volume obeys de = a11dpc. It follows therefore that parameters. The long-time limit presented in section 3.2 has
already given us three independent equations involving the six
1 unknown moduli aij. To obtain more equations, we now consider
a11  : ð19Þ a model in which the joint phase has been created when contacts
K between certain grains in the rock are not welded together. Then,
as the rock experiences deformation, points in contact may
3.2.2. Undrained test, long time. [ 20 ] The long-time separate, creating what we call ‘‘misfit porosity.’’ In particular,
undrained test for a double-porosity system should also produce it will be assumed here that phase 1 is now a mixture of two
the same physical results as a single-porosity system (assuming porous materials that have different grain minerals; for example,
only that it makes sense at some appropriate larger scale to view one constituent is made entirely of quartz grains, while the other
the medium as homogeneous). The conditions for this test are is made entirely of feldspar grains. It is along the contact between
that these two porous constituents that the crack or joint phase 2
develops. In this model the joint phase is filled entirely with
dpf
ð1Þ
¼ dpf
ð2Þ
¼ dpf ð20Þ fluid.
[25] The additional equations that lead to the determination of
the aij in this model develop in two stages: a first stage in which the
dz  dzð1Þ þ dzð2Þ ¼ 0; ð21Þ only modeling assumption is that phase 2 is entirely fluid and a
BERRYMAN AND PRIDE: DOUBLE-POROSITY MODELS IN GEOMECHANICS ECV 2-5

Figure 2. Setting p(2)


f = pc establishes a constant pressure around the matrix material which is then equivalent to the
situation experienced by a core sample of matrix material in a laboratory test.

second stage in which the specific two-mineral model of phase 1 is [ 29 ] Now we consider a thought experiment: Suppose
introduced. the confining pressure is equal to the fracture pressure, so
dpc = dp(2)
f . This situation mimics that of the matrix core sample
4.1. Laboratory Measurements on Samples in a laboratory experiment (see Figure 2). Then, by combining the
of Pure Phase 1 Material appropriate rows and columns we can telescope the 3 3 system
[26] It follows immediately from (12) and the definitions of the down to a 2 2 system of the form
constants that the pertinent equations for pure matrix material must
be given by     
deð1Þ a11 þ 2a13 þ a33 a12 þ a23 dpc
vð1Þ ¼ ð1Þ :
     dzð1Þ =vð1Þ a12 þ a23 a22 dpf
deð1Þ 1 1 að1Þ dpc
¼ ð1Þ : ð28Þ ð34Þ
dzð1Þ =vð1Þ K ð1Þ að1Þ a =Bð1Þ
ð1Þ dpf

These constants can all be estimated [Rice and Cleary, 1976; Except for an overall factor of v(1) = V (1)/V (which is generally very
Detournay and Cheng, 1993] or found in the laboratory by analysis close to unity in the cases of interest), equations (28) and (34) are
of core samples of matrix material [Hart and Wang, 1995]. As of the same form. We are therefore led to the identifications
pointed out by Berryman and Wang [1995], this fact is important
because it suggests another way of identifying certain combina- vð1Þ
tions of the aij. ¼ a11 þ 2a13 þ 
a33 ; ð35Þ
K ð1Þ
[27] To obtain the connecting equations, first note that
vð1Þ að1Þ
dV dV dVð1Þ ð2Þ  ¼ a12 þ a23 ; ð36Þ
de  ¼ vð1Þ ð1Þ þ ¼ vð1Þ deð1Þ þ vð2Þ deð2Þ ; ð29Þ K ð1Þ
V V V
vð1Þ að1Þ   vð1Þ að1Þ
ð1Þ
dV ðiÞ ð 1Þ
þ vð1Þ fð1Þ 1=Kf  1=Kf ¼ ð1Þ ð1Þ ¼ a22 : ð37Þ
deðiÞ  ð iÞ i ¼ 1; 2; ð30Þ K B K
V
[30] The stress-strain relations may now be expressed in the
ð2Þ ð2Þ ð2Þ form
dVf  dVf dV ð2Þ dVf
dzð2Þ  ¼  : ð31Þ 0 1
V V V de
@ dzð1Þ A
The final relation follows from the identity dV (2) = dV (2)
f since all dzð2Þ
fracture volume is considered here to be occupied by fluid.
[28] Thus, to obtain the desired relations in a form analogous to 0 1
(28), we must rewrite (15) as 1=K a12 a=K  a12
¼@ a12 vð1Þ að1Þ =Bð1Þ K ð1Þ vð1Þ að1Þ =K ð1Þ  a12 A
0 1 0 10 dp 1 a=K  a12 vð1Þ að1Þ =K ð1Þ  a12 a33
de a11 a12 a13 c
@ dzð1Þ A ¼ @ a21 B ð1Þ C
a22 a23 A@ dpf A; ð32Þ 0 1
dpc
vð2Þ deð2Þ a31 a32 a33 ð2Þ
dpf B ð 1Þ C

@ dpf A; ð38Þ
ð2Þ
where the third diagonal element has been modified to eliminate dpf
the fluid contribution such that
where
vð2Þ . .
a33 ¼ a33  : ð33Þ
Kf a33 ¼ vð2Þ Kf þ vð1Þ K ð1Þ  ð1  2aÞ=K þ 2a12 ; ð39Þ
ECV 2 -6 BERRYMAN AND PRIDE: DOUBLE-POROSITY MODELS IN GEOMECHANICS

and an explicit expression (that we do not need here) for a12 is


given by Berryman and Wang [1995].
[31] Combining (35) and (36) then shows that

vð1Þ =Ksð1Þ ¼ a11 þ a12 þ 2a13 þ a23 þ a33 ; ð40Þ

while combining (36) and (37) shows that


 
ð1Þ ð1Þ
vð1Þ fð1Þ 1=Kf  1=Kf ¼ a12 þ a22 þ a32 : ð41Þ

Equivalently,
ð1Þ
vð1Þ fð1Þ =Kf ¼ a12 þ a22 þ a32 ; ð42Þ

where, by analogy to (33), we define

vð1Þ fð1Þ
a22 ¼ a22  : ð43Þ
Kf

4.2. Two Weakly Bonded Minerals: An Exactly Solvable


Model
[32] So far we have considered the phenomenological equations Figure 3. A composite porous medium is composed of two
of the double porosity model at the fully macroscopic level and at distinct types of porous solid (A, B). In this model, fractures result
the core sample level. Now we want to reconsider the equations from the presence of misfit porosity between the different types of
from the microscopic point of view in a case that can be solved solid, since they are assumed to be only weakly bonded at points of
exactly. This example will provide some more insight into the contact.
nature of the equations and also provide a useful means of
evaluating the coefficients exactly in a nontrivial situation. These relations determine Ks in terms of b together with the
[33] The example we consider is the same one studied previously constituents’ moduli Km(A), Km(B) and thermal expansion coefficients
by Berryman and Milton [1992] and Berryman and Pride [1998]. b(A), b(B). One convenient form of this result is given by
Suppose a composite porous medium contains two types of porous
solid A and B occupying volume fractions v(A) = V (A)/Vand v(B) = V (B)/ ! 
V, respectively (see Figure 3). The intrinsic porosities of these two vð1Þ X vðaÞ X 1=Kmð AÞ  1=KmðBÞ
ð1Þ ðaÞ ðaÞ
constituents are f(A) and f(B), at least one of which must be nonzero ¼ ðaÞ
þ v b v b ;
Ks a¼A;B Km a¼A;B bð AÞ  bðBÞ
and composed of connected pores. The drained frame moduli of the
constituents are K (A) and K (B), while the mineral (or grain) moduli of ð46Þ
these microscopically homogeneous regions are Km(A) and Km(B). The
volume thermal expansion coefficients of the constituent grains (and where v(1) = v(A) + v(B). If the bulk moduli of the constituents are
therefore also of the frames and pores) are b(A) and b.(B) These two the same so that Km(A) = Km(B), then clearly the unjacketed modulus
constituents may not fill the entire volume. The remaining misfit satisfies Ks = Km(A), regardless of the values of the thermal
volume fraction due to the presence of cracks or fractures is therefore expansion coefficients. We refer to the case with equal constituent
given by v(C ) = 1  v(A)  v(B), while the connection to our previous bulk moduli (i.e., the microhomogeneous case) as ‘‘the Gassmann
notation is provided by the observation that v(C )  v(2). limit.’’
[34] Now we describe a gedanken experiment for an unjacketed [35] It is also easy to see that for the ratio of dpf and dq given by
(dpd  0), nonisothermal (dq 6¼ 0) configuration such that the (45), the porosity of the composite does not change. In fact,
relative change in the volumes V (A), V (B) of the constituents is the changes in pf and q also do not induce changes in intrinsic porosity
same and, by assumption, also equal to that of the total volume V. for either constituent A or B, and since the overall deformation
Then, we have the set of equalities takes the form of uniform swelling or shrinking, the overall
porosity f is also constant. Thus,
dV dpf
 ¼  bdq  
V Ks df 1 1  
 ¼  dpf  bf  b dq  0: ð47Þ
f Kf Ks
dpf
¼ ð AÞ
 bð AÞ dq
Km Combining the result (45) with (47), we find
dpf   !
¼ ð BÞ
 bðBÞ dq: ð44Þ 1 1   1 Kmð AÞ  1 KmðBÞ
Km ¼ þ bf  b : ð48Þ
Kf Ks bð AÞ  bðBÞ
It follows [Berryman and Milton, 1992] that these equalities are
satisfied if the changes in the fields have the ratio In the limit of equal constituent bulk moduli it is clear that Kf = Ks
= Km(A) = Km(B) (as expected in the Gassmann limit) regardless of the
values of the b.
dq 1=Kmð AÞ 1=KmðBÞ 1=Ks 1=KmðBÞ 1=Kmð AÞ 1=Ks
¼ ¼ ¼ : ð45Þ [36] Considering the long-time behavior of the system when
dpf bð AÞ  bðBÞ b  bðBÞ bð AÞ  b pressure equilibration is achieved so that p(1) (2)
f = pf = pf, we found
BERRYMAN AND PRIDE: DOUBLE-POROSITY MODELS IN GEOMECHANICS ECV 2-7

previously (see (19), (26), and (27)) three equations for the six showing that
unknown coefficients.
[37] The second convenient limiting case is the one related to X vðaÞ 1
measurements of core samples: dpc = dp(2)f . Then, using (28) and a13 ¼ ðaÞ
 ð57Þ
(34), we find that a¼A;B
K K

vð1Þ for this model.


a11 þ 2a13 þ a33 ¼ ð49Þ [40] Substituting (57) into (27) gives
K ð1Þ ;

1 X vðaÞ
vð1Þ að1Þ a12 ¼  ; ð58Þ
a12 þ a23 ¼  ð1Þ ð50Þ Ks a¼A;B K ðaÞ
K ;
!
vð1Þ að1Þ 1 1 which, when substituted into (50), gives
ð1Þ ð1Þ
a22 ¼ þv f  ; ð51Þ
K ð1Þ Kf K ð1Þ
f X vðaÞ vð1Þ að1Þ 1
a23 ¼ ðaÞ
 ð1Þ
 : ð59Þ
giving three more equations for the six coefficients. (Below we a¼A;B
K K Ks

show how to eliminate Kf(1) from (51) in favor of component


properties and bf.) Unfortunately, these six equations are not This completes the identification of all the off-diagonal terms.
linearly independent, so they cannot be solved for the coefficients [41] Two of the diagonal coefficients (a11 and a22) are given
until some additional relationship is found. directly by (19) and (51), respectively, and thus only a33 remains to
[38] To provide the remaining relationships needed to solve for be determined. Solving (49) for the overlined coefficient gives
the coefficients, we consider the calculation of the misfit volume
fraction v(C ) = v(2) itself in the long-time limit ( p(1) = p(2)
f ). This X vðaÞ
f
1 vð1Þ
calculation gives directly equations for both a13 and for the a33 ¼
 þ ð1Þ  2 : ð60Þ
combination a23 + a33. We need only one of these (a13) to solve K K a¼A;B
K ðaÞ
for the coefficients, and the other serves as another consistency
check. Alternatively, we can solve (26) for a33 to find
[39] The equation for the change in one of the volume fractions
is " #
1 vð1Þ X vðaÞ vðCÞ 1 vð1Þ f vð1Þ fð1Þ
  ! a33 ¼ þ ð1Þ  2 þ þ   þ ;
dvð AÞ dvð AÞ dV 1 1 1 1 K K a¼A;B
K ðaÞ Kf Ks Ksð1Þ Kf Kf
ð1Þ

¼ ð AÞ  ¼  ð AÞ dpd þ  dpf : ð52Þ


vð AÞ V V K K Ks Kmð AÞ ð61Þ

Combining this with a similar expression for the change in v(B) where we used the fact that v(C) = f  v(1)f(1). Comparing (60) and
gives (61), we see that they are in agreement if and only if the sum of the
terms in brackets in (61) vanishes identically. We will give a brief
! ! analysis to show why this is so.
vð1Þ X vðaÞ vð1Þ X vðaÞ
dvð AÞ þ dvðBÞ ¼  dpd þ  dpf [42] We consider the sum of all the coefficients in the coef-
K a¼A;B
K ðaÞ Ks a¼A;B KmðaÞ ficient matrix. Using (40), (41), (42), and (43), we find
¼ dvðcÞ :
1 f
ð53Þ a11 þ 
a22 þ 
a33 þ 2ða23 þ a13 þ a12 Þ ¼  : ð62Þ
Ks Kf
Note that from the definition of v(C ) we have that
Similarly, using equations (49), (50), and (51), we find
dV ðCÞ dV !
¼ dvðC Þ þ vðCÞ : ð54Þ 1 fð1Þ
V V a11 þ 
a22 þ 
a33 þ 2ða23 þ a13 þ a12 Þ ¼ v ð1Þ
 :
ð1Þ ð1Þ
Ks Kf
Thus the increment of fluid content in the fractures is given by
ð63Þ
ð2Þ ð2Þ ð2Þ
dVf  dVf dV ðCÞ dVf Since the left-hand side is the same for both (62) and (63), we
dzð2Þ  ¼  vðC Þ ð2Þ
V V Vf conclude that
ð2Þ
!
dV dV f
!
¼ dvðCÞ þ vðCÞ  ð2Þ : ð55Þ 1 f ð1Þ 1 fð1Þ
V Vf  ¼v ð1Þ
 ð1Þ ; ð64Þ
Ks Kf Ks Kf
After substituting for dv(C) from (53), we find that the final result is
and therefore that the expression in brackets in (61) is identically
! ! zero, consistent with (60).
X vðaÞ 1 X vðaÞ 1 vðC Þ [43] Since it is well known [Berge et al., 1993] that the unjack-
dzð2Þ ¼  dpd þ  þ dpf ;
K ðaÞ K ðaÞ Ks Kf eted pore moduli Kf and Kf(1) are very difficult to measure accu-
a¼A;B a¼A;B Km
rately, we can take advantage of (64) to eliminate Kf(1), and then we
ð56Þ can use (48) to eliminate Kf as well. We still have one constant
ECV 2 -8 BERRYMAN AND PRIDE: DOUBLE-POROSITY MODELS IN GEOMECHANICS

remaining, bf, that may be rather difficult to measure. In general, for this special model. The unjacketed modulus is found from
thermal expansion data may be more difficult to obtain on given (65) using Ks(1) = K(1)/(1  a(1)).
materials than elastic modulus data. We discuss methods of estimat- [48] From this point onward, the analysis can again follow that
ing thermal expansion coefficients from constituent information in of Berryman and Wang [1995] because all the constants needed in
Appendix A. their formulas are now available. Table 1 presents a set of input
parameters to be used in these formulas, and Table 2 displays the
4.3. Examples results of the calculations. We consider two types of material,
4.3.1. Estimates using the joints-as-misfit-porosity Chelmsford granite and Weber sandstone, since these are two of
model. [44] In the case where we have modeled the joint the rocks studied in the well-characterized laboratory data set of
phase 2 as being due to misfit porosity, the required parameters Coyner [1984]. We have used the grains-imbedded-in-cement
of the constituents are v(A), v(B), f(A), f (B), Km(A), Km(B), K(A), K(B), approach in both cases. We take the known mechanical properties
b(A), and b(B). Two other measurements are still required: (1) the and porosities from Coyner [1984], typical properties for thermal
overall bulk modulus K and (2) the bulk modulus K(1) of the expansion from Fei [1995], and values of constituent moduli from
matrix. Such data are needed for a fundamental reason: The Mavko et al. [1998]. We treat the solid grains (material B) as if they
formulas presented all have the remarkable property that they are quartz with some clay for both examples. For the granite we
do not depend explicitly on the microstructure. Yet there is no treat material A as a plagioclase feldspar (albite), while for the
doubt that the microstructure must play a role in the results. The sandstone we treat it an ‘‘average’’ feldspar [Mavko et al., 1998].
effect of the microstructure is felt most strongly in the two The results given in Table 2 depend on the values chosen for the two
constants K and K(1) (e.g., it is easy to see that fractures cause thermal expansion coefficients b and bf. However, the main
K to be much smaller than K(1)), and the microstructure therefore dependence on these coefficients is in the values of Ks, Kf, Ks(1),
affects the double-porosity parameters through the presence of and Kf(1). For the sandstone this dependence is relatively small, but
these measured parameters. for the granite it is very noticeable. In fact, we have chosen the value
[45] From this set of parameters, Ks is computed using (46), and of b to be used for the Chelmsford granite (in the absence of any
Kf is found using (48). Appendix A lists the formulas needed to data) so that the values of Ks and Ks(1) agree well with Coyner’s data.
compute the required matrix parameters. (The reader should keep The values of Kf and Kf(1) for the granite then still depend quite
in mind that Ks is a specific averaged quantity, not necessarily a strongly on the value of bf, but interestingly, it turns out that the
simple grain modulus for the composite structures that we have double-porosity coefficients aij are largely insensitive to the Kf
been addressing here.) values. This result can be traced to the fact that Kf always appears
[46] The procedure for this case is relatively straightforward with Kf in the combination 1/Kf  1/Kf, so that the small value of Kf
since all the needed equations have been listed in Appendix A. dominates such terms and therefore makes the precise value of Kf
This modeling approach is quite generic, applicable to a wide range largely unimportant for most cases of interest.
of possible microstructures. When the microstructure is special, we [49] Comparing the results obtained for the double-porosity
can simplify the analysis in some cases. We present one such coefficients aij in Table 2 with the corresponding values obtained
special case to conclude this section. by Berryman and Wang [1995] (see Table B2 in Appendix B for
4.3.2. Modeling solid grains imbedded in a cementing corrected values), we find that the agreement between the two
material. [47] The analysis of two porous minerals presented by approaches is quite good, as one might expect. We find that the
Berryman and Milton [1991] requires that at least one of the two values compare very well for both the granite and the sandstone,
constituents is porous and that the porosity is connected and in particular, we find again that the off-diagonal term a23 is
(percolating) so that overall pore pressure can equilibrate over quite small for both rocks in these examples.
long enough times. A somewhat simpler case to consider that has
all the required characteristics is that of solid grains imbedded in a 5. Model Treating the Joint Phase
porous cementing material that serves as the host for the granular
inclusions. Then we can use formulas for this case given by as a Porous Continuum
Berryman and Milton [1991] to obtain our parameter estimates. [50] The validity of the misfit model is limited to the case where
The matrix material (without fractures) is still well-bonded v(2) v(1) and requires that each porous constituent is made of a
internally and is composed of a uniform porous matrix of single isotropic mineral. To extend the utility of the model-based
material A that is imbedded with solid grains of material B. Then
it is clear that 0 = f(B) = a(B) since the solid grains have no porosity
and, similarly, that K(B) = Km(B). The results of Berryman and Table 1. Material Properties
Milton [1991, equations (58) and (59)] for such a material are
Parameter Chelmsford Granite Weber Sandstone
ð1Þ ð AÞ K ð1Þ  KmðBÞ K, GPa 8.0a 40.0a
a ¼a ð BÞ
; ð65Þ b, 106 K1 17.5 10.8b
K ð AÞ  Km bf, 106 K1 25.0 5.4
K (1), GPa 17.0a 10.0a
and the corresponding result for Kf(1) is n(1) 0.25 0.15
  K f, GPa 3.3 3.3
fð1Þ að1Þ a ð AÞ  f ð AÞ K (A), GPa
¼  f ð AÞ  f ð AÞ að AÞ  að1Þ 13.0 3.25
Kf
ð1Þ
Ks
ð1Þ
Km
ð AÞ K (A)
m , GPa 75.6c 37.5c
! a(A) 0.828 0.913
a ð AÞ b(A), 106 K1 12.2b 12.2b

ð BÞ
; ð66Þ f(A) 0.0017 0.36
K ð AÞ  Km
v(A) 0.645 0.264
where K (B)
m, GPa 39.0c 39.0c
b(B), 106 K1 24.3b 24.3b
(B)
vð AÞ v 0.334a 0.7265
f ð AÞ  ; ð67Þ v(B) = v(2) 0.011a 0.0095
vð AÞ þ vðBÞ a
From Coyner [1984].
b
From Fei [1995].
fð1Þ ¼ f ð AÞ fð AÞ ð68Þ c
From Mavko et al. [1998].
BERRYMAN AND PRIDE: DOUBLE-POROSITY MODELS IN GEOMECHANICS ECV 2-9

Table 2. Double Porosity and Other Computed Parameters


Parameter Formula Chelmsford Granite Weber Sandstone
Ks, GPa equation (45) 53.6 37.3
Kf, GPa equation (48) 37.9 36.7
a 1  K/Ks 0.85 0.89
a(1) a(A)(K (1)  K (B)
m )/(K
(A)
 K (B)
m ) 0.70 0.74
Ks(1), GPa K (1)/(1  a(1)) 56.8 38.6
Kf(1), GPa equation (66) 11.6 37.2
a11, GPa1 1/K 0.125 0.250
a12, GPa1 1/Ks  v(a)/K (a) 0.0398 0.073
a13, GPa1 v(a)/K (a)  1/K 0.0666 0.150
a22, GPa1 v(1)a(1)/K (1) + v(1)f(1)(1/K f  1/Kf(1)) 0.0410 0.0996
a22, GPa1
 a22  v(1)f(1)/Kf 0.0407 0.0708
a23, GPa1 v(1)a(1)/K (1)  a12 0.00099 0.00028
a33, GPa1 v(2)/Kf + v(1)/K (1)  (1  2a)/K + 2a12) 0.0696 0.152
a33, GPa1
 a33  v(2)/Kf 0.0663 0.149

double-porosity analysis, we now propose a second model. Suppose (note that g12 6¼ g21) and will find the six unknown coefficients
two distinct porous phases are present, as in the work of Berryman (Qi, gij) as part of the analysis.
and Milton [1991] and Berryman and Pride [1998]. One of these [53] Three equations involving the Qi and gij are immediately
porous phases may occupy significantly more total volume and also obtained by inserting the soild pressure response laws (70) and
more pore volume than the other. The second phase may occupy (71) into the definition of the total macroscopic confining
smaller total volume and have smaller pore volume as well but is pressure
distinguished by its high permeability. However, the condition
v(2) v(1) is not required, although it might hold in some cases. h  i
ð1Þ
In this model, phase 2 again represents the joints but is now treated dpc ¼ vð1Þ 1  fð1Þ dps1 þ fð1Þ dpf
as a porous continuum in fully welded contact with a uniform phase h  i ð72Þ
ð2Þ
1 porous continuum. This model is in no way restricted to the case þvð2Þ 1  fð2Þ dps2 þ fð2Þ dpf
of single-mineral porous constituents and is consistent with real
joints in which asperities and gouge-like material bridge the gap that to obtain
defines the joint thus forming a skeletal framework. The joint phase
is therefore allowed to have its own porosity f2 and drained bulk  
modulus (frame modulus) K (2). This model for the aij might be 0 ¼ vð1Þ Q1 þ vð2Þ Q2  1 dpc
considered preferable in many cases because the only information  
ð1Þ
that must be known at the macroscopic scale is K, the overall þ vð1Þ fð1Þ  vð1Þ Q1 g11  vð2Þ Q2 g21 dpf ð73Þ
 
drained modulus for the double-porosity composite. ð2Þ
þ vð2Þ fð2Þ  vð1Þ Q1 g12  vð2Þ Q2 g22 dpf :
[51] The logic of the approach is as follows. We can apply the
usual single-porosity compressibility laws to determine the
response of both phases 1 and 2 if we know the confining pressure For this to hold for any combination of the independent variables
acting on each phase which is, in general, different from the overall dpc, dp(1) (2)
f , and dpf , each coefficient must vanish and so
confining pressure acting on the double-porosity composite. Thus
we must first determine the confining pressure in each phase as a vð1Þ Q1 þ vð2Þ Q2 ¼ 1; ð74Þ
function of our three independent variables dpc, dpf(1), and dpf(2).
Inserting these results into the single-porosity compressibility laws
for phases 1 and 2 then allows the total double-porosity response to vð1Þ Q1 g11 þ vð2Þ Q2 g21 ¼ vð1Þ fð1Þ ; ð75Þ
be determined by simple addition. A similar approach was taken by
Berryman and Pride [1998] but in the more simplified setting of the vð1Þ Q1 g12 þ vð2Þ Q2 g22 ¼ vð2Þ fð2Þ : ð76Þ
long-time limit (having only one fluid pressure present in the limit).
[54] To obtain more useful equations, we next introduce the
5.1. Theoretical Development solid pressure laws (70) and (71) into the known single-porosity
[52] The confining pressure in each phase is simply the average compressibility laws (12) that hold in each phase
pressure throughout the entire extent of each phase
  1 h   
ð1Þ
i
ð1Þ
dpc1 ¼ 1  fð1Þ dps1 þ fð1Þ dpf ; ð69Þ deð1Þ ¼ ð1 Þ
1  fð1Þ dps1 þ fð1Þ  að1Þ dpf
K
1 h  
ð1Þ ð2Þ
i
with a similar expression for phase 2. Now, dp(1) f is one of the ¼ ð1Þ Q1 dpc þ fð1Þ  að1Þ  g11 Q1 dpf  g12 Q1 dpf ;
K
independent variables in our theory so in order to know dpc1, we
ð77Þ
must first express the average grain pressure in each phase dps1 in
terms of the independent variables dpc, dp(1) (2)
f , and dpf . Following    
dzð1Þ 1   1
and extending the analysis of Berryman and Pride [1998], we ð1Þ ð1Þ ð1Þ ð1Þ ð1Þ
 ð1Þ ¼ ð1Þ a 1f dps1 þ a f  ð1Þ dpf
introduce linear response laws of the form v K B
ð1Þ
   
    a ð1Þ 1 ð 1Þ ð2Þ
ð1Þ ð2Þ ¼ ð1Þ Q1 dpc þ f  ð1Þ  g11 Q1 dpf  g12 Q1 dpf
1  fð1Þ dps1 ¼ Q1 dpc  g11 dpf  g12 dpf ; ð70Þ K B
ð78Þ
   
ð1Þ ð2Þ
1  fð2Þ dps2 ¼ Q2 dpc  g21 dpf  g22 dpf ð71Þ
with comparable expressions for phase 2 (replacing 1 with 2
ECV 2 - 10 BERRYMAN AND PRIDE: DOUBLE-POROSITY MODELS IN GEOMECHANICS

everywhere). Using the definition of the total volumetric strain of [55] Upon eliminating gij, five of the double-porosity geome-
the composite chanical compliances can at last be written

de ¼ vð1Þ deð1Þ þ vð2Þ deð2Þ ; ð79Þ 1


a11 ¼ ; ð95Þ
K
the macroscopic double-porosity laws (15) of interest are thus
 
obtained by simple addition with the geomechanical constants vð1Þ að1Þ 1 að1Þ ð1  Q1 Þ
given by a22 ¼  ; ð96Þ
K ð1Þ Bð1Þ 1  K ð1Þ =K ð2Þ
vð1Þ Q1 vð2Þ Q2 1  
a11 ¼ þ ð2Þ ¼ ð80Þ vð2Þ að2Þ 1 að2Þ ð1  Q2 Þ
K ð1Þ K K a33 ¼  ; ð97Þ
K ð2Þ Bð2Þ 1  K ð2Þ =K ð1Þ
vð1Þ   vð2Þ
a12 ¼  Q1 g11 þ að1Þ  fð1Þ  ð2Þ Q2 g21 ð81Þ
K ð 1Þ K vð1Þ Q1 ð1Þ
a12 ¼  a ; ð98Þ
K ð1Þ
vð1Þ Q1 ð1Þ
a21 ¼  a ð82Þ
K ð1Þ vð2Þ Q2 ð2Þ
a13 ¼  a : ð99Þ
K ð2Þ
v ð2Þ  v  ð1Þ
a13 ¼ Q2 g22 þ að2Þ  fð2Þ  ð1Þ Q1 g12 ð83Þ The sixth can be written in a symmetrical form by considering the
K ð2Þ K
symmetrized version (i.e., 1/2(a23 + a32)) of (97) – (99). We find the
vð2Þ Q2 ð2Þ very useful form for the remaining off-diagonal term
a31 ¼  a ð84Þ
K ð2Þ  
að1Þ að2Þ K ð1Þ K ð2Þ vð1Þ vð2Þ 1
  a23 ¼ 2
þ  ; ð100Þ
vð1Þ 1 ½K ð2Þ  K ð1Þ  K ð1Þ K ð2Þ K
a22 ¼ ð1Þ að1Þ Q1 g11 þ ð1Þ  fð1Þ ð85Þ
K B
showing that there are fairly common circumstances in which this

ð2Þ
 term may either vanish identically or be very small in practice.
v 1
a33 ¼ ð2Þ að2Þ Q2 g22 þ ð2Þ  fð2Þ ð86Þ [56] These are the six relations for the aij that we have been
K B seeking. They make good use of the assumed known constants of
the constituents. The formulas obtained here are not limited in any
vð1Þ Q1 ð1Þ way by the nature of the underlying mineral moduli. The only
a23 ¼ a g 12 ð87Þ macroscopic measurement that needs to be performed on the
K ð1Þ
double-porosity composite is a drained test that gives K. All
dependence on the geometric distribution of the two porous
vð2Þ Q2 ð2Þ
a32 ¼ a g21 : ð88Þ constituents is contained within K. All dependence on the fluid
K ð2Þ properties is contained within B(1) and B(2) [Berge et al., 1993].
[57] In using these relations in the long-time results given in
If we now combine the three reciprocity conditions aij = aji along section 3.2, one can determine the Biot-Willis parameter a and the
with the three results (74) – (76), we obtain the six solid pressure Skempton coefficient B for the composite. The results so obtained
coefficients as are identical to those of Berryman and Milton [1991], who
 exploited the so-called uniform expansion (or self-similar expan-
1  K ð2Þ K sion) thought experiment that requires single-mineral porous con-
vð1Þ Q1 ¼ ð1Þ
; ð89Þ stituents. We have thus generalized the results of Berryman and
1  K ð2Þ =K Milton [1991] to include a more general class of porous composites.

ð2Þ 1  K ð1Þ K 5.2. Example Using the Joints-as-Porous-Continuum Model
v Q2 ¼ ; ð90Þ
1 K ð1Þ =K
ð2Þ [58] For our model in which the joints are treated as a separate
porous continuum in welded contact with the matrix phase 1, in
addition to the properties f(1), f(2), a(1), a(2), B(1), B(2), K (1), and
að1Þ ð1  Q1 Þ K (2) of each porous phase, we also need an estimate of the overall
Q1 g11 ¼ fð1Þ  ; ð91Þ
1  K ð1Þ =K ð2Þ drained bulk modulus of the composite K and of the overall volume
fraction v(1) of the porous matrix. For this model, in which joints are
að2Þ ð1  Q2 Þ assumed to be oriented in all three directions in order that the sample
Q2 g22 ¼ fð2Þ  ; ð92Þ is isotropic as a whole, it is reasonable to assume that K has
1  K ð2Þ =K ð1Þ contributions from both K (1) and K (2) that combine both in series
and in parallel. The geometric mean (which is well known to lie
vð1Þ að1Þ ð1  Q1 Þ between the mean and the harmonic mean in all cases) is a reasonable
Q2 g21 ¼ ; ð93Þ approximation in such circumstances, and we propose that
vð2Þ 1  K ð1Þ =K ð2Þ
 
ln K ¼ vð1Þ ln K ð1Þ þ 1  vð1Þ ln K ð2Þ ð101Þ
vð2Þ að2Þ ð1  Q2 Þ
Q1 g12 ¼ : ð94Þ
vð1Þ 1  K ð2Þ =K ð1Þ or, equivalently,
h ivð1Þ h ið1vð1Þ Þ
Note that these coefficients are independent of the fluid’s bulk K ¼ K ð1Þ K ð2Þ ð102Þ
modulus Kf.
BERRYMAN AND PRIDE: DOUBLE-POROSITY MODELS IN GEOMECHANICS ECV 2 - 11

5 6.1. Physical Constraints


[61] It is important to note that certain physical requirements
4
must never be violated by the models. This is one type of
a11 feasibility constraint on our modeling choices. For example, it is
3
nonsense to have a volume fraction that is negative. So, in the
joints-as-misfit-porosity model, if our assumptions about the
2
parameters lead to a situation wherein the predicted crack volume
a22 fraction v(C ) = v(2) is negative, then the choices that we have made
a_ij K1

1
a33 are obviously inconsistent, and better, more mutually consistent
a23 choices must be found. (Of course, changes in volume fraction can
0
a12 have either sign, but then accumulated negative changes can never
–1
exceed the starting value.) More generally, stability arguments
require that the matrix aij of geomechanical constants be positive
–2 definite; that is, the determinant of the three principal minors of aij
a13 must all be positive. Any estimate of the geomechanical constants
–3 must satisfy these three fundamental constraints.
[62] An obvious use of the formulas presented here is for
–4 construction of models to be used for testing hypotheses about
0.5 0.6 0.7 0.8 0.9 1 how various physical properties of the constituents affect the
volume fraction of phase 1 overall double-porosity system. We assume in these circumstances
that all the microstructural information is known (or has been
Figure 4. The values of the aij are normalized by the drained assumed for purposes of parameter studies) as a part of the model-
modulus of the matrix phase K (1) = 1. 1010 Pa for the model where building process.
the joint phase 2 is treated as a porous continuum in welded contact
with the matrix phase 1. Here we have taken K (2) = K (1)/20, Km(1) = 6.2. Some Constituent Information and Some
Km(2) = KSi02 = 3.6 1010 Pa, f(1) = 0.2, and f(2) = 0.4. See color Laboratory/Field Data
version of this figure at back of this issue. [63] Since the two-component composite medium paradigm
supplies us with an equation-rich environment for modeling, we
should not lose track of the fact that whenever data are available
is a good model. Another sensible choice would be to take K equal for the various constants, we should always use these data rather
to the harmonic mean of these two material constants, but this than the formulas based on the assumed two-component model.
choice is special for our results and always gives a23 = 0. Small errors in some measured parameters can propagate and
[59] In Figure 4 we plot all six aij using this model for K and become large errors in other computed parameters in some of
(95) – (99) as a function of the matrix volume fraction. The the formulas. Error analysis for the whole set of equations
parameters a(i) and B(i) have been determined assuming that both presented here would be exceedingly tedious and largely
phases 1 and 2 are single-mineral materials in which case unnecessary if enough data are available. Our general rule
. should therefore be to use the data first and then compute the
aðiÞ ¼ 1  K ðiÞ KmðiÞ ð103Þ missing pieces only.
0 . 1
ð iÞ
1 K B 1
ðiÞ K f Km C
¼ 1 þ f ð iÞ @ . A: ð104Þ 7. Conclusions
BðiÞ Kf 1  K ðiÞ KmðiÞ
[64] The standard model of the microscale in poroelasticity is
that of Gassmann [1951] in which the solid frame is assumed to be
The values of the parameters chosen correspond, for example, to composed of a single type of elastic constituent. A more general and
consolidated sandstone (phase 1) with fractures or joints running also more realistic point of view has been introduced by Berryman
through it. Observe that a12 and a13 are both negative and that and Milton [1991, 1992], Norris [1992], Berryman [1992], and
all the moduli are roughly the same order of magnitude with a23 Berryman and Pride [1998], assuming that two minerals are present
being the smallest. Although we have simply assumed a value and then working out the required modifications of the Gassmann
for K (2) in Figure 4, one could instead propose a specific model formulas. Making use of this generalization here, we have been able
for the joint phase, for example, Gangi’s [1978] bed-of-nails to show how to obtain the mechanical coefficients of a double-
model. porosity medium when data on two constituents’ mechanical and
thermomechanical behavior are available. The resulting exact for-
6. Discussion mulas for two-component media will be very useful whenever the
known field and specialized laboratory data are insufficient to
[60] In situations where the data required to evaluate the evaluate the double-porosity coefficients using the related methods
formulas of Berryman and Wang [1995] are incomplete or not of Berryman and Wang [1995]. This approach also provides a
all readily available, the formulas presented above will play an framework for modeling the geomechanical behavior and/or the
important role in the analysis. There are various situations when wave propagation characteristics of double-porosity media starting
this might arise in practice. It would become tedious to try to from realistic assumptions about their microstructure [Berryman
enumerate all of these circumstances here, so instead we have and Wang, 2000].
presented three illustrative examples of the use of the formulas in [65] One application of the present work is the problem of
sections 4.3 and 5.2. The point that we have been trying to make determining effective complex coefficients for a single-porosity
with these examples is that each of these situations does not require model (with trapped fluid in the second type of porosity) to account
more data, just different data. The total amount of information for the hidden motion and energy losses in cracks and joints.
needed is always essentially the same, but the information can Another direction for future work will be generalizations to
come in different forms and therefore needs to be processed anisotropic media. Both of these applications and generalizations
differently. are currently in progress.
ECV 2 - 12 BERRYMAN AND PRIDE: DOUBLE-POROSITY MODELS IN GEOMECHANICS

Appendix A: Parameters for the Composite can also be applied to the matrix material without fractures, as has
Matrix in the Joints as been demonstrated by Berryman and Pride [1998]. The result is
Misfit Porosity Model
1

1
   1K ð AÞ  1K ðBÞ
[66] In this appendix, we present the formulas needed to
ð1Þ
¼ þ bð1Þ  hbi m m
; ðA8Þ
estimate/compute the overall constants for a thermoelastic compo- Ks Km bð AÞ  bðBÞ
site containing two well-bonded constituents when the constituents
are porous. This model is appropriate for the matrix material in our where
double-porosity model with two minerals present. Our misfit  
model assumes that the only difference between the whole material 1 f ð AÞ f ð BÞ
 þ : ðA9Þ
and matrix material is that the matrix does not have any fractures Km Km
ð1Þ
Km
ð BÞ
(by definition) and all the fractures reside in misfit regions among
clusters of porous granular matrix material.
[67] The matrix is therefore composed of two porous materials Once b(1) is determined from (A2) then Ks(1) is determined by (A8).
A and B. The analysis of Berryman and Milton [1991] applies We have our choice of (A5) and (A8) for evaluating Ks(1), and the
directly to this material, and we will assume that the reader is choice should be made on the basis of data availability.
familiar with this part of the analysis. We will quote the results as [72] The same argument from Berryman and Milton [1992]
needed but will not give the derivations here. that gives (48) can be applied to the matrix material without
[68] What is different and not given by Berryman and Milton fractures, and the result for b(1)
f can then be rewritten as

[1991] is the fact that in addition to the results for the various bulk !
moduli, we also need corresponding results for the thermal ð1Þ bð AÞ  bðBÞ 1 1
expansions. Although the concepts follow rather easily from bf ¼ bð1Þ þ . . ð1Þ
 ð1Þ
: ðA10Þ
ð AÞ ð BÞ
1 Km  1 Km Kf Ks
previous publications, these results have not appeared anywhere
and will be presented here for the first time since they are required
in the main text. Equation (A10) shows that a measurement of K(1) f determines bf
(1)

[69] The first result we need is a standard one for volume and vice versa. One of these can be eliminated, but both are hard to
thermal expansion b(1) from the theory of composites [Rosen and measure. Fortunately, the matrix is assumed to be well bonded, so
Hashin, 1970; Christensen, 1979]: the analytical results for Kf(1) of Berryman and Milton [1991] can
  be applied. In present notation, the result is
bð AÞ  bðBÞ 1 1
bð1Þ ¼ bðBÞ þ  ðA1Þ   
1=K ð AÞ  1=K ðBÞ K ð1Þ K ðBÞ fð1Þ að1Þ af  að AÞ  aðBÞ 
ð1Þ
¼ ð1Þ
  hai  að1Þ ; ðA11Þ
or, equivalently, Kf Ks Km K ð AÞ  K ðBÞ

  
bð AÞ  bðBÞ 1 1 where a(1) is the Biot-Willis parameter for the matrix,
bð1Þ ¼ hbi þ  ; ðA2Þ
1=K ð AÞ  1=K ðBÞ K ð1Þ K
hai  f ð AÞ að AÞ þ f ðBÞ aðBÞ ; ðA12Þ
where
and furthermore
hbi  f ð AÞ bð AÞ þ f ðBÞ bðBÞ ðA3Þ
 
af a ð AÞ  f ð AÞ ð BÞ a
ð BÞ
 f ð BÞ
   f ð AÞ þ f : ðA13Þ
1 f ð AÞ f ð BÞ Km Km
ð AÞ
Km
ð BÞ
 ð AÞ þ ð BÞ : ðA4Þ
K K K
The result (A11) can be written in a number of equivalent ways.
For simplicity, we have assumed that the composite matrix is This choice is made only because it is reasonably compact.
isotropic and that the constituents are also both isotropic. Thus the [73] We have now completed our task of finding ways to
overall volume thermal expansion b(1) for the matrix can be compute all the needed constants of the composite matrix except
determined from constituent information and a measurement of one, i.e., K (1), which is the frame bulk modulus. It is clear that we
K (1), the overall bulk modulus of the matrix. cannot do better than this without specifying the microstructure of
[70] One formula for computing Ks(1) on the basis of results of the matrix. These formulas all incorporate the effects of the
Berryman and Milton [1991] is microstructure through the presence of the measured quantity
K (1), which will itself depend quite strongly on the microstructure.
     a ð BÞ  a ð AÞ 
K ð1Þ K
ð1Þ
¼  K ð1Þ  h K i ; ðA5Þ
Ks Km K ð BÞ  K ð AÞ
Appendix B: Some Corrections
where
[74] This appendix contains Tables B1 and B2 that appeared
  previously in the paper by Berryman and Wang [1995]. Repeating
K K ð AÞ K ð BÞ these tables here serves two purposes: (1) There were some errors
 f ð AÞ þ f ð BÞ ðA6Þ
Km ð AÞ
Km Km
ð BÞ in the previously published versions that are corrected here. (Of
most importance, the value of B(1) for Weber sandstone in Table B1
was incorrect, and this error propagated into a number of the values
h K i  f ð AÞ K ð AÞ þ f ðBÞ K ðBÞ : ðA7Þ for Weber in Table B2.) (2) In section 4.3 we make some explicit
comparisons between the previous results and the results of the
[71] An alternative for computing Ks(1) by making use of the present work. These comparisons are facilitated by having the
same argument from Berryman and Milton [1992] that gives (46) corrected tables present here.
BERRYMAN AND PRIDE: DOUBLE-POROSITY MODELS IN GEOMECHANICS ECV 2 - 13

Table B1. Berryman and Wang [1995] Revised Material Properties


Parameter Chelsford Granite Weber Sandstone
K, GPa 8.0a 4.0a
Ks, GPa 54.5a 37.0a
a 0.85a 0.89a
K (1), GPa 17.0a 10.0a
n(1) 0.25 0.15
Ks(1), GPa 55.5a 38.0a
a(1) 0.69a 0.74a
Kf, GPa 3.3 3.3
f(1) 0.0011 0.095a
B(1) 0.992 0.738
R(1), GPa1 0.0409 0.208
v(2), GPa 0.011a 0.0095
a
From Coyner [1984].

Table B2. Berryman and Wang [1995] Revised Double-Porosity Parameters


Parameter Formula Chelmsford Granite Weber Sandstone
a11, GPa1 1/K 0.125 0.250
a12, GPa1 a(1)Ks(1)/K (1)Ks 0.0413 0.076
a13, GPa1 a/K  a12 0.0649 0.147
a22, GPa1 v(1)a(1)/B(1)K (1) 0.0405 0.0993
a23, GPa1 v(1)a(1)/K (1)  a12 0.00119 0.00270
a33, GPa1 v(2)/K f + v(1)/K (1)  (1  2a)/K + 2a12 0.0663 0.145
a33, GPa1
 a33  v(2)/Kf 0.0630 0.142
a(2) (a33a12  a13a23)/(a11a23  a13a12) 0.997 0.994
B (a12 + a13)/(a22 + 2a23 + a33) 0.973 0.891
B(1) (a12 + a23)/a22 0.992 0.738
B[u(1)] a12/a22 1.022 0.765
BEB(1) (a23a13  a12a33)/(a22a33  a223 ) 0.993 0.738
B(2) (a33a12  a13a23)/[a33(a12 + a23)  a23(a13  a33)] 0.950 0.980
B[u(2)] a13/a33 0.978 1.011
BEB(2) (a23a12  a13a22)/(a22a33  a223 ) 0.961 0.997
Ku, GPa [a11  (a12 + a13)2/(a22 + 2a23 + a33)]1 46.3 19.34
K (2), GPa v(2)(a12 + a13)/(a11a23  a13a12) 0.174 0.0666
R, GPa1 a/BK 0.1092 0.250
R(2), GPa1 a(2)/B(2)K (2) 5.87 15.24

Notation Kf an effective pore bulk modulus (unjacketed).


Kf(1), Kf(2) pore bulk moduli of constituents (unjacketed).
f (A), f (B) volume fractions of porous components A and B R storage coefficient, equal to a/BK.
in the matrix material, i.e., f (A) = v(A)/(v(A) + v(B)). V total volume.
k(11), k(22) matrix and fracture permeabilities. V (1), V (2) total matrix and fracture volumes.
pc confining pressure. Vs solid volume, equal to (1  f)V.
pd differential pressure, equal to pc  pf. Vf pore volume, equal to fV.
pf fluid pressure. a Biot-Willis parameter.
p(1)
f , pf
(2)
matrix and fracture fluid pressures. b volume thermal expansion coefficient.
pf(d ), pf(u) drained and undrained fluid pressures. g pressure difference coefficient.
ps pressure in the solid phase. z increment of total fluid content.
v(1), v(2) volume fractions occupied by matrix and frac- z(1), z(2) increments of matrix and fracture fluid content.
tures with v(1) + v(2) = 1. q temperature.
v(A), v(B) volume fractions occupied by porous components h fluid viscosity.
A and B with v(A) + v(B) + v(2) = 1 for misfit f total porosity, equal to v(1)f(1) + v(2)f(2).
model. f(1), f(2) matrix and fracture porosity (note fracture porosity
B Skempton’s coefficient. is unity for misfit model).
G shear modulus.
K bulk modulus of drained porous frame (jacketed).
K (1), K (2) bulk moduli of drained porous constituents
( jacketed). [75] Acknowledgments. J. G. B. thanks N. G. W. Cook and S. Ita for
Kf fluid bulk modulus. helpful conversations regarding reciprocity relations. The work of J.G.B.
Km material (or grain) bulk modulus. was performed under the auspices of the U.S. Department of Energy by the
Kp an effective pore bulk modulus ( jacketed), equal University of California Lawrence Livermore National Laboratory under
to fK/a. contract W-7405-ENG-48 and supported specifically by the Geosciences
Research Program of the DOE Office of Energy Research within the Office
Ks an effective solid bulk modulus (unjacketed). of Basic Energy Sciences, Division of Engineering and Geosciences. The
Ku bulk modulus of undrained (confined) porous work of S.R.P. was also supported in part by the French government and in
frame. part by the Stanford Rock Physics and Borehole Geophysics Project and
ECV 2 - 14 BERRYMAN AND PRIDE: DOUBLE-POROSITY MODELS IN GEOMECHANICS

sponsors while on sabbatical at Stanford University. All support for this mation response of dual porosity media, in Mechanics of Jointed and
work is gratefully acknowledged. Faulted Rock: Proceedings of an International Conference, Vienna,
18 – 20 April 1990, edited by H.-P. Rossmanith, pp. 681 – 688, A. A.
Balkema, Brookfield, Vt., 1990.
References Elsworth, D., and M. Bai, Flow-deformation response of dual-porosity
Bai, M., D. Elsworth, and J.-C. Roegiers, Modeling of naturally fractured media, J. Geotech. Eng., 118, 107 – 124, 1992.
reservoirs using deformation dependent flow mechanism, Int. J. Rock Fei, Y., Thermal expansion, in Mineral Physics and Crystallography,
Mech. Min. Sci. Geomech. Abstr., 30, 1185 – 1191, 1993. A Handbook of Physical Constants, AGU Ref. Shelf, vol. 2, edited
Barenblatt, G. I., and Y. P. Zheltov, Fundamental equations of filtration of by T. J. Ahrens, pp. 29 – 44, AGU, Washington, D. C., 1995.
homogeneous liquids in fissured rocks, Sov. Phys. Dokl., 5, 522 – 525, Gangi, A. F., Variation of whole and fractured porous rock permeability
1960. (English translation of Dokl. Akad. Nauk SSSR, 132, 545 – 548, with confining pressure, Int. J. Rock Mech. Min. Sci., 15, 249 – 257, 1978.
1960.) Gassmann, F., Über die elastizität poröser medien, Veirteljahrsschr. Natur-
Berge, P. A., H. F. Wang, and B. P. Bonner, Pore pressure buildup coeffi- forsch. Ges. Zürich, 96, 1 – 23, 1951.
cient in synthetic and natural sandstones, Int. J. Rock Mech. Min. Sci. Geertsma, J., The effect of fluid pressure decline on volumetric changes of
Geomech. Abstr., 30, 1135 – 1141, 1993. porous rocks, Trans. AIME, 210, 331 – 340, 1957.
Berryman, J. G., Effective stress for transport properties of inhomogeneous Hart, D. J., and H. F. Wang, Laboratory measurements of a complete
porous rock, J. Geophys. Res., 97, 17,409 – 17,424, 1992. set of poroelastic moduli for Berea sandstone and Indiana limestone,
Berryman, J. G., and G. W. Milton, Exact results for generalized Gassmann’s J. Geophys. Res., 100, 17,741 – 17,751, 1995.
equations in composite porous media with two constituents, Geophysics, Khaled, M. Y., D. E. Beskos, and E. C. Aifantis, On the theory of
56, 1950 – 1960, 1991. consolidation with double porosity, III, A finite element formulation,
Berryman, J. G., and G. W. Milton, Exact results in linear thermomechanics Int. J. Numer. Anal. Methods Geomech., 8, 101 – 123, 1984.
of fluid-saturated porous media, Appl. Phys. Lett., 61, 2030 – 2032, 1992. Kümpel, H.-J., Poroelasticity: Parameters reviewed, Geophys. J. Int., 105,
Berryman, J. G., and S. R. Pride, Volume averaging, effective stress rules, 783 – 799, 1991.
and inversion for microstructural response of multicomponent porous Love, A. E. H., A Treatise on the Mathematical Theory of Elasticity,
media, Int. J. Solids Struct., 35, 4811 – 4843, 1998. pp. 173 – 174, Dover, Mineola, N. Y., 1927.
Berryman, J. G., and H. F. Wang, The elastic coefficients of double-porosity Mavko, G., T. Mukerji, and J. Dvorkin, The Rock Physics Handbook,
models for fluid transport in jointed rock, J. Geophys. Res., 100, 24,611 – pp. 307 – 309, Cambridge Univ. Press, New York, 1998.
24,627, 1995. Norris, A. N., On the correspondence between poroelasticity and thermo-
Berryman, J. G., and H. F. Wang, Elastic wave propagation and attenuation elasticity, J. Appl. Phys., 71, 1138 – 1141, 1992.
in a double-porosity dual-permeability medium, Int. J. Rock Mech. Min. Pride, S. R., and J. G. Berryman, Connecting theory to experiment in
Sci., 37, 63 – 78, 2000. poroelasticity, J. Mech. Phys. Solids, 46, 719 – 747, 1998.
Beskos, D. E., and E. C. Aifantis, On the theory of consolidation with Rice, J. R., On the stability of dilatant hardening for saturated rock masses,
double porosity, II, Int. J. Eng. Sci., 24, 1697 – 1716, 1986. J. Geophys. Res., 80, 1531 – 1536, 1975.
Biot, M. A., General theory of three dimensional consolidation, J. Appl. Rice, J. R., and M. P. Cleary, Some basic stress diffusion solutions for
Phys., 12, 155 – 164, 1941. fluid-saturated elastic porous media with compressible constituents, Rev.
Biot, M. A., Theory of propagation of elastic waves in a fluid-saturated Geophys., 14, 227 – 241, 1976.
porous solid, I, Low-frequency range, J. Acoust. Soc. Am., 28, 168 – 178, Rosen, B. W., and Z. Hashin, Effective thermal expansion coefficients and
1956a. specific heats of composite materials, Int. J. Eng. Sci., 8, 157 – 173, 1970.
Biot, M. A., Theory of propagation of elastic waves in a fluid-saturated Skempton, A. W. , The pore-pressure coefficients A and B, Geotechnique,
porous solid, II, Higher frequency range, J. Acoust. Soc. Am., 28, 179 – 4, 143 – 147, 1954.
191, 1956b. Tuncay, K., and M. Y. Corapciaglu, Effective stress principle for saturated
Biot, M. A., Mechanics of deformation and acoustic propagation in porous fractured porous media, Water Resour. Res., 31, 3103 – 3106, 1995.
media, J. Appl. Phys., 33, 1482 – 1498, 1962. Wang, H. F., Theory of Linear Poroelasticity With Applications to
Biot, M. A., and D. G. Willis, The elastic coefficients of the theory of Geomechanics and Hydrogeology, Princeton Univ. Press, Princeton,
consolidation, J. Appl. Mech., 24, 594 – 601, 1957. N. J., 2000.
Brown, R. J. S., and J. Korringa, On the dependence of the elastic properties Wang, H. F., and J. G. Berryman, On constitutive equations and effective
of a porous rock on the compressibility of a pore fluid, Geophysics, 40, stress for deformable, double porosity media, Water Resour. Res., 32,
608 – 616, 1975. 3621 – 3622, 1996.
Carroll, M. M., Mechanical response of fluid-saturated porous materials, in Warren, J. E., and P. J. Root, The behavior of naturally fractured reservoirs,
Theoretical and Applied Mechanics, Proceedings of the 15th International Soc. Pet. Eng. J., 3, 245 – 255, 1963.
Congress of Theoretical and Applied Mechanics, Toronto, August 17 – 23, Wilson, R. K., and E. C. Aifantis, On the theory of consolidation with
1980, edited by F. P. J. Rimrott and B. Tabarrok, pp. 251 – 262, North- double porosity, Int. J. Eng. Sci., 20, 1009 – 1035, 1982. (Correction,
Holland, New York, 1980. Int. J. Eng. Sci., 21, 279 – 281, 1983.)
Christensen, R. M., Mechanics of Composite Materials, pp. 324 – 325,
John Wiley, New York, 1979.
Coyner, K. B., Effects of stress, pore pressure, and pore fluids on bulk
strain, velocity, and permeability of rocks, Ph. D. thesis, Mass. Inst. of 
Technol., Cambridge, 1984. J. G. Berryman, Lawrence Livermore National Laboratory, University of
Detournay, E., and A. H.-D. Cheng, Fundamentals of poroelasticity, California, P. O. Box 808 L-200, Livermore, CA 94551-9900, USA.
in Comprehensive Rock Engineering, vol. 2, edited by J. A. Hudson, (berryman1@llnl.gov)
pp. 113 – 171, Pergamon, New York, 1993. S. R. Pride, Université de Rennes 1, Campus Beaulieu, Bât. 15, F-35042
Elsworth, D., and M. Bai, Continuum representation of coupled flow-defor- Rennes Cedex, France. (spride@univ-rennes1.fr)
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 107, NO. B3, 10.1029/2000JB000108, 2002

4
a11
3

a22
a_ij K1

1
a33
a23
0
a12
1

2
a13
3

4
0.5 0.6 0.7 0.8 0.9 1
volume fraction of phase 1

Figure 4. The values of the aij are normalized by the drained modulus of the matrix phase K (1) = 1. 1010 Pa for the
model where the joint phase 2 is treated as a porous continuum in welded contact with the matrix phase 1. Here we
have taken K (2) = K (1)/20, Km(1) = Km(2) = KSi02 = 3.6 1010 Pa, f(1) = 0.2, and f(2) = 0.4.

ECV 2 - 11

You might also like