A Continuum Theory of Porous Media Saturated by Multiple Immiscible Uids: I. Linear Poroelasticity

International Journal of Engineering Science 40 (2002) 1807–1833
www.elsevier.com/locate/ijengsci
A continuum theory of porous media saturated by multiple

immiscible fluids: I. Linear poroelasticity
Changfu Wei *, Kanthasamy K. Muraleetharan
School of Civil Engineering and Environmental Science, University of Oklahoma, 202 W. Boyd Street, Room 334,
Norman, OK 73019-1024, USA
Received 20 April 2002; accepted 15 May 2002
(Communicated by A. BEDFORD)
Abstract
The mechanical behavior of porous media is largely governed by the interactions among coexisting
components. These interactions occur through interfaces. In this paper, a continuum theory of multiphase
porous media is developed that can be used to characterize the interactions among various components.
Central to the theory is the implementation of the dynamic compatibility conditions microscopically rep-
resenting the constraints on the pressure jumps across the interfaces. It is shown that capillary relaxation
processes are thermodynamically associated with the changes in the volume fractions of fluids. A linear
model is developed by a formal linearization of the proposed theory. For fully saturated conditions, the
linearized theory reduces to the Biot’s poroelasticity model. A procedure to evaluate the material constants
is presented for the porous media with two fluids. The linear model is utilized to analyze the propagation of
acoustic waves in an unsaturated rock. The theoretical results are compared with the experimental data
available in the literature.
Ó 2002 Elsevier Science Ltd. All rights reserved.
Keywords: Continuum theory; Multiphase porous media; Theory of mixtures with interfaces; Dynamic compatibility
conditions on interfaces; Poroelasticity; Acoustic wave
1. Introduction
Study of the behavior of porous media is of great interest in a number of diverse fields, such as
civil and environmental engineering, petroleum engineering, geophysics, chemical engineering,
*
Corresponding author. Tel.:+1-405-325-4228; fax: +1-405-325-4217.
E-mail address: wei-1@ou.edu (C. Wei).
0020-7225/02/$ - see front matter Ó 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 2 0 - 7 2 2 5 ( 0 2 ) 0 0 0 6 8 - X
1808 C. Wei, K.K. Muraleetharan / International Journal of Engineering Science 40 (2002) 1807–1833
and biomechanics. A porous medium is an assemblage of solid particles forming a matrix (or
skeleton) whose voids are filled with several fluids, e.g., water, oil, air and gas. The overall be-
havior of a porous medium is by and large controlled by the interactions among various coex-
isting components. These interactions must be properly characterized in an analysis procedure of
porous media.
The first theory of porous media was developed by Biot [1–3] and can be used to analyze the
linear elastic behavior of porous media. This theory was later generalized to account for the
nonlinear effects by Biot [4], Prevost [5], and Zienkiewicz et al. [6]. Thus far, most of the pro-
cedures used in analyzing the behavior of porous media are based on these models and include
two key components, one the effective stress principle [7] and the other the Darcy’s law [8]. The
effective stress principle has found many successful applications in modeling saturated geoma-
terials. In some cases, however, it is very difficult to apply the effective stress principle. For
instance, in modeling the rock masses with occluded porosity and swelling soils, it has been
found that the effective stress cannot be defined in the customary sense [9–11]. In addition, the
difficulties in developing an effective stress equation for unsaturated soils have been recognized
[12–14].
The models cited above are heuristic in nature since they impose no restrictions on the con-
stitutive relationships of the porous media. To apply these models, the constitutive relationships
must be intuitively introduced. In order to take into account the irreversible phenomena such as
plastic deformation, Biot [15] proposed an important concept in poromechanics called the prin-
ciple of virtual dissipation. Although heuristic, the principle of virtual dissipation is very general
and accounts for finite deformation, plastic deformation, molecular diffusion, heat conduction
and fluid flow. Recently, viewing porous media as open continua, Coussy [16] generalized Biot’s
theory into the modern context of thermodynamics.
In spite of their simplicity, the Biot’s theory and its generalizations fail to consider the mi-
croscopic structure of porous media that plays critical roles in the overall behavior of porous
materials. In addition, it remains uncertain whether this kind of theories can be used to model the
plastic deformation of multiphase porous media. Alternatively, the theories of mixtures can be
used to describe the behavior of porous media and they are particularly suited to take various
coupling phenomena into account. Comprehensive reviews on this subject can be found in Ref.
[17,18]. In the mixture theory-based models, some variables such as volume fractions are intro-
duced to represent the microstructure of porous media. As a consequence, complementary
equations must be developed so that a complete description of porous media is achieved. These
complementary equations are usually called the closure equations.
Conventionally, the closure equations are obtained in one of the following ways based on the
theory of mixtures. Complementary equations are introduced for the volume fractions or other
variables representing the microstructure [19,20]; the volume fractions are treated as internal
variables and corresponding evolution equations are developed [21,22]. Alternatively, additional
balance equations can also be directly derived [23–25]. Although these approaches have found
successful applications in some particular circumstances, they fail to account for some important
mechanisms occurring in porous media. For instance, the hysteresis of capillary pressure in
multiphase porous media has not been satisfactorily simulated based on any of the above theories,
and the effects of capillary pressure on the behavior of multiphase porous media remain poorly
understood.
C. Wei, K.K. Muraleetharan / International Journal of Engineering Science 40 (2002) 1807–1833 1809
In this paper, a continuum theory of multiphase porous media is developed that is capable of
characterizing the microscopic interactions between the coexisting components. The key step in
this development is the implementation of the dynamic compatibility conditions on interfaces
[26]. In order to reflect the inclusion of these interfacial conditions, the theoretical frame-
work presented in this paper is named the theory of mixtures with interfaces (TMI). Parallel to
Hassanizadeh and Gray’s procedure [27], a set of general constitutive relationships for the po-
rous media with two fluids has been presented by Muraleetharan and Wei [28]. In this paper,
the results previously obtained are generalized to incorporate the dynamic compatibility con-
ditions on interfaces and a linear thermoporoelastic model is presented. In the companion
paper [29], the variational structure and lagrangian formulations of the proposed theory are
developed.
2. Dynamic compatibility conditions on interfaces
One of the salient features of porous media is the existence of interfaces among various co-
existing components. At a microscopic scale, the mechanical interactions occurring on the in-
terface between two components can be categorized into two groups. One is the drag force due to
the relative motions of bulk components in the direction tangential to the interface; the other is
due to the material impenetrability and represents some sort of force equilibrium, i.e. the capillary
equilibrium, in the direction normal to the interface. The first type of interactions has a macro-
scopic counterpart, i.e. the hydrodynamic drag that can be taken into account by using the flow
equations of Darcy’s type [8]. An example of the second type of interactions at macroscale is the
relationship between the capillary pressure and degree of saturation. Although this relationship is
usually introduced in the analysis of two-phase flow in porous media, its characters (e.g., hys-
teresis) remain poorly understood. Thus far, little is known about the coupling effects between
deformation and capillary pressure.
To gain an insight into the problem, we consider the capillary equilibrium from a microscopic
a
point of view (see Fig. 1). Let pmicro be the microscopic pressure of a-component. At equilib-
a b
rium, the pressure difference (pmicro pmicro ) on ab-interface cannot arbitrarily change, and instead
it is a function of the surface tension, the temperature, the local geometry of pores, and the local
fluid content. This is the so-called dynamic compatibility condition on an interface [26]. These
Fig. 1. Microscopic structure of porous media with two-phase fluids.

dynamic compatibility conditions can be averaged onto the macroscopic scale to yield the rela-
tionships between the capillary pressures and some macroscopic state variables N (e.g., the degree
of saturation) [30]. Hence, we can write pa pb ¼ vab ðNÞ, where pa is the (averaged) pressure of a-
phase and vab is a function of state variables N. In the limit case where the porous medium is
constituted by incompressible solid grains and saturated by a single fluid, the material impene-
S W
trability yields the constraint pmicro ¼ pmicro on the interface. As a consequence, pS ¼ pW , i.e. the
capillary pressure is zero. Based on the theory of porous media, it can be easily proven that the
zero-capillary pressure condition yields the Terzaghi’s effective stress equation for the saturated
porous media [5,10]. In general, however, the capillary pressure (pa pb ) does not vanish.
Capillary equilibrium is achieved through local fluid flow. The local fluid flow is a capillary
relaxation process due to the fluid exchange driven by the nonequilibrium capillary forces between
the pores of different sizes. Although the local fluid flow may be substantially influenced by the
macroscopic fluid flow driven by pressure gradient or vice versa, both have little in common. In
general, the capillary relaxation time is of the same order as or even smaller than the characteristic
times of macroscopic fluid flow [31] and the skeleton deformation. For swelling soils, where
the effects of the capillary relaxation become significant on the overall behavior of the material,
the effective stress equation cannot be defined in a usual sense [10,11]. Furthermore, due to the
capillary relaxation and irreversibility, the change of capillary pressures with the material state is
irreversible and accompanied by hysteresis. These phenomena have been extensively observed in
the laboratory.
Clearly, the effective stresses and the capillary pressures are closely correlated through the
dynamic compatibility conditions on interfaces. At the macroscopic scale, the dynamic compat-
ibility conditions on interfaces can be represented by the relationships between the capillary
pressures and state variables. Since these relationships are independent of any balance equations
and constitutive relationships for an individual component, they can be utilized as the closure
equations. Hence, in order to develop a continuum model of porous media, it becomes crucial to
properly simulate the change of the capillary pressures with the material state.
3. Kinematics and balance equations
The porous media of concern here consist of a solid skeleton (S) and k immiscible fluids. In
what follows, character a (or b) is used to denote an individual bulk component, and unless
otherwise explicitly specified, a, or b ¼ S; 1; 2; . . . ; k. Juxtaposed character ab represents an in-
terface between a and b components. Character f denotes a fluid, and unless otherwise specified,
f ¼ 1; 2; . . . ; k. Standard solid mechanics sign convention is used, i.e. tensile stresses are positive.
Let Ba R3 be a fixed but otherwise arbitrary reference configuration of a-phase. As usual in
the theory of mixtures, each individual component is assigned an independent motion defined by
vat : Ba 7!St R3 ; i:e: x ¼ vat ðX a Þ ¼ va ðX a ; tÞ; ð1Þ
where St R3 is the current configuration, and t 2 ½0; T

Rþ the elapsing time; X a represents
the coordinate of a particle (denoted by pa ) of a-phase in its reference configuration; x is the
spatial position of pa at time t. Function va ðX a ; tÞ is assumed to be invertible and differentiable as

many times as necessary.
The velocity and acceleration of pa are respectively defined by
ova ðX a ; tÞ
va ðx; tÞ ¼ V a ðX a ; tÞ ¼
ot
and
o2 va ðX a ; tÞ ova ðx; tÞ
aa ðx; tÞ ¼ Aa ðX a ; tÞ ¼ ¼ þ va rva ;
o2 t ot
where r represents the spatial gradient, and the operator (o=ot þ va r) is the material derivative
with respect to the motion of a-phase and denoted by Da =Dt.
The relative velocity of a fluid is
wf ðx; tÞ ¼ vf ðx; tÞ vS ðx; tÞ: ð2Þ
The deformation gradient of the solid skeleton F is defined by
ovS ouS
F¼ ¼ I þ ; ð3Þ
oX S oX S
where uS is the displacement of the solid skeleton and equals vS ðX S ; tÞ X S ; I is the second-order
isotropic tensor with component dij , where dij is the Kronecker delta. In the following, we assume
that transformation between configurations preserves the orientation, and hence J ¼ det F > 0.
The Green–Lagrangian strain tensor is
E ¼ 12ðF T F IÞ: ð4Þ
The local forms of balance equations introduced in the following were developed in [32,33]
based on a local averaging procedure. From now onwards, we assume that an interface carries no
averaged thermodynamical properties. This assumption is general enough for the development of
a continuum model, since the macroscopic properties of the interfaces are much smaller than their
bulk-phase counterparts. For simplicity, mass exchange (or phase change) between two compo-
nents is excluded and it is assumed that all the components have the same temperature, i.e.
hS ¼ hf ¼ hðx; tÞ.
Mass balance equations are
Da ðna qa Þ
þ na qa r va ¼ 0; ð5Þ
Dt
where na and qa denote the volume fraction and true mass density of a-component, respectively.
Linear momentum balance equations are
Da v a X
a a
n q r ðna ta Þ na qa ba ¼ ba ;
T ð6Þ
ab
Dt b6¼a
P b a represents the rate of
where b6¼a is the summation over S; 1; 2; . . ., and k but excludes a; T ab
linear momentum transferring to a-component through ab-interface;. ba is the external supply of
linear momentum, and we assume for simplicity that ba ¼ bðx; tÞ; ta is the true averaged Cauchy
stress tensor of a-component. Assuming that there are no moment of momentum interactions
among various components, it follows that ta ¼ ðta ÞT .
Energy balance equations are
Da E a X
na qa na ta : d a þ r ðna qa Þ na qa ha ¼ ba ;
Q ð7Þ
ab
Dt b6¼a
where Ea is the internal energy of a-component; ha is the external supply of heat; qa the heat flux;
b a the rate of energy transferring to a-component through ab-interface; ta : d a ¼ ta dijb and
Q ab ij
va va ¼ vai vai , i (¼1, 2, or 3) represents the spatial direction; d a is the symmetrical part of the
velocity gradient.
Entropy balance equations are
Da ga X
na qa þ r ðna ua Þ na qa xa ¼ b a þ Ka ;
U ð8Þ
ab
Dt b6¼a
where ga is the entropy density of a-component; ua the surface flux of entropy and it is assumed
that ua ¼ qa =h; h is the temperature; xa is the external supply of entropy and xa ¼ ha =h; Ka the
net production of entropy in a-phase; U b a the rate of entropy transferring to a-component
ab
through ab-interface.
Due to the singularity of interfaces, (5)–(8) are subjected to the following constraints
0¼T bb ;
ba þ T ð9Þ
ab ab
X
0¼ Q b c wc ;
bc þ T ð10Þ
ab ab
c¼a and b

0¼ Uba þ U
bb ; ð11Þ
ab ab
and
X
K¼ Ka P 0: ð12Þ
a
As the requirement of the second law of thermodynamics, inequality (12) states that the total net
production of entropy in porous media is nonnegative in any thermodynamic process.
For later discussions, it is useful to define the total mass density q, the total (Cauchy) stress
tensor r, and the total entropy density g of the porous medium as
X
q¼ na qa ; ð13Þ
a
X
r¼ na t a ; r ¼ rT ; ð14Þ
a
and
X
qg ¼ na qa ga : ð15Þ
a
In addition, the Helmholtz free energy Aa of a-component is defined by the following Legendre
transformation
Aa ¼ Ea hga : ð16Þ
4. General constitutive relationships
The state of a porous medium can be determined by independent state variables na , qa , va and h.
To achieve a complete description of the porous medium, constitutive relationships
P must be de-
b a . Note
veloped for the following dependent state parameters: Ea (or Aa ), ga , qð¼ a na qa Þ, ta , T ab
a f
that since h ¼ h ¼ h, we only need to develop a constitutive relationship for the total heat flux of
the mixture, i.e. q.
For simplicity, it is assumed that the porous media consist of an elastic matrix saturated by k
inviscid fluids. By introducing the principle of local action [34] and in a way similar to that fol-
lowed in [21], the dependent state variables in general can be assumed as functions of
n o
N ¼ h; rh; nf ; rnf ; n_ f ; qa ; rqa ; F; wf :
Among the above constitutive variables, the volume fractions of fluids are introduced to represent
the microstructure of the porous medium; the rates of volume fractions are also included so that
capillary relaxation can be taken into account. In the proposed theory, no internal constraints are
introduced and all the bulk components are arbitrarily compressible. Hence, the real mass den-
sities of bulk components qa are introduced above as constitutive variables. As shown in [35], any
internal constraints on compressibility will exclude certain modes of acoustic waves in the porous
medium.
The free energy functions of individual components are postulated as
AS ¼ A^S ðh; qS ; EÞ ð17Þ
and

Af ¼ A^f h; qf ; nf : ð18Þ
In postulating the above constitutive equations, we have assumed that the free energy density of
an individual component is solely determined by the state variables of that component. Since the
volume fraction of solid can be determined by integrating the mass balance equation of the solid
phase with E and qS given, nS is not explicitly included in (17). Due to the requirement of the
material objectivity, E is used in (17) instead of F. In addition, since the influence of deforma-
tion on a fluid can be implicitly taken into account by using the volume fraction, E is excluded in
(18).
Following a way similar to that in [28] and noting that interfaces carry no macroscopic ther-
modynamic properties, we can show that the sufficient and necessary conditions for (17) and (18)
to satisfy the entropy inequality (12) are
oAa
ga ¼ ; ð19Þ
oh
tf ¼ pf I; ð20Þ
tS ¼ te pS I; ð21Þ
and
X X q
DC ¼ n_ f Pf wf rf rh P 0; ð22Þ
f f
h
P
where DC ¼ hK ¼ a Ka ,
2 oAa
pa ¼ ðqa Þ ; ð23Þ
oqa
oAS T
t e ¼ qS F F ; ð24Þ
oE
oAf
Pf ¼ pf pS nf qf ; ð25Þ
onf
X
f
f
rf ¼ b f þ pf nf qf oA rnf nf qf oA rh;
T ð26Þ
fa
a6¼f
onf oh
and
X
q¼ na qa : ð27Þ
a
Parameter pa is the thermodynamic pressure of a-component. Eq. (20) implies that pf equals to the
true material pressure of an inviscid fluid.
When the porous medium arrives at equilibrium, the constitutive variables become

Neq ¼ hþ ; 0; nfþ ; 0; 0; qaþ ; 0; F þ ; 0 ;
where ð Þeq denotes the set of state variables at equilibrium. Furthermore, at equilibrium, the
dissipative energy function DC attains its minimum. It follows that

oDC
¼0 ð28Þ
ozc Neq
and

o2 DC
det P 0; ð29Þ
ozc ozd Neq
where zc or zd is an element in fn_ f ; wf ; rhg, f ¼ 1; 2; . . . ; k. Eq. (28) yields
Pfeq ¼ 0; ð30Þ
!
X
rfeq ¼ bf
T ¼ 0; ð31Þ
fa
a6¼f eq
and
qeq ¼ 0: ð32Þ
Clearly, Pf has two contributions: an equilibrium part Pfeq and a dissipative part P b f . From Eq.
(25), we may conclude that P b represents the capillary relaxation processes discussed in Section 2.
f
It is remarkable that in general the capillary pressures (pf pS ) are nonzero due to the dependence
of the free energy of a fluid on its volume fraction. This result is different from that in [21], where
zero capillary pressures are predicted at the so-called shifting equilibrium. In practice, zero
capillary pressure has never been observed for a multiphase porous medium. The capillary
pressures now can be expressed as
oAf b f:
pf pS ¼ nf qf þP ð33Þ
onf
Eq. (33) shows that some kind of constraints is imposed on the pressure difference between a fluid
and the solid (i.e. the capillary pressure). It is quite obvious that this equation represents the
macroscopic counterpart of the dynamic compatibility conditions on interfaces discussed in
Section 2. P
Similarly, (26) and (31) imply that the linear momentum exchanges terms a6¼f T b f have two
fa
contributions and can be generally expressed as
X
f
f
b f f f f oA f f f oA
Tfa ¼ p n q f
rn n q rh ^rf ; ð34Þ
a6¼f
on h
where ^rf is the dissipative part of the total linear momentum exchange between f-fluid and other
components.
Inserting (34) into the linear momentum balance equations (6) for f-fluid, we obtain the fol-
lowing flow equation
Df vf
nf qf nf qf b þ nf qf rGf ¼ ^rf ; ð35Þ
Dt
where Gf is the chemical potential of f-fluid defined by
pf
Gf ¼ A f þ : ð36Þ
qf
We now specify the dissipative forces P b f , ^rf and ^q, where ^q is the dissipative part of q=h.
Keeping in mind the equilibrium conditions (30)–(32), one obtains the following residual dissi-
pation inequality
X X
bf
n_ f P wf ^rf rh ^
q P 0: ð37Þ
f f
It is important to note that the capillary relaxation represented by P b f is thermodynamically

associated with n_ f . Assuming the volume fraction of a fluid can be additively decomposed into a
recoverable part and an irrecoverable part, we can show that the hysteresis in the capillary
pressure for the multiphase flow in the porous medium can be simulated within a framework
similar to the classical visco-elastoplasticity. Further development goes beyond the scope of this
paper, and the interested readers are referred to [35] for a detailed account.
As usual in the theory of mixtures (e.g., [21]), we assume that there exists at least one admissible
thermodynamic process so that (37) is satisfied. It follows that
X
^rf ¼ l^fr wr l
^fh rh; ð38Þ
r¼1;...;k
X
q¼
^ ^hr wr l
l ^hh rh; ð39Þ
r¼1;...;k
and
X c^f
bf ¼
P r r
n_ ; ð40Þ
s f
r¼1;...;k r
where coefficients l ^fr and l ^hh represent fluid drag force interactions and heat conductivity effects,
^hr and l
respectively; l ^fh are the coefficients representing the cross effects between hydraulic and
thermal effects; c^r are material constants and sfr represent the characteristic times for the capillary
f
relaxation processes. Inserting (38)–(40) into (37) yields the following dissipative energy
!
XX c^rf r f X
D¼ n_ n_ þ wr l
^rf wf þ wf l
^fh rh þ rh l
^hf wf þ rh lhh rh P 0:
f r
srf f
ð41Þ
For a physically possible thermodynamic process, certain restrictions must be imposed on the
coefficients in (38)–(40). In fact, inequality (41) is satisfied for any values of fn_ f ; wf ; rhg if and
only if (29) is fulfilled. This condition can also be deduced from the Onsager’s principle [36].
5. Linearization of the theory
In this section, we develop a linear thermoporoelasticity model based on the theory presented
above. In the above theory, the free energy functions of individual constituents, i.e. A^S ðh; qS ; EÞ
and A^f ðh; qf ; nf Þ, remain to be specified. Without losing generality, we assume that the porous
medium is initially at an equilibrium state represented by

fh; nf ; qa ; E; wf g ¼ h0 ; nf0 ; qa0 ; 0; 0 :
Under a small external disturbance, the porous medium arrives at a new state represented by

h0 þ h; nf0 þ nf ; qa0 þ qa ; E; wf :
From now onwards, h, nf , qa , E, and wf are all viewed as incremental quantities.

The linear model is obtained by taking quadratic polynomials for the free energies AS and Af .
Consequently, we postulate that

S 2
S S ^S 1 S 1 S 2 1 S q qS qS
n0 q0 A ¼ n0 E : D : E n0 cS h þ n0 KS S nS0 kSqh h S nS0 hJ She : E þ nS0 S J Sqe : E
2 2 2 q0 q0 q0
ð42Þ
and

f 2
1 f 2 1 f q 1 f f f 2 qf f q
f
nf0 qf0 A^f ¼ n0 cf h þ n0 Kf f
þ n0 H n ðn Þ nf f
0 kqh h f
þ nf f
0 kqn n f
nf0 kfhn hnf ; ð43Þ
2 2 q0 2 q0 q0
where A^S and A^f are the incremental values of the corresponding free energy functions. Note that
in general the coefficients in (42) and (43) are the functions of the initial temperature h0 , the initial
volume fractions nf0 , and the initial mass density qa0 .
By introducing (42) and (43), it follows from (23) and (24) that
qS
p S ¼ KS þ J Sqe : E kSqh h; ð44Þ
qS0
qf
p f ¼ Kf þ kfqn nf kfqh h; ð45Þ
qf0
and
qS
te ¼ D : E þ J Sqe : I 4 J She : I 4 h; ð46Þ
qS0
where pa and te are incremental values; I 4 is the isotropic fourth-order tensor with components
ðI 4 Þijkl ¼ 1=2ðdik djl þ dil djk Þ; Ka is the bulk modulus of an individual component; J Sqe represents
the coupling between the skeleton deformation and the compression of the solid component; kfqn
accounts for the coupling between the mass density change and the volume fraction change of
a fluid; kaqh can be decomposed as kaqh ¼ 3Ka aah , where aah is the thermal dilation coefficient of
an individual component; J She represents the thermal dilation effect of the skeleton; D is the
isothermal elastic tangent modulus of the skeleton and it is a fourth-order tensor.
Assuming that the material is isotropic, we have
J Sqe ¼ kSqe I; ð47Þ
J She ¼ kShe I; ð48Þ
and
D ¼ kS I I þ 2lS I 4 ; ð49Þ
where kS and lS are the Lame’s constants of the solid skeleton; I is the isotropic second-order
tensor with components dij ; denotes the tensor product.
The total Cauchy stress, i.e. (14), now reads
qS X qa X
r ¼ nS0 kS kSqe ðI : EÞI þ 2nS0 lS E þ nS0 kSqe S I na0 Ka a I nf0 kfqn nf I
q0 a
q 0 f
!
X
þ nS0 kShe þ na0 kaqh hI: ð50Þ
a
The specific entropy of a bulk component is obtained by linearizing (19), i.e.

qS
qS0 gS ¼ cS h þ kSqh þ kShe I : E ð51Þ
qS0
and
qf
qf0 gf ¼ cf h þ kfqh þ kfhn nf : ð52Þ
qf0
By definition, the specific heat capacity of an individual component is given by cav ¼ h0 ca .

The linear form of the total specific entropy density of the mixture, i.e. (15), is given by
X X qa X
q0 g ¼ na0 ca h þ na0 kaqh a þ nS0 kShe I : E þ nf0 kfhn nf : ð53Þ
a a
q0 f
Closure equations (33), which represent the macroscopic counterparts of the dynamic com-
patibility conditions on interfaces, are linearized and it follows that
qf b f;
pf pS ¼ nf0 Hfn nf þ nf0 kfqn nf0 kfhn h þ P ð54Þ
qf0
where Hfn is a parameter accounting for the effects of the volume fraction changes on the capillary
pressure. Finally, the incremental chemical potential of a fluid is obtained by linearizing (36) and
qf
qf0 Gf ¼ Kf þ kfqn nf kfqh h: ð55Þ
qf0
The constitutive relationships described above include the following material parameters: kS ,
lS , kSqe , Ka , kfqn , Hfn , ca , kShe , kfhn , and kaqh , where a ¼ S; 1; 2; . . . ; k and f ¼ 1; 2; . . . ; k. It must be
noted that for a model to be physically possible, certain restrictions must be imposed on the above
material constants. These restrictions come from the requirements that the free energies repre-
sented by (42) and (43) always have nonnegative values. It follows that
2
3kS þ 2lS > 0; lS > 0; KS > 0; kS þ 23lS KS P kSqe ;
2
Kf > 0; Hfn P 0; Kf Hfn P kfqn ;

ðk þ 2=3k Þ kSqe kShe
S S

kSqe KS kShe > 0;

kShe kShe cS
and

Kf kfqn kSqh

f
kqn Hfn kShn > 0:
S
kqh kShn cf
The mass balance equation, i.e. (5), now can be cast into the following linear forms
nS0 oqS onS

þ þ nS0 r vS ¼ 0 ð56Þ
qS0 ot ot
and
nf0 oqf onf

þ þ nf0 r vf ¼ 0: ð57Þ
qf0 ot ot
The equation of motion for the porous medium is obtained by summing up the linear mo-
mentum balance equations over all the phases. By incorporating (9), the linear counterpart of this
equation is expressed by
X
S
o2 ua S S S S S S S q
na0 qa0 2
¼ n0 kS þ lS kqe rr u þ n0 lS r ru þ n0 kqe r S
a
ot q0

a X !
X q X
a f f f S S a a
n0 Ka r a n0 kqn rn þ n0 khe þ n0 kqh rh: ð58Þ
a
q0 f a
Substituting (55) in the linear form of (35), one obtains the following linear flow equations:

f
o2 uf q
nf0 qf0 ¼ n0 kqn rn n0 Kf r f þ nf0 kfqh rh þ ^rf ;
f f f f
ð59Þ
ot2 q0
where ^rf is given by (38). Subtraction of (58) from (59) repeatedly over all the fluids yields the
linear form of the linear momentum balance equation of the solid skeleton, i.e.
o2 uS
qS
S S
nS0 qS0 S S S S S
¼ n0 kS þ lS kqe rr u þ n0 lS r ru n0 KS kqe r S
ot2 q0
X
þ nS0 kSqh nS0 kShe rh ^rf : ð60Þ
f
Incorporating (9), (10), (16), (19)–(21), (23)–(27), (33), (34) and (36), we can derive the total
energy balance for the porous medium by summing up (7) over all the phases, that is,
S !
D ðqgÞ X X X
h þ qgdiv vS þ div q þ hgf nf qf wf qh ¼ b f n_ f
P wf ^rf :
Dt f f f
The linear form of the above total energy balance equation is

" #
og X
f f f f
h0 q0 þ h0 g0 n0 q0 div w þ div q ¼ 0; ð61Þ
ot f
where the change in the total external supply of heat h has been omitted. Eq. (61) shows that the
change in the total entropy of the mixture has two contributions: the fluid flow and the heat
conduction. Inserting (53) into (61) and noting that q ¼ h0 ^q for a linear model, we obtain
X
X
" #
oh X na0 kaqh oqa f f on
f
S S
X
a
n0 ca þ a
þ n0 khn þ n0 khe n0 q0 g0 div vS
f f f
a
ot a
q0 ot f
ot f
X
f f f f
þ n0 q0 g0 div v þ div ^
q ¼ 0; ð62Þ
f
where gf0 is the initial entropy density of a fluid; q^ is given by (39).

Thus far, all the equations governing the linear thermoelasticity of the porous media saturated
by k fluids have been presented. These governing equations include the mass balance (56) and
(57), the linear momentum balance (59) and (60), the energy balance (62), and the closure
equation (54) as well as constitutive equations (38)–(40), (44) and (45). Note that the total number
of the governing equations is equal to the number of the independent state variables including qa ,
va , h and nf , where a ¼ S; 1; 2; . . . ; k and f ¼ 1; 2; . . . ; k.
6. Linear poroelasticity of unsaturated porous media
In this section, the linearized theory presented above will be used to develop a linear elastic model
of porous media saturated by two immiscible fluids––a wetting fluid (W) and a nonwetting fluid
(N). Here, k ¼ 2, a ¼ S; W; N and f ¼ W; N. It is assumed for simplicity that the material is iso-
tropic and in an isothermal condition. The effect of capillary relaxation is omitted. That is, capillary
equilibrium can be achieved immediately and characteristic times srf in (40) become infinite so that
Pf ¼ 0. In addition, cross effects in the fluid drag forces are neglected and (38) becomes
^rf ¼ ^
lff wf ¼ ^
lf wf ; ð63Þ
where use of l^ff ¼ l^f I has been made.

In the isothermal condition, the closure equation (54) becomes
qf
pf pS ¼ nf0 Hfn nf þ nf0 kfqn : ð64Þ
qf0
Governing equations can be cast into a more compact form by eliminating nf and qa from (59)
and (60). Integrating (56) and (57) with respect to time yields, respectively,
qS
nS0 þ nS þ nS0 r uS ¼ 0 ð65Þ
qS0
and
qf
nf0 þ nf þ nf0 r uf ¼ 0; ð66Þ
qf0
where nS ¼ nW nN . With incorporation of (44), (45) and (64)–(66), eliminating qa and nf from
(59) and (60) yields
o2 uS
nS0 qS0 ¼ M SS þ n S
0 l S rr uS þ nS0 lS r ruS þ MSW rr uW þ MSN rr uN
ot2
X
ouf ouS
þ l^f ; ð67Þ
f¼W;N
ot ot

W
o2 uW W ou ouS
nW W
0 q0
S W N
¼ MSW rr u þ MWW rr u þ MWN rr u l^ ; ð68Þ
ot2 ot ot
and

N
o2 uN N ou ouS
nN N
0 q0
S W N
¼ MSN rr u þ MWN rr u þ MNN rr u l^ ; ð69Þ
ot2 ot ot
where the coefficients MSS , MSW , MSN , MWW , MWN , and MNN are given in Appendix A. Note that
the lineal elastic model represented by (67)–(69) possesses a symmetrical structure.
Eqs. (67)–(69) are similar to the generalized formulations of Biot’s theory used in the analyses
of the wave propagation in porous media, see, for instance, Brutsaert [37], Bowen [21], Garg and
Nayfeh [38], and Tuncay and Corapcioglu [39] among others. However, the model presented here
differs from those cited above in that it rigorously takes into account the dynamic compatibility
conditions on interfaces. It is noted that, in the generalized Biot’s models, the relationship be-
tween the capillary pressure and moisture content, a macroscopic counterpart of the dynamic
compatibility conditions on interfaces, was introduced intuitively. Moreover, this relationship was
used in the generalized Biot’s models only to help obtain some material constants. In the model
presented above, however, the relationships between capillary pressures and state variables are
considered as the original components of the model and this treatment is thermodynamically
consistent.
Eqs. (67)–(69) are applicable only to the porous materials where each fluid phase is intercon-
nected. When the content of the wetting fluid increases up to a large value, the nonwetting fluid
will be disconnected and trapped in the wetting one and both fluids will macroscopically move
together. In such cases, one may set nN 0 ¼ 0, and (67)–(69) reduce to
o2 uS
nS0 qS0 ¼ r½ðA þ NÞr uS þ Qr uW
þ N r ruS þ l^W wW ð70Þ
ot2
and
o2 uW
n0 qW
0 ¼ rðQr uS þ Rr uW Þ l^N wW : ð71Þ
ot2
This set of equations takes the same form as the Biot’s poroelasticity model [3]. For comparison,
the same notation is used here. Coefficients A, N, Q, and R are given in Appendix B.
We now evaluate the material constants in the governing equations (67)–(71). Suppose that the
initial porosity n0 of the porous medium and the degree of saturation of the wetting fluid Sr are
known. The following materials parameters remain to be evaluated:
kS ; lS ; kSqe ; Ka ; kfqn ; Hfn ; l^f ;
where a ¼ S; W; N; f ¼ W; N. The bulk modulus Ka of an individual component can be deter-

mined in the laboratory. In general, however, other material parameters are seldom transparent in
experiments. To evaluate these parameters relationships must be established between the above
material constants and the phenomenological parameters that can be measured through experi-
ments.
We first note that the coefficients l^f of drag forces can be related to the permeability of the
porous medium by the following relationship
ðnf Þ2 mf
l^f ¼ ;
kkrf
where k is the intrinsic permeability of the porous medium; krf is the relative permeability of f-fluid
and it is a function of the degree of saturation; mf is the viscosity of f-fluid. Both k and krf can be
evaluated in the laboratory [40].
Parameter kfqn represents the coupling between the mass density and the volume fraction of a
fluid. In the following, we assume that such coupling effects are negligible, i.e. kfqn 0. Equiva-
lently, we assume that in isothermal conditions the mass density of a fluid is a function of its
pressure only. Material parameters kS , lS , and kSqe account for the elastic deformation of the solid
skeleton. Since these parameters are independent of moisture content, they can be obtained
through the following three tests on saturated samples.
The first test is to measure the shear modulus G of a saturated sample. Constant lS is obtained by
G ¼ nS0 lS : ð72Þ
The other two tests are the compression tests under fully drained and undrained conditions, which
can used to measure the drained modulus KD , the undrained bulk modulus KU , and the pore
pressure coefficient B.
In an isothermal condition, if the porous medium becomes fully saturated, nS0 ¼ 1 n0 , nW0 ¼ n0
and nN0 ¼ 0, where n 0 is the initial porosity of the porous medium. The mass densities and volume
fractions can be eliminated from (50) with introducing (64)–(66), and we obtain the following
linear stress-strain relationship for the fully saturated porous media
h i
r þ aB pW I ¼ nS0 kS þ k0 ðI : EÞI þ 2nS0 lS E; ð73Þ
where aB is the Biot’s coefficient [43]

S
nS0 n0 nS0 HWn þ kqe
aB ¼ 1 ; ð74Þ
n0 nS0 HWn þ KS
and

S S
n0 nS0 HW
n k qe þ 2n 0 n S W
0 n kqe
H
k0 ¼ :
n0 nS0 HW
n þ KS
In the fully drained compression test, the pore liquid pressure is constant and therefore pW
vanishes. Using (73), we show that the drained modulus is

2
KD ¼ nS0 kS þ lS þ k0 : ð75Þ
3
For the fully undrained compression test, it can be easily proven that the pressure change
depends on the volumetric strain of the skeleton through
pW ¼ paB ðI : EÞ; ð76Þ
where

KS þ n0 nS0 HW
n KW
p¼ S : ð77Þ
n0 KW þ n0 KS þ n20 nS0 HW
n
As a by-product, we note that for KS HW

n , p reduces to the Biot’s modulus Q for saturated
porous media [43], i.e.
1 1 nS n0
¼ ¼ 0þ : ð78Þ
p Q KS KW
Substituting (76) in (73) and taking the isotropic part of the resultant equation, we obtain
1
ðI : rÞ ¼ KD þ pa2B ðI : EÞ: ð79Þ
3
Hence, the undrained bulk modulus is
KU ¼ KD þ pa2B : ð80Þ
Again, take the isotropic part of (73) and eliminate the volumetric strain with incorporating
(76). It follows that

1 KD
ðI : rÞ ¼ aB þ pW : ð81Þ
3 paB
Eq. (81) implies that the pore pressure coefficient B is given by

1
KD paB
B¼ aB þ ¼ ; ð82Þ
paB KU
that is,
paB ¼ BKU : ð83Þ
Substituting (83) in (80), we obtain the Biot’s coefficient

1 KD
aB ¼ 1 : ð84Þ
B KU
Suppose that KD , KU , and B have been determined by the above tests, p and aB are evalu-
ated through (83) and (84), respectively. Then Eqs. (74), (75) and (77) are solved for kSqe , kS , and
HWn . It is worthy to note that for some porous media such as the fully saturated geomaterials,
parameter HW n could be very small, since the solid component has very low compressibility and
the impenetrability on the interface requires that pW pS ¼ 0. In this case, setting HWn ¼ 0, and
S W
we obtain kqe and kS by solving (74) and (75). In general, Hn is a function of the fluid content
and it must be evaluated by the tests on the unsaturated samples with known degree of satu-
ration.
To evaluate HW N
n and Hn , two more tests need to be performed. One is to determine the change
in the capillary pressure (i.e. matric suction [41]) of the unsaturated sample with varying specific
moisture content (i.e. nW ). In this test, the volume of the sample is kept constant. Hence,
nW þ nN ¼ n 0, where n is the porosity change of the sample. It follows from (64) that
N N W W W
W W N N
W
pN pW ¼ nN
0 Hn n n0 Hn n ¼ n0 Hn þ n0 Hn n ; ð85Þ
where use has been made of kfqn ¼ 0. The specific saturation capacity CC is defined by
N

W W 1
CC ¼ nN
0 Hn þ n0 Hn : ð86Þ
Clearly, CC can be measured through the above test.

For the unsaturated porous medium, we can easily prove that the volumetric strain is evaluated
by the following equation
ðr : I=3 þ pN Þ ðpN pW Þ

I : E ¼ þ ; ð87Þ
KM H
where H is the bulk modulus associated with capillary pressure; KM is the constant-suction bulk
modulus and
2 2 3
2 kSqe nW N W N
6 0 n 0 Hn Hn 7
KM ¼ nS0 4kS þ lS þ N W W
5: ð88Þ
3 KS nN
0 Hn þ n0 Hn
In addition, we have
N
nS0 nN
0 Hn
vW ¼ H 1 KM ¼ nW
0 þ W N N
; ð89Þ
nW
0 Hn þ n0 Hn
where vW is the Bishop’s effective stress coefficient [42]. Both KM and H can be experimentally
determined ([41]: ch. 12). Note that the last term in the bracket of (88) is generally much smaller
than the previous terms. Hence, we can approximately write
2 2 3
6 2 kSqe 7
KM ¼ nS0 4kS þ lS 5: ð90Þ
3 KS
For the other test to evaluate HW N

n and Hn , the total confining pressure and air pressure are kept
N
constant, i.e. r : I=3 þ p ¼ 0, and the volumetric strain of the sample is measured with varying
moisture content. This test is performed to determine parameter H. Finally, Eqs. (86) and (89) can
be solved for HW N
n and Hn .
Thus far, all the elastic parameters have been evaluated. Other procedures to determine the
material constants can also be used. The procedure proposed above, however, seems to be the
simplest. It includes five tests: three on the saturated samples and two on the unsaturated samples
with known moisture content. Compared to [43], where only the saturated porous media were
considered, the procedure proposed above needs only one additional test.
7. Acoustic waves in unsaturated rocks
In this section, as an application of the proposed theory, the linear model represented by (67)–
(69) is used to analyze the propagation conditions and characteristics of acoustic waves in an
unsaturated rock. Let r ua ¼ ea , and r ua ¼ Xa . The dispersion relationships for elastic waves
can be obtained by applying the divergence and curl operators, respectively, to both sides of (67)–
(69) and assuming
ea ¼ Aa exp½iðfn x xtÞ
ð91Þ
and
Xa ¼ Ba exp½iðfn x xtÞ
; ð92Þ
pffiffiffiffiffiffiffi
where f is the wave number; x is the angular frequency; i ¼ 1. In general, f is a complex
number, i.e. f ¼ fr þ ifi , where fr and fi are real numbers. fi is the attenuation coefficient. The
phase velocity is defined as v ¼ x=fr .
The conditions of existence of nontrivial solutions (91) and (92) give rise to the following
dispersion relationships for compressional waves and the shear wave

det x2~z þ f2 j~ ixd~ ¼ 0 ðcompressional wavesÞ ð93Þ
and

det x2~z þ f2 ~
h ixd~ ¼ 0 ðshear waveÞ; ð94Þ
where

~z ¼ diag nS0 qS ; nW W N N
0 q ; n0 q ;
2 3
MSS þ 2nS0 lS MSW MSN
~¼4
j MSW MWW MWN 5;
MSN MWN MNN
2 3
l^W þ l^N ^ lW ^lN
6 7
d~ ¼ 4 ^ lW l^N 0 5;
^lN 0 l^N
and
2 3
nS0 lS 0 0
h¼4 0
~ 0 0 5:
0 0 0
Given an angular frequency x, (93) and (94) can be solved for f, and the phase velocity and the
attenuation coefficient are evaluated. For a given angular frequency x, the polynomial equation
expanded from (93) has three complex roots for ðx=fÞ2 and hence the wave number f has six
roots. However, only three of these roots physically make sense, since the amplitude of the waves
must decrease with time and the imaginary part of f (i.e. fi ) is always nonnegative. Therefore,
there generally exist three compressional waves in an unsaturated porous medium, respectively
denoted as P1, P2, and P3 such that vP1 > vP2 > vP3 .
Before solving (93) and (94), it is instructive to examine these two equations in the limit cases.
As x ! 0 so that f ! 0 and x=f becomes finite, it follows from (93) that
h i
4 2
ðx=fÞ 1 vc0 ðf=xÞ ¼ 0;
where
MSS þ 2nS0 lS þ MWW þ MNN þ 2ðMSW þ MWN þ MSN Þ

v2c0 ¼ : ð95Þ
nS0 qS þ nW W N N
0 q þ n0 q
Physically, this situation corresponds to those wherein all the three components move together,
i.e. l^W ! þ1 and l^N ! þ1. As x ! þ1 such that f ! þ1 and x=f remains finite, (93) de-
scribes purely elastic waves, equivalent to the case with no viscous coupling, i.e. l^W ! 0 and
l^N ! 0.
Similarly, as x ! 0 and x=f becomes finite, (94) gives
nS0 lS
v2s ¼ v2s0 ¼ ; ð96Þ
nS0 qS þ nW W N N
0 q þ n0 q
where vs0 is the velocity of the shear wave when all the three components move together. As
2
x ! þ1 and x=f remains finite, Eq. (94) degenerate to nS0 qS ðx=fÞ ¼ nS0 lS . Clearly, this corre-
sponds to the case wherein viscous coupling vanishes.
Table 1
Material properties for the Massilon sandstone saturated by water and air
Material parameters Symbol Value Unit
Porosity n0 0.23 –
Density of solid grain qS 2650.0 kg/m3
Density of water qW 997.0 kg/m3
Density of gas qN 1.10 kg/m3
Bulk modulus of solid grain KS 3:5 107 kPa
Bulk modulus of water KW 2:25 106 kPa
Bulk modulus of air KN 0:11 103 kPa
Viscosity of water mW 1:0 103 Pa s
Viscosity of air mN 1:8 105 Pa s
Intrinsic permeability k 2:5 1012 m2
Elastic constants kS 2:94 106 kPa
lS 1:87 106 kPa
kSqe 1:0 107 kPa
The material properties of the Massilon sandstone are summarized in Table 1. These properties
were obtained based on Murphy’s data [44]. To evaluate HW N
n and Hn , the Brooks–Corey rela-
tionship [45] given below is used
SM ðSr Þ ¼ pd Se1=k ; ð97Þ
where the matric suction SM , i.e. capillary pressure, is a function of Sr and

8
< 0:0 Sr 6 SrW ;
Sr SrW
Se ¼ SrW < Sr < SrN ;
: SrN SrW
1:0 Sr P SrN ;
pd is the air-entry pressure value, k a positive constant; SrW and SrN are the residual and air-entry
degree of saturation, respectively. For sandstone, we choose pd ¼ 50 kPa, k ¼ 1:5, SrW ¼ 0:1, and
SrN ¼ 0:95. Eq. (97) is used to evaluate Cc . The bulk modulus H is not available for the Massilon
sandstone. However, the parameter vW in (89) is generally located in the vicinity of Sr for un-
saturated porous media [12,14]. For convenience, we assume vW ¼ Sr .
Fig. 2 illustrates the influence of the degree of saturation on the phase velocities of waves for
four different frequencies, i.e. 500, 1000, 1500, and 2000 Hz. In Fig. 2a and b, the measured phase
velocities of the P1 wave and the shear wave S are also given. It can be seen that the phase ve-
locities of P1 and S waves do not change with the frequency in the range of 500–2000 Hz, and the
model predictions agree favorably with the experimental results. Fig. 2a shows that the phase
velocity of S-wave decreases gently with increasing moisture content. A similar trend is observed
for the first compressional wave P1 (Fig. 2b). Unlike the former, however, vP1 increases when
Sr > SrN ð¼ 0:95Þ. Two mechanisms can be used to explain this phenomena: on one hand, vP1
decreases with increase in the density of the mixture; on the other hand, vP1 increases with decrease
in the compressibility of the mixture. The compressibility of the mixture is minimal when it be-
comes fully saturated and hence vP1 reaches its maximum value at Sr ¼ 1:0.
Fig. 2. Variations of phase velocities with varying degree of saturation for the Massilon sandstone.
Fig. 2c shows that vP2 decreases with the degree of saturation up to 85% and then increases
rapidly when the porous material approaches the fully saturated condition. It can be seen that vP2
is significantly influenced by the frequency. Fig. 2d shows that vP3 increases with Sr up to about
75% and then decreases. When Sr < SrW or Sr > SrN , vP3 vanishes and P3 disappears. This implies
that the third compressional wave coexists with the capillary pressures and when the capillary
pressure becomes zero, P3 disappears.
Fig. 3 illustrates the effects of the degree of saturation and the frequency on the attenuation
coefficient (i.e. fi ). As expected, in all the cases, the waves of higher frequency always attenuate
more rapidly. For S-wave (Fig. 3a) and P1-wave (Fig. 3b), the attenuation is trivial for the fre-
quency lower that 500 Hz. When Sr > 50%, S and P1 are attenuated rapidly with the degree of
saturation and the attenuation reaches a peak value at around Sr ¼ SrN , where the air phase is
trapped in the water. The attenuation of P2 is described in Fig. 3c, showing that there is a peak
value when saturation is in between 0.8 and 0.9. As shown in Fig. 3d, P3 attenuates strongly when
one of the fluid phases tends to disconnect and the attenuation of P3 has smaller values in between
SrW and SrN . This again implies that there exists an affinity between P3 and the capillary pressure.
In general, P2 and P3 waves are elusive in measurements since they are strongly attenuated in
propagation.
Fig. 3. Variations of attenuation coefficients with varying degree of saturation for the Massilon sandstone.
8. Conclusions
A continuum theory of porous media saturated by multiple fluids is developed within the
framework of the theory of mixtures. The proposed theory can be used to characterize the dy-
namic compatibility conditions on interfaces. These compatibility conditions microscopically
represent the mechanical constraints on the pressure jumps across interfaces and play important
roles in the overall behavior of porous media with multiple fluids. Volume fractions are intro-
duced as independent state variables to represent the microstructure of porous media. It is shown
that the capillary relaxation is thermodynamically associated with the changes in the volume
fractions of fluids.
A linear elastic model of porous media with multiple fluids is developed by a formal lineari-
zation of the proposed theory. For the fully saturated porous media, the linearized theory reduces
to the Biot’s poroelasticity model. A simplified model is developed for porous media saturated by
two-phase fluids and a procedure to obtain the material constants is presented.
As an application and validation of the proposed theory, propagation of acoustic waves in po-
rous media with two fluids is analyzed by using the linear model. It is shown that there generally
exist three compressional waves in the unsaturated porous media, and the third compressional
wave is associated with the capillary pressure. The predicted velocities of the first compressional
and shear waves are compared with the measurements and a favorable comparison is obtained.
Acknowledgements
This research was supported by the US National Science Foundation (grant no. CMS-9501718)
and this support is acknowledged.
Appendix A. Coefficients in the governing equations (67)–(69)

2
nS0 KS kSqe ðDW þ DN 2DW DN Þ
MSS ¼ nS0 kS þ KS 2kSqe ;
KS ð1 DN DW Þ

nS0 kSqe KS nW k W
0 qn K W ð1 DN ÞDW
MSW ¼ ;
KS ð1 DN DW Þ

nS0 kSqe KS nN N
0 kqn KN ð1 DW ÞDN
MSN ¼ ;
KS ð1 DN DW Þ
2
W
nS0 nW k
0 qn KW DW
MWW ¼ nW
0 KW ;
KS ð1 DN DW Þ

W N N
nS0 nW k
0 qn KW n k
0 qn K N DW DN
MWN ¼ ;
KS ð1 DN DW Þ
2
N
nS0 nN
0 kqn KN DN
MNN ¼ nN
0 KN ;
KS ð1 DN DW Þ
where
nW
0 KS
DW ¼ ;
W W W
nS0 KW þ nW
0 K S þ nS W
n
0 0 n 0 H n 2kqn
nN
0 KS
DN ¼ ;
N N N
nS0 KN þ nN
0 K S þ nS N
n
0 0 n0 H n 2k qn
EW ¼ kSqh þ nW W W
0 khn kqh ;
and
EN ¼ kSqh þ nN N N
0 khn kqh :
Appendix B. Coefficients in the governing equations (70) and (71)

2
n0 nS0 KS kSqe
A ¼ nS0 kS þ KS 2kSqe ;
W
n0 KS þ nS0 KW þ n0 nS0 n0 HW n 2kqn
N ¼ nS0 lS ;

n0 nS0 kSqe KS n0 kWqn K W
Q¼ ;
W
n0 KS þ nS0 KW þ n0 nS0 n0 HWn 2k qn
and
2
n20 KS KW nS0 kW
qn þ n 0 K W HW
n
R¼ :
W
n0 KS þ nS0 KW þ n0 nS0 n0 HW
n 2kqn
References
[1] M.A. Biot, J. Appl. Phys. 12 (1941) 155–164.

[2] M.A. Biot, J. Appl. Phys. 33 (1962) 1482–1498.
[3] M.A. Biot, J. Acoust. Soc. Am. 28 (1956) 168–178.
[4] M.A. Biot, Ind. Univ. Math. J. 21 (1972) 597–620.
[5] J.H. Prevost, Int. J. Eng. Sci. 18 (1980) 787–800.
[6] O.C. Zienkiewicz, C. Humpheson, R.W. Lewis, A unified approach to soil mechanics problems (including
plasticity and viscoplasticity), in: G. Gudehus (Ed.), Finite Element in Geomechanics, Wiley, New York, 1977, pp.
151–178.
[7] K. Terzaghi, The shear resistance of saturated soils, in: Proc. 1st Int. Conf. Soil Mech. Found. Eng., Cambridge,
MA, vol. 1, 1936, pp. 54–56.
[8] E.C. Aifantis, Acta Mechanica 37 (1980) 265–296.
[9] O. Coussy, Euro. J. Mech. A/Solid 8 (1989) 1–14.
[10] M.A. Murad, Int. J. Numer. Anal. Mech. Geomech. 23 (1999) 673–696.
[11] L.S. Bennethum, M.A. Murad, J.H. Cushman, Comp. Geotech. 20 (1997) 245–266.
[12] G.D. Aitchison, I.B. Donald, Some preliminary studies of unsaturated soils, in: Proc. 2nd Austr.-N. Zealand Conf.
Soil Mech. Found. Eng., Christchurch, 1956, pp. 192–199.
[13] A.W. Bishop, I.B. Donald, The experimental study of partly saturated soils in triaxial apparatus, in: Proc. 5th Int.
Conf. Soil Mech. Found. Eng., vol. 1, 1961, pp. 13–21.
[14] A.W. Bishop, G.E. Blight, Geotechnique 13 (3) (1963) 177–197.

[15] M.A. Biot, Int. J. Solids Struct. 13 (1977) 579–597.
[16] O. Coussy, Mechanics of Porous Continua, Wiley, New York, 1995.
[17] A. Bedford, Drumheller, Int. J. Eng. Sci. 21 (1983) 863–960.
[18] R. de Boer, Appl. Mech. Rev. 49 (4) (1996) 201–262.
[19] L.M. Morland, J. Geophys. Res. 77 (5) (1972) 890–900.
[20] R. de Boer, J. Bluhm, Int. J. Solid Struct. 36 (1999) 4805–4819.
[21] R.M. Bowen, Int. J. Eng. Sci. 20 (6) (1982) 253–268.
[22] B. Svendsen, K. Hutter, Int. J. Eng. Sci. 33 (14) (1995) 2021–2054.
[23] M.A. Goodman, S.C. Gowin, Arch. Rat. Mech. Anal. 48 (1972) 249–266.
[24] S.L. Passman, Int. J. Eng. Sci. 15 (1977) 117–129.
[25] K. Wilmanski, Arch. Mech. 48 (4) (1996) 591–628.
[26] K. Wilmanski, J. Nonequilib. Thermodyn. 20 (1995) 50–77.
[27] S.M. Hassanizadeh, W.G. Gray, Adv. Water Resour. 13 (4) (1990) 169–186.
[28] K.K. Muraleetharan, C. Wei, Int. J. Numer. Anal. Meth. Geomech. 23 (1999) 1579–1608.
[29] C. Wei, K.K. Muraleetharan, Int. J. Eng. Sci. 40 (2002) 1835–1854.
[30] W.G. Gray, Int. J. Multiphase Flow 26 (3) (2000) 467–501.
[31] Y.A. Buyevich, Transp. Porous Media 21 (1995) 145–162.
[32] S.M. Hassanizadeh, W.G. Gray, Adv. Water Resour. 2 (1979) 131–203.
[33] W.G. Gray, S.M. Hassanizadeh, Int. J. Multiphase Flow 15 (1) (1989) 81–95.
[34] R.W. Ogden, Non-linear Elastic Deformation, Ellis Harwood Ltd, New York, 1984.
[35] C. Wei, Static and dynamic behavior of multiphase porous media: governing equations and finite element
implementation, Ph.D. Thesis, The University of Oklahoma, 2001.
[36] S.R. Groot, P. Mazur, Nonequilibrium Thermodynamics, McGraw-Hill, New York, 1962.
[37] W. Brutsaert, J.N. Luthin, J. Geophys. Res. 69 (4) (1964) 643–652.
[38] S.K. Garg, A.H. Nayfeh, J. Appl. Phys. 60 (9) (1986) 3045–3056.
[39] K. Tuncay, M.Y. Corapcioglu, J. Geophys. Res. 101 (b11) (1996) 25149–25159.
[40] J. Bear, Dynamics of Fluids in Porous Media, American Elsevier, New York, 1972.
[41] D.G. Fredlund, H. Rahardjo, Soil Mechanics for Unsaturated Soils, John Wiley and Sons, New York, 1993.
[42] A.W. Bishop, Tek. Ukebl. 106 (39) (1959) 113–143.
[43] M.A. Biot, D.G. Willis, J. Appl. Mech. ASME 24 (1957) 594–601.
[44] W.F. Murphy, J. Acoust. Soc. Am. 71 (1982) 1458–1468.
[45] H. Brooks, A.T. Corey, Hydraulic properties of porous media, Colorado State Univ. Hydrol. Paper, no. 3, March
1964.

A Continuum Theory of Porous Media Saturated by Multiple Immiscible Uids: I. Linear Poroelasticity

Uploaded by

Copyright:

Available Formats

A Continuum Theory of Porous Media Saturated by Multiple Immiscible Uids: I. Linear Poroelasticity

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A Continuum Theory of Porous Media Saturated by Multiple Immiscible Uids: I. Linear Poroelasticity

Uploaded by

Copyright:

Available Formats

International Journal of Engineering Science 40 (2002) 1807–1833

A continuum theory of porous media saturated by multiple

2. Dynamic compatibility conditions on interfaces

Fig. 1. Microscopic structure of porous media with two-phase ﬂuids.

3. Kinematics and balance equations

vat : Ba 7!St  R3 ; i:e: x ¼ vat ðX a Þ ¼ va ðX a ; tÞ; ð1Þ

where St  R3 is the current conﬁguration, and t 2 ½0; T

spatial position of pa at time t. Function va ðX a ; tÞ is assumed to be invertible and diﬀerentiable as

The deformation gradient of the solid skeleton F is deﬁned by

4. General constitutive relationships

AS ¼ A^S ðh; qS ; EÞ ð17Þ

where zc or zd is an element in fn_ f ; wf ; rhg, f ¼ 1; 2; . . . ; k. Eq. (28) yields

It is important to note that the capillary relaxation represented by P b f is thermodynamically

5. Linearization of the theory

From now onwards, h, nf , qa , E, and wf are all viewed as incremental quantities.

J Sqe ¼ kSqe I; ð47Þ

J She ¼ kShe I; ð48Þ

The speciﬁc entropy of a bulk component is obtained by linearizing (19), i.e.

By deﬁnition, the speciﬁc heat capacity of an individual component is given by cav ¼ h0 ca .

nS0 oqS onS

nf0 oqf onf

The linear form of the above total energy balance equation is

where gf0 is the initial entropy density of a ﬂuid; q^ is given by (39).

6. Linear poroelasticity of unsaturated porous media

where use of l^ff ¼ l^f I has been made.

kS ; lS ; kSqe ; Ka ; kfqn ; Hfn ; l^f ;

where a ¼ S; W; N; f ¼ W; N. The bulk modulus Ka of an individual component can be deter-

where aB is the Biot’s coeﬃcient [43]

pW ¼ paB ðI : EÞ; ð76Þ

As a by-product, we note that for KS  HW

Hence, the undrained bulk modulus is

Eq. (81) implies that the pore pressure coeﬃcient B is given by

paB ¼ BKU : ð83Þ

Substituting (83) in (80), we obtain the Biot’s coeﬃcient

Clearly, CC can be measured through the above test.

ðr : I=3 þ pN Þ ðpN pW Þ

For the other test to evaluate HW N

7. Acoustic waves in unsaturated rocks

MSS þ 2nS0 lS þ MWW þ MNN þ 2ðMSW þ MWN þ MSN Þ

SM ðSr Þ ¼ pd Se1=k ; ð97Þ

where the matric suction SM , i.e. capillary pressure, is a function of Sr and

Appendix A. Coeﬃcients in the governing equations (67)–(69)

Appendix B. Coeﬃcients in the governing equations (70) and (71)

[1] M.A. Biot, J. Appl. Phys. 12 (1941) 155–164.

[14] A.W. Bishop, G.E. Blight, Geotechnique 13 (3) (1963) 177–197.

You might also like

vat : Ba 7!St R3 ; i:e: x ¼ vat ðX a Þ ¼ va ðX a ; tÞ; ð1Þ

where St R3 is the current conﬁguration, and t 2 ½0; T

As a by-product, we note that for KS HW