Vascoand Minkoff

On the propagation of a disturbance in a heterogeneous, deformable,
porous medium saturated with two fluid phases

D. W. Vasco
1
and Susan E. Minkoff
2
ABSTRACT
The coupled modeling of the flow of two immiscible fluid
phases in a heterogeneous, elastic, porous material is formu-
lated in a manner analogous to that for a single fluid phase.
An asymptotic technique, valid when the heterogeneity is
smoothly varying, is used to derive equations for the phase
velocities of the various modes of propagation. A cubic
equation is associated with the phase velocities of the
longitudinal modes. The coefficients of the cubic equation
are expressed in terms of sums of the determinants of 3 3
matrices whose elements are the parameters found in the
governing equations. In addition to the three longitudinal
modes, there is a transverse mode of propagation, a general-
ization of the elastic shear wave. Estimates of the phase
velocities for a homogeneous medium, based upon the for-
mulas in this paper, agree with previous studies. Further-
more, predictions of longitudinal and transverse phase
velocities, made for the Massilon sandstone containing vary-
ing amounts of air and water, are compatible with laboratory
observations.
INTRODUCTION
Multiphase flow is an important physical process that underlies
many critical activities such as waste disposal, geothermal produc-
tion, oil and gas production, agriculture, and groundwater manage-
ment. Geophysical imaging methods are used increasingly to
monitor the flow of liquids and gases in the subsurface (Calvert,
2005; Rubin and Hubbard, 2006). Therefore, it is important to have
accurate and efficient techniques for modeling wave propagation in
heterogeneous porous media saturated by one or more fluid phases.
It is particularly helpful to have methods that provide insight into
the various physical factors controlling the propagation of a wave in
a poroelastic medium.
There are several ways to approach the coupled modeling of de-
formation and multiphase fluid flow in a heterogeneous porous
medium, each with its own advantages and limitations. A numerical
method is the most general, and there are several studies based upon
numerical techniques (Noorishad et al., 1992; Rutqvist et al., 2002;
Minkoff et al., 2003, 2004; Dean et al., 2006). Numerical methods
can require significant computer resources CPU time and com-
puter memory. Also, numerical methods have difficulty modeling
the wide range of behaviors in the coupled multiphase problem,
which can include hyperbolic elastic wave propagation as well
as fluid diffusion, involving a broad range of time scales: from milli-
seconds to hours or even days. Numerical methods tailored to seis-
mic frequencies can improve the computational efficiency (Masson
et al., 2006) but still face challenges in treating multiple fluid phases
and 3D problems. Finally, numerical methods do not provide ex-
plicit expressions for observed quantities such as the arrival time
of a propagating disturbance or its amplitude. Analytic methods
can be efficient and can provide explicit expressions for observed
quantities. However, analytic methods are typically limited to rela-
tively simple situations, such as a homogeneous half-space and a
single fluid phase (Levy, 1979; Simon et al., 1984; Gajo and Mon-
giovi, 1995). As the medium is generalized for example, by
including layering analytic methods become increasingly com-
plicated and require significantly more computation time, facing the
same limitations as numerical techniques (Wang and Kumpel,
2003). Thus, analytic methods may not provide the generality re-
quired for solving commonly encountered inverse problems. For
example, in many inverse problems, one is interested in determining
smoothly varying heterogeneous properties in a 3D setting.
In this paper, we formulate and validate governing equations for
deformation in a porous medium containing two fluid phases and
present an asymptotic, semianalytic technique for their solution.
The equations, presented below, are generalizations of those for
a medium containing a single fluid, given by Pride et al. (1992)
Manuscript received by the Editor 2 April 2011; revised manuscript received 12 December 2011; published online 9 April 2012.
1
Lawrence Berkeley Laboratory, Earth Sciences Division, Berkeley, California, USA. E-mail: dwvasco@lbl.gov.
2
University of Maryland Baltimore County, Department of Mathematics and Statistics, Baltimore, Maryland, USA. E-mail: sminkoff@umbc.edu.
2012 Society of Exploration Geophysicists. All rights reserved.
L25
GEOPHYSICS, VOL. 77, NO. 3 (MAY-JUNE 2012); P. L25L44, 10 FIGS.
10.1190/GEO2011-0131.1
Downloaded 06 Jun 2012 to 216.63.246.110. Redistribution subject to SEG license or copyright; see Terms of Use at http://segdl.org/
and Pride (2005). The governing equations share many character-
istics with earlier work by Tuncay and Corapcioglu (1996, 1997)
and Lo et al. (2002, 2005). The asymptotic approach used in our
investigation is similar in many respects to the technique applied
by Vasco (2009) in a study of broadband propagation in a deform-
able porous medium containing a single fluid. The asymptotic
approach provides semianalytic expressions for the phase velocity
of a propagating disturbance and methods for the efficient solution
of the governing equations. The asymptotic solution is valid in a
medium with smoothly varying heterogeneity of arbitrary magni-
tude and thus is more general than a purely analytic solution.
METHODOLOGY
In this section, we discuss the equations governing the solid and
fluid displacements in a porous medium containing two fluids. We
also outline an asymptotic methodology that can provide a semi-
analytic solution of the governing equations. As indicated below,
the lowest-order terms in the asymptotic series produce equations
for the phase velocities of the various modes of propagation.
The governing equations
Consider the case in which two fluid phases are present in the
pore space of the solid matrix. The porosity of the material is de-
noted by , and the saturations of fluids one and two are denoted by
S
1
and S
2
, respectively. The two phases are assumed to fill the entire
pore space; hence, the saturations sum to unity: S
1
S
2
1. For
the time interval of interest, the elements of the porous matrix be-
have elastically, but the components of the fluid are described by the
constitutive law for a Newtonian liquid. The fluids are assumed to
behave immiscibly and one fluid can block the flow of the other.
Therefore, each fluid obeys the two-phase version of Darcys law
(de Marsily, 1986), in which the flow velocity of the ith fluid re-
lative to the solid matrix _ w
i
is proportional to the gradient of the
fluid pressure:
_ w
i

k
ri
k
i
P
i
; (1)
where k
ri
S
i
is the relative permeability of the ith phase, k is the
absolute permeability,
i
is the fluid viscosity, and P
i
is the fluid
pressure. The relative permeability k
ri
S
i
is a function of the sa-
turation of ith fluid phase and provides a measure of the ability of
the other fluid phase present in the pore space to block the flow.
An important point is that although we are modeling the propa-
gation of a transient disturbance in the fluid-filled porous solid, dis-
turbance is not to be identified with the continuous flow of the fluid.
Rather, the disturbance is associated with the propagation of a wave
in the poroelastic medium. The wave typically will propagate much
faster than any advancing fluid-saturation front. When modeling the
two-phase fluid flow, there will be two timescales: the scale asso-
ciated with the saturation change and the scale associated with the
propagating elastic and pressure disturbance. We assume that one
can model the actual field history incrementally, modeling any
rapid transient disturbance in pressure and solid displacement using
the equations derived here and modeling the saturation changes
using quasi-static two-phase flow. As in a loosely coupled approach
to modeling deformation and flow (Minkoff et al., 2003, 2004), one
can introduce feedback between the solution to the poroelastic
equations derived here and the reservoir simulator used to model
the long-term saturation changes. This should become clearer after
we introduce the full set of coupled equations for two-phase flow.
The conservation equations
The approach used in this section is a straightforward generaliza-
tion of the method of averaging. This technique, developed for a
single fluid phase by Bear et al. (1984), de la Cruz and Spanos
(1985), and Pride et al. (1992), is generalized to two-phase flow by
Tuncay (1995) and Tuncay and Corapcioglu (1996, 1997). As in the
case of a single fluid, one averages the microscopic conservation
equations for the elastic solid matrix and the Newtonian fluids,
making use of Slatterys (1968, 1981) theorem. The application of
Slatterys theorem to the conservation equations results in govern-
ing equations for the displacements in the solid phase u
s
and in the
fluid phases u
i
, i 1, 2:
s
_ u
s
t

s

s
d
1
d
2
(2)
and
i
_ u
i
t

i

i
d
i
. (3)
The dots over the displacement vectors denote the derivative with
respect to time. Note that the summation convention, summation
over repeated indices, is not used in this paper. In an effort to keep
the presentation compact, we are representing the two equations for
the fluid phases by a single indexed equation (equation 3), allowing
i to take the values 1 and 2 for the respective fluid phases. The index
notation, introduced above, will be implemented in much of this
paper. In equations 2 and 3, the parameter
s
is the volume fraction
of the solid phase, and the parameter
i
is the volume fraction of the
ith fluid phase. Note that the volume fraction of the fluid phase may
be written in terms of the porosity and the fluid saturation as
i
S
i
. The solid and fluid densities are denoted by
s
and
i
,
respectively. The quantities
s
and
i
are the stress tensors asso-
ciated with the solid and fluid phases. Explicit expressions for
the stress tensors, in terms of the solid and fluid displacements,
are given in Appendix A. The vectors d
1
and d
2
, referred to as
the momentum transfer or interaction terms, represent drag forces
from the interaction of the solid and fluids within the porous med-
ium (Pride et al., 1993).
Pride et al. (1992) argue that the drag force can be expressed in
the form
d
i

i
Nf
i
; (4)
where f
i
is the macroscopic applied force vector and N is a dimen-
sionless tensor operator independent of f
i
. For an isotropic medium,
N I; (5)
where, in the most general setting, is an integro-differential op-
erator. According to Pride et al. (1992, 1993), the three primary
macroscopic forces influencing the motion of the fluids are a pres-
sure gradient due to a spatially varying flow field, the relative mo-
tion of the elastic solid frame and the fluid, and fluid body forces.
One can think of as a convolutional operator or, in the frequency
domain, as a term that depends upon the frequency . Thus, N can
L26 Vasco and Minkoff
change the nature of the differential operators in f
i
. As shown by
Pride et al. (1992), by substituting d
i
into the fluid equation for the
ith phase, one arrives at a specific form for d
i
:
d
i

i
f
i

i
i
1
1
_ w
i
t
; (6)
where _ w
i
is the flow velocity of fluid i, given in equation 1. The
flow velocity of the fluid is measured relative to the position of
the solid, given by _ w
i
_ u
i
_ u
s
. The quantity 1
1
, termed
the dynamic tortuosity by Johnson et al. (1987), controls how much
relative fluid flow occurs in response to applied forces. In the case
of simplified pore models, analytic methods can be used to calculate
1
1
explicitly (Johnson et al., 1987; Pride et al., 1993). We
can substitute the expressions for d
i
, equation 6, into the macro-
scopic equations for linear momentum (equations 2 and 3). The
resulting governing equations are
s
_ u
s
t

s

s
D
1
_ w
1
t
D
2
_ w
2
t
; (7)
i
_ u
i
t

i

i
D
i
_ w
i
t
; (8)
where
D
i

i
i
1
1
: (9)
Adding and subtracting
i
i
multiplied by the partial derivative of
_ u
s
with respect to time from the left side of equation 8 produces the
following system of equations in u
s
and w
i
:
s
_ u
s
t
D
1
_ w
1
t
D
2
_ w
2
t

s

s
(10)
i
_ u
s
t

i
i
D
i
_ w
i
t

i

i
; (11)
three vector differential equations for the three unknown vectors u
s
,
w
1
, and w
2
. By combining these three equations with the expres-
sions for the stress tensors (see Appendix A) and appropriate
boundary conditions, we can solve for the displacement of
each phase.
There are advantages to writing the equations in the frequency
domain by applying the Fourier or the Laplace transform (Brace-
well, 1978). One advantage is that the time derivatives reduce to
multiplication of the transformed variables by the frequency . This
removes the time derivatives from the equations, leading to a system
of equations containing only spatial derivatives. Furthermore, the
convolutional operator is converted to multiplication by some func-
tion of the frequency . Applying the Fourier transform to each of
the three equations, we can write equations 10 and 11 in the fre-
quency domain:
s
U
s

1
W
1

2
W
2

s

s
0; (12)
1
U
s

1
W
1

1

1
0; (13)
2
U
s

2
W
2

2

2
0; (14)
where we now write the fluid equations explicitly. In these three
equations, denotes the stress tensor transformed into the fre-
quency domain, and we define
s

s
2
; (15)
i

i
2
; (16)
i

i
i
1
1
2
; (17)
and
i

i
i
1 1
1
2
(18)
for i 1, 2.
To finish the statement of the governing equations, we need ex-
pressions for the divergence of the stress tensors in terms of the
solid and fluid displacements. The specification of the solid and
fluid stress tensors, based upon a reformulation of the work of
Tuncay and Corapcioglu (1997), is given in Appendix A. Fourier
transforming the expressions for the solid and fluid stress tensors
given by equations A-24, A-25, and A-26, we can substitute them
into equations 12, 13, and 14 to arrive at

_
G
m
_
U
s
U
s
T
2
3
U
s
I
__
K
u
U
s
C
s1
W
1
C
s2
W
2

s
U
s

1
W
1

2
W
2
0; (19)
C
1s
U
s
M
11
W
1
M
12
W
2

1
U
s

1
W
1
0;
(20)
C
2s
U
s
M
21
W
1
M
22
W
2

2
U
s

2
W
2
0;
(21)
where the coefficients of these equations are given at the end of
Appendix A. Note that the coefficients depend upon the properties
of the components of the medium, the saturations of the fluids S
1
and S
2
, and the capillary pressure function P
cap
that is determined
by the pressure difference in the two fluids (see equation A-4). Also,
the parameter depends upon the flow properties of the medium
and thus upon the medium permeability. For example, as indicated
in the Applications section, for the model considered by Tuncay and
Corapcioglu (1996), is given by

C
i
C
i

i
i
i
; (22)
where C
i
is given by
Propagation in a heterogeneous porous medium L27
C
1

2
S
1
2
1
kk
r1
; (23)
C
2

2
S
2
2
2
kk
r2
(24)
for Darcy flow.
The set of equations 1921 is of the same general form as the
governing equations for displacements in a porous media saturated
by a single fluid (Pride, 2005; Vasco, 2009). There are three extra
terms (those involving W
2
) in the first two equations (19 and 20)
and one additional equation (21) governing the evolution of the sec-
ond fluid phase.
An asymptotic analysis of the governing equations and
semianalytic expressions for the phase
The three expressions, equations 1921, represent a formidable
set of coupled vector differential equations. Because the coefficients
are functions of the spatial variables and the frequency, a closed-
form, analytic solution is generally not possible. Furthermore,
the equations govern both elastic deformation and diffusive fluid
flow, covering a range of spatial and temporal scales. Thus, even
numerical methods for solving these equations can encounter dif-
ficulties due to the wide range of scales. It is possible to gain some
insight and to develop a semianalytic solution to the coupled equa-
tions by means of an asymptotic expression. The solution will be
valid for a medium in which the heterogeneity is smoothly varying.
In Appendix B, we use the method of multiple scales (Whitham,
1974; Anile et al., 1993) to obtain a set of equations that can be used
to determine the phase of a disturbance propagating in a heteroge-
neous porous medium containing two fluids. The technique has
been applied to a number of problems (Korsunsky, 1997), and a
variant of the technique has been used to rederive the governing
equations for poroelasticity (Burridge and Keller, 1981). The meth-
od of multiple scales was used recently to construct a solution for
coupled deformation and flow in a heterogeneous poroelastic med-
ium saturated by a single fluid (Vasco, 2008, 2009). Furthermore, it
has been applied to nonlinear problems involving fluid flow, such as
flow in a heterogeneous medium with pressure-sensitive properties
(Vasco and Minkoff, 2009) and multiphase fluid flow involving
large saturation changes (Vasco, 2011).
Asymptotic expressions for displacements
There are several ways to motivate an asymptotic treatment of the
governing equations 1921. For example, one might adopt an ex-
pansion in powers of the frequency and consider solutions for
which is large. However, because the coefficients contain com-
plicated expressions in frequency and because of the diffusive and
wavelike behaviors contained in the governing equations, it is best
not to make specific assumptions regarding the frequency. An alter-
native approach is provided by the method of multiple scales, which
is based upon a separation of length scales (Anile et al., 1993;
Korsunsky, 1997). In particular, because we are interested in mod-
eling propagation in a smoothly varying medium, we assume that
the scale length of the heterogeneity is much greater than the scale
length of the disturbance. The scale length of the disturbance, which
we denote by l, is the length over which a field, such as the fluid
pressure or the displacement of the porous matrix, varies from the
background value to the value associated with the disturbance. The
scale length of the heterogeneity is denoted by L, and it is assumed
that L is much larger than l. Thus, the ratio lL is assumed to be
much smaller than one. In the method of multiple scales, an asymp-
totic solution is constructed in terms of the ratio . The first step in
this approach involves transforming the spatial scale from physical
coordinates x to slow coordinates, denoted by X, where
X x: (25)
Representing the solution in terms of the slow coordinates X intro-
duces an implicit dependence on the scale variable . Because the
scale parameter is assumed to be small, we can represent the solu-
tion as a power series in :
U
s
X; ; e
i
n0
n
U
n
s
X; ; (26)
W
i
X; ; e
i
n0
n
W
n
i
X; ; (27)
where the superscript n on U
n
s
and W
n
s
denotes additional terms in
the summation and not exponents. The function x; is referred
to as the local phase and is related to the kinematics of the propa-
gating disturbance. As noted by Anile et al. (1993, p. 50), the local
phase is a fast or rapidly varying quantity. Because is small
less than one only the first few terms of the power series are
significant. The series 26 and 27 are in the form of generalized
plane-wave expansions of U
s
X; ; and W
i
X; ; , similar
to that used in modeling electromagnetic and elastic waves (Fried-
lander and Keller, 1955; Luneburg, 1966; Kline and Kay, 1979; Aki
and Richards, 1980; Chapman, 2004).
The coordinate transformation 25 has implications for the spatial
derivatives in the governing equations 1921. For example, using
the chain rule, we can rewrite the partial derivative of the solid
displacement U
s
with respect to the spatial variable x
i
as
U
s
x
i
X
i
x
i
U
s
X
i

x
i
U
s
. (28)
Hence, making use of equation 25,
U
s
x
i

U
s
X
i

x
i
U
(29)
(Anile et al., 1993). Thus, the differential operators, which are
defined in terms of the partial derivatives with respect to the spatial
coordinates, are likewise written as
U
s

X
U
s

U
s
; (30)
where
X
denotes the gradient with respect to the components of
the slow variables X.
To derive an asymptotic solution, we rewrite the differential
operators in the governing equations 1921 in terms of the slow co-
ordinates X. We then substitute the power series representations 26
and 27 for U
s
and W
i
, producing three equations with terms of var-
ious orders in . Because is assumed to be much smaller than one,
we consider terms of the lowest order in . The procedure is outlined
in Appendix B, where terms of order
0
1 are presented. In the
subsections that follow, we discuss these terms in greater detail,
deriving explicit expressions for the phase x; .
Before delving into the details of the expressions for the phase,
we should comment as to what constitutes a smoothly varying med-
ium. As mentioned above, a medium is smoothly varying if the
scale length of the heterogeneity is much larger than the length scale
of the propagating disturbance. However, the length scale of the
disturbance will depend upon its frequency content. Thus, there
is an implicit dependence of the scale length upon frequency
and the smoothness of a medium will depend upon the frequency
under consideration.
Terms of zeroeth-order: The phase of the disturbance
The zeroeth-order terms are presented in Appendix B, equa-
tions B-14 and B-15. These equations can be collected into the
matrix equation
_
_
_
_
I ll I
1
I C
s1
ll I
2
I C
s2
ll I
1
I C
1s
ll I
1
I M
11
ll I M
12
ll I
2
I C
2s
ll I M
21
ll I
2
I M
22
ll I
_
_
_
_
_
_
_
U
0
s
W
0
1
W
0
2
_
_
_
_
_
_
0
0
0
_
_
_; (31)
where l is the local phase gradient,

s
G
m
l
2
; (32)
K
u

1
3
G
m
; (33)
l is the magnitude of the local phase gradient vector l, ll I is a
dyadic formed by the outer product of the vector l (Ben-Menahem
and Singh, 1981; Chapman, 2004) (see equation B-11), and the
coefficients are given in (equations 1518) and in Appendix A.
Alternatively, one may think of the dyadic ll as the vector outer
product ll
T
, where l
T
signifies the transpose of l, converting the
column vector l to the row vector l
T
.
The system of equation 31 has a nontrivial solution if and only if
the determinant of the coefficient matrix vanishes (Noble and
Daniel, 1977, p. 203). For a given set of coefficients, the determi-
nant of the matrix is a polynomial in the components of the vector l.
Given that the components of the vector l are the elements of the
gradient of the phase , the polynomial equation is also a partial
differential equation for the phase function. This nonlinear differ-
ential equation is the eikonal equation associated with propagation
in a porous medium saturated with two fluid phases (Kravtsov and
Orlov, 1990; Chapman, 2004). Although we could attempt to find
the roots of the ninth-order polynomial equation directly, that ap-
proach would involve some rather lengthy algebra. In Appendix C,
we describe an approach based upon the eigenvectors of the system
of equation 31. In that approach, the modes of propagation are
partitioned into longitudinal displacements (displacement in the
direction of l), and transverse displacements (displacement in a
direction perpendicular to l). The results of that approach are
discussed next.
Longitudinal displacements
As shown in Appendix C, for the longitudinal modes of propa-
gation, the condition that equation 31 has a nontrivial solution is
det
_

s
Hl
2
1
C
s1
l
2
2
C
s2
l
2
1
C
1s
l
2
1
M
11
l
2
M
12
l
2
2
C
2s
l
2
M
21
l
2
2
M
22
l
2
_
0; (34)
where we have used the definitions 32 and 33 and defined the para-
meter H as
H K
u

4
3
G
m
: (35)
Equation 34 is a cubic equation in l
2
, the square of the magnitude of
the slowness vector l. Solving this cubic equation for l
2
allows one
to determine the permissible modes of longitudinal displacement.
Equation 34 is much more complicated than the single-phase
constraint, which is the determinant of a 2 2 matrix (Vasco,
2009). Therefore, one must exercise care when calculating the de-
terminant in equation 34. This calculation is given in some detail in
Appendix D, where we apply, in a recursive fashion, a rule for com-
puting the determinant of a matrix whose columns are sums. As
shown in Appendix D, we can write the cubic equation for
s l
2
as
Q
3
s
3
Q
2
s
2
Q
1
s Q
0
0; (36)
where the coefficients are given by
Q
3
det
_
H C
s1
C
s2
C
1s
M
11
M
12
C
2s
M
21
M
22
_
; (37)
Q
2
det
_
_
_
s
C
s1
C
s2
1
M
11
M
12
2
M
21
M
22
_
_
_ det
_
_
_
H
1
C
s2
C
1s

1
M
12
C
2s
0 M
22
_
_
_
det
_
_
_
H C
s1

2
C
1s
M
11
0
C
2s
M
21

2
_
_
_; (38)
Q
1
det
_
_
_
s

1
C
s2
1

1
M
12
2
0 M
22
_
_
_ det
_
_
_
s
C
s1

2
1
M
11
0
2
M
21

2
_
_
_
det
_
_
_
H
1

2
C
1s

1
0
C
2s
0
2
_
_
_; (39)
Q
0
det
_
s

1

2
1

1
0
2
0
2
_
: (40)
The roots of the cubic equation 36 determine the value of l, the
magnitude of the phase gradient vector l . To find the roots,
we first put equation 36 in a canonical form by dividing through
by Q
3
:
s
3
Q
2
Q
3
s
2
Q
1
Q
3
s
Q
0
Q
3
0. (41)
Or, if we define the coefficients
2

Q
2
Q
3
; (42)
1

Q
1
Q
3
; (43)
and
0

Q
0
Q
3
; (44)
we can write equation 41 as
s
3

2
s
2
1
s
0
0: (45)
Note that, in dividing by Q
3
, we are assuming the determinant of the
coefficient array is not zero. The determinant Q
3
vanishes if any of
the rows of the coefficients of the l
2
terms in the determinant 34 are
linear dependent. This can occur if the properties of the two fluids
are similar and it becomes difficult to distinguish between the fluids.
The solution of the cubic equation 45 can be written explicitly as
a function of the coefficients (Stahl, 1997, p. 47). We begin by
defining

1
3
2

1
(46)
and

2
27
1
3
2

0
: (47)
Furthermore, if we define the parameter as

4
27
2

3
; (48)
then we can write the solution of equation 45 in the form
s l
2
2
_
1
1
_
_
3

1
3
2
_
1
1
p
_
3

2
3
;
(49)
an expression for the squared phase gradient magnitude in terms of
the medium parameters. Although the first term in equation 49
shares a formal similarity to the phase of a disturbance propagating
in a porous medium containing a single liquid (Vasco, 2009), the
overall expression is decidedly more complicated.
Because l is the magnitude of the phase gradient vector l ,
we can use equation 49 to formulate a differential equation for :

2
_
1
1
_
_
3

1
3
2
_
1

1
p
_
3

2
3
;
(50)
which is an eikonal equation for the phase of the propagating dis-
turbance (Kravtsov and Orlov, 1990). Equation 50 provides all of
the information necessary for modeling the kinematics, i.e., the tra-
veltime, of the propagating disturbance. For example, one may
solve the nonlinear partial differential equation 50 numerically,
using a fast marching method (Sethian, 1985, 1999), which was
introduced to seismology by Vidale (1988). The fast marching
approach has proven to be stable, even in the presence of rapid
changes in medium properties. Or one may use the method of char-
acteristics (Courant and Hilbert, 1962) to derive a related set of or-
dinary differential equations, the ray equations (Anile et al., 1993;
Chapman, 2004) that can be solved numerically (Press et al., 1992).
Transverse displacements
Now we consider the case in which the displacements are per-
pendicular to the propagation direction. In that situation, the eigen-
vector is given by the solution of an equation similar to equation
C-5 in Appendix C:
e
0; (51)
where
is the associated eigenvalue. Invoking similar arguments

to those used in the analysis in Appendix C but tailored to transverse
displacements, we can show that the vanishing of the determinant of
the matrix reduces to
det
_
_
_
s
G
m
l
2
1

2
1

1
0
2
0
2
_
_
_ 0; (52)
a quadratic equation for l, whose coefficients depend upon the fre-
quency and the properties of the porous medium and the fluids. The
determinant 52 is a straightforward calculation, resulting in the
quadratic equation
2
G
m
l
2

s
2

1
2

2
2
0; (53)
which can be solved for l:
l
s

1
1

1
2
G
m
: (54)
Thus, there is a single solution for phase gradient magnitude asso-
ciated with the transverse mode of displacement. The different signs
indicate propagation in the forward and reverse directions along l.
The expression for l in the case of transverse displacements
(equation 54) is much simpler than that for longitudinal displace-
ments (equation 49). We shall rewrite it to bring out some simi-
larities to the expression for a single fluid (Vasco, 2009). If we
factor out
1
2
and use the definitions for
s
,
1
, and
2
, equation 54
can be written as
l
2
G
m
(55)
or as
l
1
s

f
G
m
; (56)
where
f

1
1
S
1
1

2
2
S
2
2
(57)
is a weighted fluid density whose weights are a function of fre-
quency through the dependence upon . Equation 56 is a direct
modification of the expression for the slowness of an elastic shear
wave. Equation 56 generalizes the expression for wave propagation
in a porous medium saturated by a single fluid phase (Pride, 2005;
Vasco, 2009), where one has
l
f

f
G
m
;
(58)
where is i
f
k for a fluid of density
f
and viscosity
f
.
APPLICATIONS
Comparison with previous studies
Here we compare our results with Tuncay and Corapcioglus
(1996) and Lo et al.s (2005) studies of wave propagation in a
homogeneous porous medium containing two fluid phases and with
experimental data (Murphy, 1982). First, in the case of transverse
displacements, we establish the equivalence of our expression for
the phase velocity to that of Tuncay and Corapcioglu (1996) when
the medium is homogeneous and when we define in a particular
fashion. Second, we compare numerical predictions of complex ve-
locities for the three longitudinal modes of propagation in a porous
medium saturated by two fluid phases. We compare predictions
derived using our formulation with those by Tuncay and
Corapcioglu (1996) and Lo et al. (2005). Finally, we calculate
the primary longitudinal and the transverse velocities for the porous
Massilon sandstone partially saturated by water, as described in
Murphy (1982).
Transverse (shear) displacements
It is simplest to begin our comparisons with the expression for
the phase gradient magnitude of the shear component, given by
equation 54. Before we begin, it must be noted that our phase func-
tion , introduced in the power series 26 and 27, is defined slightly
differently from the conventional use in seismic applications.
Specifically, we include the frequency term as part of . Thus,
our definition of l will contain an additional factor, and the square
of the phase velocity will be given by
c
2

2
l
2
: (59)
Let us begin with an expression for 1l
2
, where l is given by
equation 54:
1
l
2

1
2
G
m
s

1
1

1
2
: (60)
The coefficients are given by expressions 1518. However, the coef-
ficients
1
,
2
,
1
, and
2
contain the operator , which depends
upon the fluid response to applied forces (Johnson et al., 1987;
Pride et al., 1993). To compare our predicted velocities with those
of Tuncay and Corapcioglu (1996), we need to relate these coeffi-
cients to those used in their paper. By comparing the coefficients in
their governing equations 13 with the coefficients in the governing
equations 1214, after accounting for the slightly different formu-
lation and after transforming their equations to the frequency
domain, we find that
1
iC
1
; (61)
2
iC
2
; (62)
1

2

1
iC
1
; (63)
and
2

2

2
iC
2
; (64)
where
i
is the volume averaged density, given by

i

i
i
S
i
i
, and C
1
and C
2
are coefficients defined in
Tuncay and Corapcioglu (1996), related to the fluid flow. The coef-
ficients take the form
C
1

2
S
1
2
1
kk
r1
; (65)
C
2

2
S
2
2
2
kk
r2
(66)
for Darcy flow, where S
i
is the saturation of the ith fluid,
i
is the
fluid viscosity for phase i, k is the absolute permeability, and k
ri
is
the relative permeability for fluid i. Using relationships 6164, we
can rewrite the product as
2

2

2
2
C
1
C
2
C
1

2
C
2

1
i; (67)
as well as the other terms in expression 60. As a result, the square of
the phase velocity c
2
can be expressed as the ratio
c
2

2
l
2

Y
2
Y
1
; (68)
where
Y
1

C
1
C
2

s

1

2

s

1

2
2
i
C
2

1

s

2
C
1

2

s

1
(69)
and
Y
2
G
m
C
1
C
2

1

2
2
iG
m
C
2

1
C
1

2
: (70)
These expressions agree with those of Tuncay and Corapcioglu
(1996) for the phase velocity of the shear wave (see their equa-
tions 2830).
Note that, using equations 17 and 61, one can derive an explicit
expression for the frequency-dependent variable that determines
the dynamic tortuosity 1
1
. Recall that the dynamic tortuosity
controls the amount of fluid flow in response to the applied forces.
The variable also appears in the definition of the coefficients
i
and
i
(see equations 17 and 18), which are part of the
governing equations. Equating the expressions for
i
given in
equations 17, 61, and 62, we have
1
1
_
i
_
C
i
i
: (71)
Solving equation 71 for gives

C
i
C
i

i
i
i
; (72)
where C
i
is given by equations 65 and 66. Equation 72 indicates
that is actually a function of the specific fluid. Thus, in the most
general case, should vary for each fluid. This makes physical
sense because the flow characteristics of each fluid can differ.
Longitudinal displacements
Although it is possible to apply the previous mathematical ana-
lysis to the longitudinal modes, that approach would be signifi-
cantly more complicated. It is far simpler to proceed with a
direct numerical comparison of the predictions provided by the ex-
pressions of Tuncay and Corapcioglu (1996) and Lo et al. (2005)
with those from the solutions of the cubic equation 36. Because the
phase velocities are generated by the cubic equation with complex
coefficients given by expressions 3740, there will be three com-
plex longitudinal velocities in general.
A comparison with Tuncay and Corapcioglu (1996).For the
first comparison, with the results of Tuncay and Corapcioglu
(1996), the poroelastic parameters for the medium are representative
of the properties of the Massilon sandstone described by Murphy
(1982). Thus, the bulk modulus K
fr
of the frame is 1.02 GPa, the
bulk modulus of the grains K
s
is 35.00 GPa, shear modulus G
fr
is
1.44 GPa, the density of the solid grains
s
is 2650.00 kgm
3
, the
intrinsic permeability of the sandstone k is 9.0 10
13
m
2
, and the
volume fraction of the solid phase
s
is 0.77. The two fluid phases
are air (fluid 1) and water (fluid 2), with respective viscosities
1
and
2
of 18 10
6
and 1.0 10
3
Pa-s. For fluid 1 (air), the bulk
modulus K
1
is 0.145 MPa and the density
1
is 1.10 kgm
3
.
For fluid 2 (water), K
2
is 2.25 GPa and
2
is 997.00 kgm
3
.
The capillary function P
cap
used here (see equation A-4) was first
proposed by van Genuchten (1980). The exact form of the capillary
function is
P
cap
S
2

100
__
1
S
2
S
rw
S
m2
S
r2
_
m
_
n
; (73)
where S
r2
is the residual water saturation, taken to be 0.0, and S
m2
is the upper limit of water saturation m 1 1n, where n 10
and 0.025. The relative permeability functions associated with
the two fluids in the porous matrix, those postulated by
(Mualem, 1976; van Genuchten, 1980), are shown in Figure 1.
An examination of the functions C
1
and C
2
, given by equations 65
and 66, reveals that they are singular when the relative permeabil-
ities vanish. Thus, some care is required as the fluid saturations
approach the end points of the curves shown in Figure 1. For this
reason, we shall avoid those saturations at which the relative
permeabilities approach zero.
The phase velocities predicted using expression 59, where l
2
is a
root of the cubic equation 36, are plotted in Figure 2 along with
the phase velocities predicted using the formulas of Tuncay and
Corapcioglu (1996). The phase velocities, the real component of the
complex number c, are associated with a frequency of 1000 Hz. The
cubic equation predicts three complex velocities varying as a func-
tion of fluid saturation. We should note that a
11
in expressions
2123 of Tuncay and Corapcioglu (1996) should be replaced by
the variable a
11
a
11
4G
fr
3 for the three formulas to agree
Figure 1. The relative permeability curves of Mualem (1976) for
fluid flow in a porous medium saturated by two fluids. The curves
describe the variation of the relative permeability k
ir
S
i
, the func-
tion appearing in the expressions 65 and 66 for C
1
and C
2
. The
algebraic expressions for these curves are given by equations 75
and 76. Fluid 1 is the gas phase (air); fluid 2 is the liquid phase
(water).
with their previous equation 18, which contains a
11
. As indicated in
Figure 2, there is excellent agreement between our predicted long-
itudinal velocities and those given by the formulas of Tuncay and
Corapcioglu (1996). The qualitative features noted by Tuncay and
Corapcioglu (1996) are present in the velocities plotted in Figure 2.
For example, the velocities of the first two modes of longitudinal
propagation drop significantly as the water saturation decreases
from one. This is due to the much higher compressibility of air com-
pared with that of water. With further decreases in water saturation,
there is a gradual increase in the phase velocities of the two modes.
As noted by Tuncay and Corapcioglu (1996), the third longitudinal
mode arises due to the pressure difference between the fluid phases.
Thus, this phase velocity approaches zero when the fluid saturations
approach fully saturated or fully unsaturated conditions because the
capillary pressure vanishes when only a single fluid occupies the
pore space.
The imaginary component of the phase velocity c, given by equa-
tion 59, provides a measure of the attenuation. The attenuation var-
ies as expc
i
r, where c
i
is the imaginary component of the phase
velocity and r the radial distance from the source. In Figure 3, we
plot the attenuation coefficient for the three longitudinal modes, cal-
culated using the expressions of Tuncay and Corapcioglu (1996)
and the roots of cubic equation 36. There is excellent agreement
between the two approaches. The attenuation overall is quite small
for the first longitudinal mode of propagation, which propagates
much like an elastic wave in the solid. As noted in Tuncay and
Corapcioglu (1996), the attenuation is due to energy dissipation
induced by the relative motion of the fluids and the solid. Further-
more, they note that the end-point attenuation in the second mode of
longitudinal propagation is determined by the kinematic viscosity,
the ratio of the fluid viscosity to the fluid density. The attenuation
coefficient of the third longitudinal mode of propagation is quite
large and increases as the fraction of either phase becomes large.
Perhaps this is due to the increased flow as the pore space is domi-
nated by a single fluid. Also, the capillary pressure which drives this
mode decreases as one phase begins to vanish, resulting in a rapidly
decreasing amplitude for the third mode of propagation.
A comparison with Lo et al. (2005).Recently, Lo et al. (2002,
2005) have followed an alternative approach in deriving the equa-
tions governing coupled poroelastic deformation and flow. Speci-
fically, they use the mass-balance equations for the three-phase
system (two fluids and one solid) coupled with a closure relation
for porosity change to derive the governing equations. The resulting
system of equations is similar to those given above. In particular,
their stress-strain relations (given by equations 32a32c of Lo et al.
[2005]) are equivalent to our equations A-18, A-20, and A-21. The
coefficients in their stress-strain relations, also denoted by a
ij
(given
by equations 30d30j in Lo et al. [2005]), are identical to those of
Tuncay and Corapcioglu (1996) and to expressions A-7A-17. The
primary differences between the equations presented here and those
of Lo et al. (2005) are because of the inclusion of temperature
effects and inertial coupling terms between the fluids that we do
not include. Lo et al. (2005) note that their equations include inertial
terms due to the differential movement between the solid and the
fluids. They contrast their results with the expressions of Tuncay
and Corapcioglu (1996) that do not contain such inertial terms.
However, our expressions, in particular the coefficients
i
given
Figure 2. The three phase velocities associated with the longitudi-
nal modes of propagation in a porous medium saturated with two
fluids. The phase velocities are plotted as functions of water satura-
tion. The frequency used in the computations was 1000 Hz. The
phase velocities are determined by the real component of the roots
of cubic equation 36.
by equation 18, do contain inertial effects due to the coupling
between the solid and the fluids, in the form of
2
terms.
For a qualitative comparison of our expressions and those of Lo
et al. (2005), we have computed the three longitudinal velocities for
the two simulations described in their paper. The porous solid prop-
erties are based upon experimental data for an unconsolidated Co-
lumbia fine sandy loam (Chen et al., 1999; Lo et al., 2005). The
primary difference between this porous material and the sandstone
described above is that the fine sandy loam is unconsolidated and
hence much weaker. Thus, the bulk modulus of the rock frame K
fr
is only 0.008 GPa, a fraction of the value of 1.02 GPa for the sand-
stone. Similarly, the shear modulus of the frame for the loam G
fr
is
quite low, 0.004 GPa, compared to a value of 1.44 GPa for the con-
solidated sandstone. Note that both materials are primarily com-
posed of silica grains and the bulk modulus of the solid particles
is 35 GPa. Therefore, we would expect quite different bulk veloci-
ties for the consolidated sandstone and the unconsolidated
sandy loam.
Lo et al. (2005) consider two pairs of fluids: an air-water system,
similar to that of Tuncay and Corapcioglu (1996), and an oil-water
mixture. The properties of the constituents in the air-water system
are similar to those used in Tuncay and Corapcioglu (1996): the
bulk modulus of air is 0.145 MPa, the bulk modulus of water is
2.25 GPa, and the densities of air and water are 1.1 and
997.0 kgm
3
, respectively. The viscosity of air is 18 10
6
Pa-s,
and the viscosity of water is 0.001 Pa-s. For the oil, the bulk mod-
ulus is given by 0.57 GPa, the density is given by 762 kgm
3
, and
the viscosity is given by 0.00144 Pa-s. The derivative of the
capillary pressure with respect to changes in saturation is given ex-
plicitly by
dP
cap
dS
1

2
g
mn
_
1 S
1
n
n1
1
_
1n
n
1 S
1
_
2n1
n1
_
(74)
(Lo et al., 2005), where g is the gravitational acceleration. The quan-
tities , m, and n are fitting parameters with values 1 m
1
,
n 2.145, and m 1 1n 0.534 for the air-water system.
For the oil-water system, the parameters are given by
2.39 m
1
, n 2.037, and m 1 1n 0.509 (Lo et al.,
2005). This is the same model of capillary pressure put forward
by van Genuchten (1980) and used above, though in a slightly dif-
ferent formulation (see equation 73).
The relative permeability functions are those of Mualem (1976),
which were used in the previous comparison. The exact algebraic
expressions are
k
r1
S
2
1 S
2
1 S
2
1
m
2m
; (75)
k
r2
S
2
S
2
f1 1 S
2
1
m
m
g
2
; (76)
where is a fitting parameter (Mualem, 1976). The fitting parameter
associated with the air-water and the oil-water mixtures is 0.5, as
noted in Lo et al. (2005).
We computed the three longitudinal velocities for the air-water
system and the oil-water mixture as a function of the water satura-
tion. The velocities of the first longitudinal wave are plotted in
Figure 4 for the air-water and the oil-water fluid mixtures. The
Figure 3. The attenuation of a propagating longitudinal mode of
displacement, plotted as a function of water saturation. The fre-
quency used in the computations was 1000 Hz. The attenuation
is determined by the imaginary components of the three roots of
cubic equation 36.
velocities are computed at a single frequency of 100 Hz for this
phase. As expected, the bulk velocities for the air-water system are
much lower for the unconsolidated sandy loam (on average,
100 ms; Figure 4), than for the sandstone (between 1140 and
1200 ms; Figure 2). In Figure 4, the velocities calculated using
the expressions of Tuncay and Corapcioglu (1996) are indicated
by the dashed (oil-water) and solid (air-water) lines, whereas our
calculated values are indicated by the filled squares (air-water)
and the open circles (oil-water). The behavior of the air-water and
the oil-water systems is rather different as the saturations are varied.
When the water saturation is near zero, the bulk velocities of the two
systems are distinctly different, with the velocity of the oil-water
mixture approximately eight times larger than the velocity of the
air-water mixture. This reflects the influence of the pore fluids
because the systems are saturated by two very different fluids.
The velocity in the oil-water system gradually increases as the water
saturation increases. This contrasts with the behavior of the air-
water system in which the velocity remains nearly constant until
rather high water saturation. The velocity increases dramatically
when the porous medium is almost completely water saturated.
When the material is entirely saturated by water, the velocities
are nearly equal for the air-water and oil-water systems. This makes
physical sense because both systems are in identical water-saturated
states. However, there may be slight differences because, as noted in
Lo et al. (2005), two different permeabilities are used by Chen et al.
(1999) to fit the observational data. Our calculated velocities agree
with those computed using the expressions of Tuncay and Corap-
cioglu (1996). Furthermore, the variations of the first longitudinal
velocity, denoted by P1, agree with those of Lo et al. (2005) (see
their Figure 1a).
The longitudinal mode of intermediate velocity, often referred to
as the P2 mode, is associated with diffusive propagation in the man-
ner of a transient pressure variation (Pride, 2005). The propagating
disturbance corresponds to the out-of-phase motion of the solid and
fluid mixtures (Lo et al., 2005; Pride, 2005). The disturbance is
known as the slow wave in the study of wave propagation in a med-
ium saturated with a single fluid (Pride, 2005; Vasco, 2009). In
Figure 5, we plot the calculated intermediate velocities for the
air-water and oil-water fluid mixtures in the sandy loam at a fre-
quency of 100 Hz. The average velocities are much lower for this
mode of propagation, on the order of 1 2 ms. The computed var-
iations (filled squares and open circles) generally agree with the pre-
dications of Tuncay and Corapcioglu (1996) (solid and dashed
lines). There is some deviation for the oil-water system at high
water saturations. The estimates are of the same order as those of
Lo et al. (2005). However, the exact values differ by a factor of
two or more, and there are differences in the nature of the variation
for the air-water system at high water saturations. As in the case of
the P1 mode, the velocities of the two systems approach each other
as the medium becomes water saturated.
The third mode of propagation P3, with the lowest velocity,
arises from the pressure difference between the two fluid phases
(Tuncay and Corapcioglu, 1996). Thus, the disturbance is the result
of the presence of a second fluid in the pore space and is not ob-
served in systems with a single fluid phase. Such a phase is extre-
mely difficult to observe experimentally (Tuncay and Corapcioglu,
1996) due to its high attenuation and extremely low velocity. In
Figure 6, we plot the calculated velocities of the P3 mode for
the air-water and oil-water mixtures at a frequency of 100 Hz.
The velocities are extremely low, generally less than 0.1 ms. As
noted in other studies (Tuncay and Corapcioglu, 1996; Lo et al.,
2005), the velocities approach zero at high and low water satura-
tions. As in the case of the other two modes, the air-water and
oil-water velocities approach a common value (in this case, zero)
as the water saturation approaches one. There is general agreement
Air-Water
Oil-Water
Figure 4. Velocities for the first (fastest) longitudinal mode, known
as P1, plotted as a function of the water saturation. Two fluid mix-
tures are shown in this figure: an air-water mixture and an oil-water
mixture. The velocities calculated using the expressions in this pa-
per are plotted as open circles for the oil-water system and solid
squares for the air-water system. In addition, the values computed
using the formulas of Tuncay and Corapcioglu (1996) are plotted as
a dashed line (oil-water) and a solid line (air-water).
Oil-Water
Air-Water
Figure 5. The velocities associated with the second or intermediate
P2 mode of propagation, plotted as a function of water saturation.
The values computed using the formulas of Tuncay and Corapcio-
glu (1996) are plotted as a dashed line (oil-water) and a solid line
(air-water).
between our estimates and those of Tuncay and Corapcioglu (1996)
and Lo et al. (2005).
The velocities of the three modes of propagation are frequency
dependent. To compare the variation with frequency we have com-
puted the velocities for three frequencies: 50, 100, and 200 Hz. In all
of the computations, we only consider the air-water system. For the
first longitudinal mode P1, as noted in Lo et al. (2005); (see their
Figure 1a), the velocities do not change over this range of frequen-
cies. Thus, we have not plotted the velocities because they are iden-
tical to those shown in Figure 4. In Figure 7, we plot the velocities
for the intermediate model P2 at the three frequencies of interest. As
in Lo et al. (2005), the velocities increase as the frequencies
increase. A similar pattern of higher velocities with increasing
frequency is observed for the third longitudinal mode of propaga-
tion P3, as shown in Figure 8.
A comparison with laboratory observations
Here, we compare data from a series of experiments by Murphy
(1982) to predictions based upon our formulation. These and other
experimental studies have shown that fluid saturations can have a
significant influence upon the phase velocities of extensional (lon-
gitudinal) and rotational (transverse) waves in a sample (Domenico,
1974, 1976; Murphy, 1982). Although many laboratory experi-
ments, such as those of Domenico (1974, 1976) are conducted at
high frequency, the resonance bar experiments of Murphy (1982)
span a wide frequency range, from 300 Hz to 14 kHz. In addition,
a torsional pendulum technique was used to measure rotational
(transverse) wave attenuation at low acoustic frequencies (Murphy,
1982). The flow experiments of Murphy (1982) were conducted in a
sample of Massilon sandstone. The properties of this porous mate-
rial are identical to those noted above for the Massilon sandstone
(Murphy, 1982; Tuncay and Corapcioglu, 1996).
The relative-permeability curves used to represent the flow of the
two fluids, air and water, in the sandstone are those published by
Wyckoff and Botset (1936) and shown in Figure 9. These curves are
also used in the analysis of the Massilon data conducted by Tuncay
and Corapcioglu (1996). Because these curves have no analytic
representation, we digitized the curves plotted in Tuncay and
Corapcioglu (1996) and interpolated between the points using cubic
splines. Thus, the relative permeabilities are approximate at the high
and low saturation values where it is difficult to resolve the small
Air-Water
Oil-Water
Figure 6. The velocities associated with the third mode of propaga-
tion P3, plotted as a function of water saturation. The values com-
puted using the formulas of Tuncay and Corapcioglu (1996) are
plotted as dashed (oil-water) and solid (air-water) lines.
50 Hz
100 Hz
200 Hz
Figure 7. The velocities associated with the second or intermediate
mode of propagation P2, plotted as a function of water saturation.
The velocities are shown for three different frequencies: 50, 100,
and 200 Hz. As in Figures 5 and 6, the values computed using meth-
ods in this paper are symbols; the values computed using the meth-
ods in Tuncay and Corapcioglu (1996) are solid and dashed lines.
5
0

H
z
1
0
0

H
z
2
0
0

H
z
Figure 8. The velocities associated with the third mode of propaga-
tion P3, plotted as a function of water saturation. The velocities are
shown for three different frequencies: 50, 100, and 200 Hz. The
values computed using methods in this paper are symbols; the
values computed using the methods in Tuncay and Corapcioglu
(1996) are indicated by solid and dashed lines.
values. The capillary pressure function given above (see equa-
tion 73) was used in our modeling. These observations have been
used in studies involving elastic wave propagation in a partially
saturated porous medium (Tuncay and Corapcioglu, 1996;
Berryman et al., 2002).
Using the parameters given above, the relative permeability
curves plotted in Figure 9, and the capillary pressure function
73, we calculated the phase velocities for the longitudinal and trans-
verse modes of propagation. Because of the diffusive nature of the
second and third longitudinal modes, it is difficult to observe them
experimentally. Thus, Murphy (1982) only has observations asso-
ciated with the primary or first longitudinal model and the trans-
verse mode, as shown in Figure 10. We calculated the first
longitudinal mode using cubic equation 36 with the parameters
given above and the coefficients given in Appendix A and
equations 3740. The phase velocity associated with the transverse
mode was estimated using equation 54. For comparison, we also
computed the values using the formulas given in Tuncay and
Corapcioglu (1996), plotted in Figure 10. Because of the difficulties
in estimating the relative permeabilities and the sensitivity of the
coefficients C
1
and C
2
at high and low saturations, we avoided mak-
ing predictions of phase velocities for saturations near zero and one.
Both techniques give sharp increases in longitudinal phase velocity
for water saturations near one, but the exact values are fairly sen-
sitive to how the relative permeabilities are calculated. Overall,
there is good agreement between the observations of longitudinal
and transverse phase velocity, the predictions by Tuncay and
Corapcioglu (1996), and our predictions (Figure 10).
DISCUSSION
Following the approach of Tuncay (Tuncay, 1995; Tuncay and
Corapcioglu, 1996, 1997) but using the formulation of Pride
(1992, 1993), we have obtained governing equations for coupled
deformation and two-phase flow. These equations are similar to cor-
responding expressions for coupled deformation and single-phase
flow (Pride, 2005; Vasco, 2009). This similarity should aid in
the interpretation of the coefficients and terms of the more compli-
cated two-phase equations. Furthermore, the approach should make
the extension to three phase conditions, such as oil, water, and gas,
less difficult. Such an extension will result in a more complicated
quartic equation for the longitudinal velocities. However, the solu-
tion of a quartic equation still has a closed-form expression in terms
of its coefficients (Stahl, 1997, p. 124).
The equations given here unify several earlier investigations of
two-phase flow in a deformable medium (Berryman et al., 1988;
Santos et al., 1990; Tuncay and Corapcioglu, 1996, 1997; Lo
et al., 2005) in which different assumptions were made regarding
the inclusion of capillary pressure, the consideration of inertial
effects of the fluid, and the presence or absence of heterogeneity
in the medium. Here, as in Pride et al. (1992, 1993), we allow
for the effects of capillary pressure as well as inertial effects due
to the relative acceleration of the fluids with respect to the solid
matrix. These effects are contained in the complex, integro-differ-
ential operator , whose form may vary, depending on the various
forces included in the formulation. The methods presented in this
paper are also applicable to more general models of fluid flow. For
example, one could develop a model for wave propagation in a med-
ium with patchy saturation (Dvorkin and Nur, 1998; Johnson,
2001). In addition, one could consider the mechanism of Biot flow
and squirt flow, in which fluid movement into microcracks is
accounted for (Dvorkin et al., 1994).
CONCLUSIONS
The asymptotic analysis presented in this paper leads to a semi-
analytic solution for a medium with smoothly varying properties.
Our preliminary analysis, restricted to zeroeth-order terms, provides
explicit expressions for the slowness of the longitudinal and trans-
verse modes of propagation. As in a homogeneous medium, a cubic
equation determines the slowness for the three modes of longitudi-
nal propagation. The coefficients of the cubic equation are
Figure 9. Relative permeability curves based upon the experiments
of Wyckoff and Botset (1936) on the flow of gas-liquid mixtures
through unconsolidated sands. The relative permeability functions
are based upon cubic spline fits to a set of digitized points.
Transverse Mode
First Longitudinal Mode
Figure 10. Observed (squares) and calculated (circles, crosses)
phase velocities of the phase velocity of the transverse mode
(lower curve) and the first longitudinal model (upper curve). The
observed values were obtained from the experiments of
Murphy (1982).
expressed as sums of determinants of 3 3 matrices. The elements
of the matrices are the coefficients in the governing equations.
These determinant-based formulas for the coefficients are much
simpler than previous explicit forms, are easy to implement in a
computer program, and should reduce the occurrence of algebraic
errors. Most importantly, the results are valid in the presence of
smoothly varying heterogeneity. Thus, the explicit expressions
for slowness provide a basis for traveltime calculations and ray tra-
cing in a heterogeneous poroelastic medium containing two fluid
phases.
The asymptotic results pertaining to the phase velocity are the
first steps toward a full solution of the coupled equations governing
deformation and two-phase flow. Following earlier work on single-
phase flow, it is straightforward, though rather laborious, to derive
an expression for the amplitudes of the disturbances. Thus, one can
derive the zeroeth-order solution, obtained by considering the terms
in the asymptotic power series corresponding to n 0. The solution
is valid for a medium with smoothly varying heterogeneity.
However, the exact definition of smoothness is with respect to the
scale length of the propagating disturbance. Thus, the notion of the
medium smoothness does depend upon the frequency range of in-
terest. If layering is present, it can be included as explicit boundaries
within a given model, as can fault boundaries. The full expression
for the zeroeth-order asymptotic solution may be used for the effi-
cient forward modeling of deformation and flow. Such modeling
encompasses the hyperbolic, wavelike propagation of the elastic
compressional and shear waves and the diffusive propagation that
occurs primarily due to the presence of the fluid.
ACKNOWLEDGMENTS
This work was supported by the Assistant Secretary, Office of
Basic Energy Sciences of the U. S. Department of Energy under
contract DE-AC03-76SF00098. We would like to thank Jim Berry-
man, and by extension William Murphy, for supplying the data from
the Massilon sandstone experiments.
APPENDIX A
THE CONSTITUTIVE EQUATIONS
In this appendix, we discuss the stress-strain relationships used in
this paper. These equations have a long history and have evolved
from the early constitutive equations for an elastic solid. First, the
equations of elasticity were generalized to include a fluid (Kosten
and Zwikker, 1941; Frenkel, 1944; Biot, 1956a, 1956b, 1962a,
1962b; Garg, 1971; Auriault, 1980; Pride et al., 1992). Then,
two immiscible fluids were allowed to occupy the pore space (Bear
et al., 1984; Garg and Nayfeh, 1986; Berryman et al., 1988; Santos
et al., 1990; Tuncay and Corapcioglu, 1997; Lo et al., 2002, 2005)
The stress-strain relationships depend on the elastic properties of the
solid matrix and on the properties of the two fluids contained within
the pore space. Specifically, the constitutive relationships depend
upon the bulk modulus of the solid material comprising the grains
of the matrix K
s
and the bulk modulus of the solid skeleton as a
whole, or the bulk modulus of the frame K
fr
. In addition, the
stress-strain relationship depends upon the shear modulus of the
solid grains G
s
and upon the shear modulus of the frame G
fr
.
The mechanical behavior of the poroelastic fluid-saturated body
also depends upon the bulk moduli of the fluids, as denoted by
K
1
and K
2
. The behavior of the fluid-filled porous body is also
a function of the pore fraction, as represented by the porosity
and the fluid phase saturations S
1
and S
2
. The quantity
i
is the
volume fraction of the fluid phase i and is related to the fluid
saturation according to
i
S
i
: (A-1)
Note that the fluid saturations sum to one, S
1
S
2
1, because
they fill the entire pore space. As is well known in the theory of
the flow of immiscible fluids, in general there is a pressure differ-
ential between the two fluids occupying the pore space, the capillary
pressure P
cap
P
1
P
2
(Bear, 1972). This pressure differential,
which is a function of the saturations, is responsible for the curva-
ture of the interface between the pore fluids. Because the fluid
saturations sum to unity, we can write the capillary pressure as a
function of one of the fluid saturations say, S
1
. Because we
will be considering incremental pressures and saturations, changes
with respect to some background average pressures and saturations,
we can linearize the relationship between the incremental saturation
change and the incremental pressure differences. Thus, we can write
P
1
P
2

dP
cap
dS
1
S
1
; (A-2)
assuming that one considers a small enough time increment such
that the saturation change S
1
is small.
The macroscopic stress-strain equations were derived by Tuncay
and Corapcioglu (1997) using the method of averaging. This work
generalizes the single-phase analysis of Pride et al. (1992). The
coefficients in the equations are written in terms of the properties
of the porous skeleton and the fluids:
N
1
K
s
1 K
fr
; (A-3)
N
2
S
1
S
2
dP
cap
dS
1
; (A-4)
N
3
N
1
K
1
S
1
N
2
K
2
S
2
N
2
K
1
K
2
K
2
s
K
1
S
2
K
2
S
1
N
2
: (A-5)
In terms of these coefficients, the stress-strain relationship for the
solid phase is given by
1
s
a
11
u
s
a
12
u
1
a
13
u
2
I
G
fr
_
u
s
u
s
2
3
u
s
I
_
; (A-6)
where
a
11

K
s
N
1
1 K
1
N
2
S
1
K
2
N
2
S
2
K
1
K
2
N
3
K
2
s
K
fr
K
1
S
2
K
2
S
1
N
2
N
3
; (A-7)
a
12

K
1
K
s
N
1
S
1
K
2
N
2
N
3
; (A-8)
a
13

K
2
K
s
N
1
S
2
K
1
N
2
N
3
. (A-9)
Similarly, the full stress-strain relations for the two fluid compo-
nents are
S
1
1
a
21
u
s
a
22
u
1
a
23
u
2
I; (A-10)
where
a
21
a
12
; (A-11)
a
22

K
1
S
1
K
2
s
K
2
S
1
K
2
s
N
2
K
2
N
1
N
2
S
2
N
3
; (A-12)
a
23

K
1
K
2
S
2
S
1
K
2
s
N
1
N
2
N
3
; (A-13)
and
S
2
2
a
31
u
s
a
32
u
1
a
33
u
2
I; (A-14)
with
a
31
a
13
; (A-15)
a
32
a
23
; (A-16)
a
33

K
2
S
2
K
2
s
K
1
S
2
K
2
s
N
2
K
1
N
1
N
2
S
1
N
3
. (A-17)
We need the stress-strain relationships in terms of the solid dis-
placements u
s
and the relative fluid displacements w
i
u
i
u
s
, the
fluid displacement relative to the current position of the solid ma-
trix. Thus, we add and subtract appropriately weighted u
s
terms. For
example, equation A-6 can be written
1
s
a
1s
u
s
a
12
w
1
a
13
w
2
I
G
fr
_
u
s
u
s
2
3
u
s
I
_
; (A-18)
where
a
1s
a
11
a
12
a
13
: (A-19)
Similarly for the two fluid phases, we can write
S
1
1
a
2s
u
s
a
22
w
1
a
23
w
2
I (A-20)
and
S
2
2
a
3s
u
s
a
32
w
1
a
33
w
2
I; (A-21)
where
a
2s
a
21
a
22
a
23
(A-22)
and
a
3s
a
31
a
32
a
33
: (A-23)
We rename the coefficients in the stress-strain relationships given
above to bring them closer to the form of the stress-strain relation-
ships for a single fluid phase in a poroelastic medium (Pride, 2005):
1
s
K
u
u
s
C
s1
w
1
C
s2
w
2
I
G
m
u
s
u
s
2
3
u
s
I; (A-24)
S
1
1
C
1s
u
s
M
11
w
1
M
12
w
2
I;
(A-25)
and
S
2
2
C
2s
u
s
M
21
w
1
M
22
w
2
I;
(A-26)
where the coefficients are given by G
fr
and the parameters a
ij
:
K
u
a
1s
; (A-27)
C
s1
a
12
; (A-28)
C
s2
a
13
; (A-29)
G
m
G
fr
; (A-30)
C
1s
a
2s
; (A-31)
M
11
a
22
; (A-32)
M
12
a
23
; (A-33)
C
2s
a
3s
; (A-34)
M
21
a
32
; (A-35)
M
22
a
33
: (A-36)
APPENDIX B
APPLICATION OF THE METHOD OF MULTIPLE
SCALES TO EQUATIONS GOVERNING COUPLED
DEFORMATION AND TWO-PHASE FLOW
In this appendix, we use the method of multiple scales to obtain a
system of equations constraining the zeroth-order amplitudes of the
solid displacement U
0
s
and the fluid velocities W
0
1
and W
0
2
in a
heterogeneous poroelastic medium saturated by two fluids. The
condition that these equations have a nontrivial solution is sufficient
to provide equations for the phase velocities of the various modes of
propagation. The motivation for the method of multiple scales is
presented in the main body of the text. In particular, see the discus-
sion surrounding equations 2530.
Let us begin with the first of the governing equations 19, after
expanding all of the spatial derivatives:
G
m
U
s
G
m
U
s
2
3
G
m
U
s
I
G
m
U
s
G
m
U
s
2
3
G
m
U
s
I
K
u
U
s
K
u
U
s
C
s1
W
1
C
s1
W
1
C
s2
W
2
C
s2
W
2

s
U
s

1
W
1

2
W
2
0: (B-1)
The first step involves reformulating the governing equations in
terms of the slow variables, introduced in equation 25. To do this,
we rewrite the differential operators in slow coordinates, as in
equation 29. We then substitute the series representations for the
vectors U
s
and W
i
(see equations 26 and 27), retaining only those
terms containing
0
1 and
1
. We use the definition of l to
arrive at
G
m

_
l
U
s
_
G
m

_
l
U
s
_
T

2
3
G
m

__
l
U
s
_
I
_
G
m

_
l
U
s
_
G
m
l
_
U
s
_
G
m
l
_
l
2
U
s
2
_
G
m

_
l
U
s
_
T
G
m
l
_
U
s
_
T
G
m
l
_
l
2
U
s
2
_
T

2
3
G
m

_
l
U
s
_
I
2
3
G
m
l
_

U
s
_
I
2
3
G
m
l
_
l

2
U
s
2
_
I K
u
_
l
U
s
_
K
u
_
l
U
s
_
K
u
l
_

U
s
_
K
u
l
_
l

2
U
s
2
_
C
s1
_
l
W
1
_
C
s1
_
l
W
1
_
C
s1
l
_

W
1
_
C
s1
l
_
l

2
W
1
2
_
C
s2
_
l
W
2
_
C
s2
_
l
W
2
_
C
s2
l
_

W
2
_
C
s2
l
_
l

2
W
2
2
_

s
U
s

1
W
1

2
W
2
0: (B-2)
We can write equation B-2 more compactly if we use the fact that
U
s
iU
s
(B-3)
and
W
i
iW
i
; (B-4)
which follows from the form of solutions 26 and 27. Making these
substitutions, we can rewrite equation B-2 as
G
m
ilU
s
G
m
ilU
s
T

2
3
G
m
il U
s
I
G
m
ilU
s
G
m
l iU
s
G
m
l lU
s
G
m
ilU
s
T
G
m
l iU
s
T
G
m
l lU
s
T

2
3
G
m
il U
s
I
2
3
G
m
l iU
s
I
2
3
G
m
l l U
s
I K
u
il U
s
K
u
il U
s
K
u
l iU
s
K
u
ll U
s
C
s1
il W
1
C
s1
il W
1
C
s1
l iW
1
C
s1
ll W
1
C
s2
il W
2
C
s2
il W
2
C
s2
l iW
2
C
s2
ll W
2
s
U
s

1
W
1

2
W
2
0: (B-5)
From equation B-5, we can obtain all of the terms necessary for
the first of the three governing equations 19. In particular, we
can extract all terms of order
0
1 that are required to determine
an expression for phase.
We also need the zeroth-order terms for the two fluid equations 20
and 21. We use index notation to represent the pair of equations by a
single expression. The expanded version of the index equation is
given by
C
is
U
s
C
is
U
s
M
i1
W
1
M
i1
W
1
M
i2
W
2
M
i2
W
2
i
U
s

i
W
i
0; (B-6)
where the index i signifies the fluid that is under consideration. Sub-
stituting the differential operators and retaining terms of order
0
and
1
, and using the definition of l,
C
is
_
l
U
s
_
C
is
_
l
U
s
_
C
is
l
_

U
s
_
C
is
l
_
l

2
U
s
2
_
M
i1
_
l
W
1
_
M
i1
_
l
W
1
_
M
i1
l
_

W
1
_
M
i1
l
_
l

2
W
1
2
_
M
i2
_
l
W
2
_
M
i2
_
l
W
2
_
M
i2
l
_

W
2
_
M
i2
l
_
l

2
W
2
2
_

i
U
s

i
W
i
0: (B-7)
Using the property of the partial derivatives given by equations B-3
and B-4, we can write equation B-7 as
iC
is
l U
s
iC
is
l U
s
l U
s
C
is
ll U
s
iM
i1
l W
1
iM
i1
l W
1
l W
1
M
i1
ll W
1
iM
i2
l W
2
iM
i2
l W
2
l W
2
M
i2
ll W
2
i
U
s

i
W
i
0 (B-8)
for i 1, 2 for the two fluids, respectively.
Terms of order zero
In this subsection, we consider terms of the lowest order in ,
terms of order zero. For smoothly varying heterogeneity, such terms
are the most significant. Gathering terms of zeroeth-order from
equation B-5 leads to
G
m
l
2
U
0
s
G
m
ll U
0
s

2
3
G
m
ll U
0
s
K
u
ll U
0
s

s
U
0
s
C
s1
ll W
0
1
C
s2
ll W
0
2

1
W
0
1

2
W
0
2
0; (B-9)
where
ll U
0
s
ll U
0
s
: (B-10)
Note that we can represent equation B-10 as an operator, a dyadic
(Ben-Menahem and Singh, 1981; Chapman, 2004) applied to U
0
s
:
ll U
0
s
ll IU
0
s
; (B-11)
where I is the identity matrix with ones on the diagonal and zeros
off the diagonal. Alternatively, one may think of the dyadic ll as the
vector outer product ll
T
, where l
T
signifies the transpose of l, con-
verting the column vector l to the row vector l
T
.
Combining like terms and defining the coefficients

s
G
m
l
2
(B-12)
and
K
u

1
3
G
m
; (B-13)
we can rewrite equation B-9 as
U
0
s
ll U
0
s

1
W
0
1
C
s1
ll W
0
1

2
W
0
2
C
s2
ll W
0
2
0: (B-14)
We can treat the equations pertaining to the two fluids, as expressed
in B-6, similarly. Collecting the zeroeth-order terms in equation B-8
produces
i
U
0
s
C
is
ll U
0
s
M
i1
ll W
0
1
M
i2
ll W
0
2

i
W
0
i
0;
(B-15)
where the index i takes the values one or two, depending on the
fluid under consideration.
APPENDIX C
REDUCTION OF THE DETERMINANT
In this appendix, we demonstrate that the vanishing of the deter-
minant of the 9 9 coefficient matrix in equation 31,

_
I ll I
1
I C
s1
ll I
2
I C
s2
ll I
1
I C
1s
ll I
1
I M
11
ll I M
12
ll I
2
I C
2s
ll I M
21
ll I
2
I M
22
ll I
_
;
(C-1)
is equivalent to the vanishing of the determinant of a much smaller
3 3 matrix. The determinant of a matrix is given by the product of
its eigenvalues (Noble and Daniel, 1977). Thus, the vanishing of the
determinant is equivalent to the vanishing of one or more eigen-
values of the matrix . Furthermore, there will be an eigenvector
associated with the zero eigenvalue.
Based upon physical considerations, in particular the polariza-
tions of the modes of propagation in a poroelastic medium and
the structure of the matrix C-1, the vectors
e
l
_
y
1
l
y
2
l
y
3
l
_
; (C-2)
e
1
_
z
1
l
1
z
2
l
1
z
3
l
1
_
; (C-3)
and
e
2
_
t
1
l
2
t
2
l
2
t
3
l
2
_
(C-4)
are suggested as potential eigenvectors of the matrix . Here, l
1
and l
2
are two orthogonal vectors lying in the plane perpendicular

to l. Physically, the vector e
l
corresponds to longitudinal propaga-
tion when the fluid and solid displacements are parallel to the
direction of propagation. Conversely, the vectors e
1
and e
2
cor-
respond to transverse motion in which the direction of fluid and
solid displacement is perpendicular to the direction of propagation.
The physical motivation is from wave propagation in a homoge-
neous medium. In a homogeneous medium, one can use potentials
to decompose an elastic disturbance into a longitudinal mode of
propagation and two transverse modes of propagation (Aki and Ri-
chards, 1980). The structure of the matrix also suggests that the
vectors C-2, C-3, and C-4 are potential eigenvectors. Specifically,
each 3 3 submatrix in contains the terms I and ll I. When these
terms are multiplied by l, the results are proportional to l. When the
terms are multiplied by l
1
and l
2
, the first term gives the same vector
but the second term vanishes. Thus, the vector l, and vectors per-
pendicular to it provide special directions for the matrix .
For illustration, we consider the eigenvector e
l
associated with
the longitudinal modes of propagation, displacement in the direc-
tion of propagation l. Because it is an eigenvector, the vector e
l
satisfies the equation
e
l
e
l
: (C-5)
Furthermore, because we are interested in the case in which the
determinant vanishes, the eigenvalue of interest is the one that
vanishes, reducing equation C-5 to
e
l
0: (C-6)
From the algebraic form of the coefficient matrix and the form of
the eigenvector e
l
, equation C-6 is equivalent to
_
_
_
_
l
2
I
1
C
s1
l
2
I
2
C
s2
l
2
I
1
C
1s
l
2
I
1
M
11
l
2
I M
12
l
2
I
2
C
2s
l
2
I M
21
l
2
I
2
M
22
l
2
I
_
_
_
_
_
_
_
y
1
l
y
2
l
y
3
l
_
_
_ 0: (C-7)
The requirement that this equation have a nontrivial solution is the
vanishing of the determinant of the coefficient matrix,
det
_
_
_
l
2
I
1
C
s1
l
2
I
2
C
s2
l
2
I
1
C
1s
l
2
I
1
M
11
l
2
I M
12
l
2
I
2
C
2s
l
2
I M
21
l
2
I
2
M
22
l
2
I
_
_
_ 0:
(C-8)
At this point, we invoke a theorem from linear algebra regarding the
determinant of a matrix composed of block submatrices (Silvester,
2000). The theorem states that the determinant of a matrix tensor
product
L Q
_
l
11
Q l
12
Q l
13
Q
l
21
Q l
22
Q l
23
Q
l
31
Q l
32
Q l
33
Q
_
; (C-9)
where Q is a 3 3 matrix and
L
_
l
11
l
12
l
13
l
21
l
22
l
23
l
31
l
32
l
33
_
; (C-10)
is given by
det L Q det Ldet Q
3
: (C-11)
Applying this theorem to the coefficient matrix in equation C-8 and
making use of the fact that Q is the identity matrix, we derive the
condition
det
_
l
2
1
C
s1
l
2
2
C
s2
l
2
1
C
1s
l
2
1
M
11
l
2
M
12
l
2
2
C
2s
l
2
M
21
l
2
2
M
22
l
2
_
0;
(C-12)
the vanishing of the determinant of a 3 3 matrix.
APPENDIX D
COMPUTING THE DETERMINANT FOR THE
LONGITUDINAL MODE OF PROPAGATION
In this appendix, we detail the computation of the determinant of
the matrix
M
_

s
Hs
1
C
s1
s
2
C
s2
s
1
C
1s
s
1
M
11
s M
12
s
2
C
2s
s M
21
s
2
M
22
s
_
; (D-1)
where s l
2
. The principle that we apply repeatedly relates to the
determinant of a matrix containing a column in which each element
is the sum of two terms. A theorem in linear algebra shows that the
determinant of such a matrix can be written as the sum of two de-
terminants, each of which contains one element of the sum (Noble
and Daniel, 1977, p. 200). We illustrate this principle by an appli-
cation to the matrix M given above. Each element of the first col-
umn of this matrix is the sum of two terms. Thus, we can write the
determinant of M as
det
_
_
_
s
Hs
1
C
s1
s
2
C
s2
s
1
C
1s
s
1
M
11
s M
12
s
2
C
2s
s M
21
s
2
M
22
s
_
_
_
det
_
_
_
s

1
C
s1
s
2
C
s2
s
1

1
M
11
s M
12
s
2
M
21
s
2
M
22
s
_
_
_
s det
_
_
_
H
1
C
s1
s
2
C
s2
s
C
1s

1
M
11
s M
12
s
C
2s
M
21
s
2
M
22
s
_
_
_: (D-2)
We can apply this principle recursively, first to the second column of
each of the component matrices in equation D-2 and then to the
third column of each of the component determinants, obtaining a
cubic equation in s. We can write the cubic equation compactly as
Q
3
s
3
Q
2
s
2
Q
1
s Q
0
0 (D-3)
if we define the coefficients
Q
3
det
_
H C
s1
C
s2
C
1s
M
11
M
12
C
2s
M
21
M
22
_
; (D-4)
Q
2
det
_
_
_
s
C
s1
C
s2
1
M
11
M
12
2
M
21
M
22
_
_
_ det
_
_
_
H
1
C
s2
C
1s

1
M
12
C
2s
0 M
22
_
_
_
det
_
_
_
H C
s1

2
C
1s
M
11
0
C
2s
M
21

2
_
_
_; (D-5)
Q
1
det
_
_
_
s

1
C
s2
1

1
M
12
2
0 M
22
_
_
_ det
_
_
_
s
C
s1

2
1
M
11
0
2
M
21

2
_
_
_
det
_
_
_
H
1

2
C
1s

1
0
C
2s
0
2
_
_
_; (D-6)
Q
0
det
_
s

1

2
1

1
0
2
0
2
_
: (D-7)
REFERENCES
Aki, K., and P. G. Richards, 1980, Quantitative seismology: W. H. Freeman
& Co.
Anile, A. M., J. K. Hunter, P. Pantano, and G. Russo, 1993, Ray methods for
nonlinear waves in fluids and plasmas: Longman Scientific and Technical.
Auriault, J. L., 1980, Dynamic behavior of a porous medium saturated by a
Newtonian fluid: International Journal of Engineering Science, 18, no. 6,
775785, doi: 10.1016/0020-7225(80)90025-7.
Bear, J., 1972, Dynamics of fluids in porous media: Elsevier.
Bear, J., M. Y. Corapcioglu, and J. Balakrishna, 1984, Modeling of centri-
fugal filtration in unsaturated deformable porous media: Advances in
Water Resources, 7, no. 4, 150167, doi: 10.1016/0309-1708(84)
90013-7.
Ben-Menahem, A., and S. J. Singh, 1981, Seismic waves and sources:
Springer Verlag.
Berryman, J. G., P. A. Berge, and B. P. Bonner, 2002, Estimating rock por-
osity and fluid saturation using only seismic velocities: Geophysics, 67,
391404, doi: 10.1190/1.1468599.
Berryman, J. G., L. Thigpen, and R. C. Y. Chin, 1988, Bulk elastic wave
propagation in partially saturated porous solids: Journal of the Acoustical
Society of America, 84, 360373, doi: 10.1121/1.396938.
Biot, M. A., 1956a, Theory of propagation of elastic waves in a fluid-satu-
rated porous solid. I. Low-frequency range: Journal of the Acoustical So-
ciety of America, 28, 168178, doi: 10.1121/1.1908239.
Biot, M. A., 1956b, Theory of propagation of elastic waves in a fluid-satu-
rated porous solid. II. Higher-frequency range: Journal of the Acoustical
Biot, M. A., 1962a, Mechanics of deformation and acoustic propagation in
porous media: Journal of Applied Physics, 33, 14821498, doi: 10.1063/
1.1728759.
Biot, M. A., 1962b, Generalized theory of acoustic propagation in porous
dissipative media: Journal of the Acoustical Society of America, 34,
12541264, doi: 10.1121/1.1918315.
Bracewell, R. N., 1978, The Fourier transform and its applications:
McGraw-Hill Book Company.
Burridge, R., and J. B. Keller, 1981, Poroelasticity equations derived from
microstructure: Journal of the Acoustical Society of America, 70, 1140
1146, doi: 10.1121/1.386945.
Calvert, R., 2005, Insights and methods for 4D reservoir monitoring and
characterization: EAGE/SEG Distinguished Instructor Short Course 8.
Chapman, C. H., 2004, Fundamentals of seismic wave propagation:
Cambridge University Press.
Chen, J., J. W. Hopmans, and M. E. Grismer, 1999, Parameter estimation of
two-fluid capillary pressure Saturation and permeability functions: Ad-
vances in Water Resources, 22, no. 5, 479493, doi: 10.1016/S0309-1708
(98)00025-6.
Courant, R., and D. Hilbert, 1962, Methods of mathematical physics: Wiley
Interscience.
Dean, R., X. Gai, C. Stone, and S. Minkoff, 2006, A comparison of tech-
niques for coupling porous flow and geomechanics: Society of Petroleum
Engineers Journal, 11, no. 1, 132140, doi: 10.2118/79709-PA.
de la Cruz, V., and T. J. T. Spanos, 1985, Seismic wave propagation in a
porous medium: Geophysics, 50, 15561565, doi: 10.1190/1.1441846.
de Marsily, G., 1986, Quantitative hydrogeology: Academic Press.
Domenico, S. N., 1974, Effects of water saturation on seismic reflectivity of
sand reservoirs encased in shale: Geophysics, 39, 759769, doi: 10.1190/
1.1440464.
Domenico, S. N., 1976, Effect of brine-gas mixture on velocity in an un-
consolidated sand reservoir: Geophysics, 41, 882894, doi: 10.1190/1
.1440670.
Dvorkin, J., R. Nolen-Hoeksema, and A. Nur, 1994, The squirt-flow me-
chanism: Macroscopic description: Geophysics, 59, 428438, doi: 10
.1190/1.1443605.
Dvorkin, J., and A. Nur, 1998, Acoustic signatures of patchy saturation: In-
ternational Journal of Solids and Structures, 35, 48034810, doi: 10.1016/
S0020-7683(98)00095-X.
Frenkel, J., 1944, On the theory of seismic and seismoelectric phenomena in
a moist soil: Journal of Physics (in Russian), 8, 230241.
Friedlander, F. G., and J. B. Keller, 1955, Asymptotic expansions of solu-
tions of
2
k
2
u 0: Communications on Pure and Applied Mathe-
matics, 8, no. 3, 387394, doi: 10.1002/cpa.3160080306.
Gajo, A., and L. Mongiovi, 1995, An analytical solution for the transient
response of saturated linear elastic porous media: International Journal
of Numerical and Analytic Methods in Geomechanics, 19, no. 6, 399
413, doi: 10.1002/nag.1610190603.
Garg, S. K., 1971, Wave propagation effects in a fluid-saturated porous so-
lid: Journal of Geophysical Research, 76, 79477962, doi: 10.1029/
JB076i032p07947.
Garg, S. K., and A. H. Nayfeh, 1986, Compressional wave propagation in
liquid and/or gas saturated elastic porous media: Journal of Applied Phy-
sics, 60, 30453055, doi: 10.1063/1.337760.
Johnson, D. L., 2001, Theory of frequency dependent acoustics in patchy-
saturated porous media: Journal of the Acoustical Society of America,
110, 682694, doi: 10.1121/1.1381021.
Johnson, D. L., J. Koplik, and R. Dashen, 1987, Theory of dynamic perme-
ability and tortuosity in fluid-saturated porous media: Journal of Fluid
Mechanics, 176, no. 1, 379402, doi: 10.1017/S0022112087000727.
Kline, M., and I. W. Kay, 1979, Electromagnetic theory and geometrical
optics: John Wiley and Sons.
Korsunsky, S., 1997, Nonlinear waves in dispersive and dissipative systems
with coupled fields: Addison Wesley Longman.
Kosten, C. W., and C. Zwikker, 1941, Extended theory of the absorption of
sound by compressible wall-coverings: Physica, 8, no. 9, 968978, doi:
10.1016/S0031-8914(41)80004-4.
Kravtsov, Y. A., and Y. I. Orlov, 1990, Geometrical optics of inhomogeneous
media: Springer-Verlag.
Levy, T., 1979, Propagation of waves in a fluid-saturated porous elastic so-
lid: International Journal of Engineering Science, 17, no. 9, 10051014,
doi: 10.1016/0020-7225(79)90022-3.
Lo, W.-C., G. Sposito, and E. Majer, 2002, Immiscible two-phase fluid flows
in deformable porous media: Advances in Water Resources, 25, no. 8-12,
11051117, doi: 10.1016/S0309-1708(02)00050-7.
Lo, W.-C., G. Sposito, and E. Majer, 2005, Wave propagation through elastic
porous media containing two immiscible fluids: Water Resources Re-
search, 41, W02025, doi: 10.1029/2004WR003162.
Luneburg, R. K., 1966, Mathematical theory of optics: University of
California Press.
Masson, Y. J., S. R. Pride, and K. T. Nihei, 2006, Finite difference modeling
of Biots poroelastic equations at seismic frequencies: Journal of Geophy-
sical Research, 111, B10305, doi: 10.1029/2006JB004366.
Minkoff, S. E., C. Stone, S. Bryant, and M. Peszynska, 2004, Coupled geo-
mechanics and flow simulation for time-lapse seismic modeling: Geophy-
sics, 69, 200211, doi: 10.1190/1.1649388.
Minkoff, S. E., C. Stone, S. Bryant, M. Peszynska, and M. Wheeler, 2003,
Coupled fluid flow and geomechanical deformation modeling: Journal of
Petroleum Science and Engineering, 38, no. 1-2, 3756, doi: 10.1016/
S0920-4105(03)00021-4.
Mualem, Y., 1976, A new model for predicting the hydraulic conductivity of
unsaturated porous media: Water Resources Research, 12, 513522, doi:
10.1029/WR012i003p00513.
Murphy, W. F., 1982, Effects of partial water saturation on attenuation in
Massilon sandstone and Vycor porous glass: Journal of the Acoustical
Noble, B., and J. W. Daniel, 1977, Applied linear algebra: Prentice-Hall.
Noorishad, J., C.-F. Tsang, and P. A. Witherspoon, 1992, Theoretical and
field studies of coupled behaviour of fractured rocks 1. Development
and verification of a numerical simulator: International Journal of Rock
Mechanics and Mining Sciences & Geomechanics Abstracts, 29, no. 4,
401409, doi: 10.1016/0148-9062(92)90515-2.
Press, W. H., S. A. Teukolsky, W. T. Vetterling, and B. P. Flannery, 1992,
Numerical recipes: Cambridge University Press.
Pride, S. R., 2005, Relationships between seismic and hydrological proper-
ties, in S. R. Pride, ed., Hydrogeophysics: Springer, 253291.
Pride, S. R., A. F. Gangi, and F. D. Morgan, 1992, Deriving the equations of
media for isotropic motion: Journal of the Acoustical Society of America,
92, 32783290, doi: 10.1121/1.404178.
Pride, S. R., F. D. Morgan, and A. F. Gangi, 1993, Drag forces of porous-
medium acoustics: Physical Review B, 47, no. 9, 49644978, doi: 10
.1103/PhysRevB.47.4964.
Rubin, Y., and S. S. Hubbard, 2006, Hydrogeophysics: Springer.
Rutqvist, J., Y. S. Wu, C. F. Tsang, and G. Bodvarsson, 2002, A modeling
approach for analysis of coupled multiphase fluid flow, heat transfer, and
deformation in fractured porous rock: International Journal of Rock
Mechanics and Mining Sciences & Geomechanics Abstracts, 39, no.
4, 429442, doi: 10.1016/S1365-1609(02)00022-9.
Santos, J. E., J. M. Corbero, and J. Douglas Jr., 1990, Static and dynamic
behavior of a porous solid saturated by a two-phase fluid: Journal of
the Acoustical Society of America, 87, 14281438, doi: 10.1121/1
.399439.
Sethian, J. A., 1985, Curvature and the evolution of fronts: Communications
in Mathematical Physics, 101, no. 4, 487499, doi: 10.1007/
BF01210742.
Sethian, J. A., 1999, Level set and fast marching methods: Cambridge Uni-
versity Press.
Silvester, J. R., 2000, Determinants of block matrices: The Mathematical
Gazette, 84, 460467.
Simon, B. R., O. C. Zienkiewicz, and D. K. Paul, 1984, An analytical solu-
tion for the transient response of saturated porous elastic solids: Interna-
tional Journal for Numerical and Analytic Methods in Geomechanics, 8,
no. 4, 381398, doi: 10.1002/nag.1610080406.
Slattery, J. C., 1968, Multiphase viscoelastic fluids through porous media:
American Institute of Chemical Engineering Journal, 14, no. 1, 5056,
doi: 10.1002/aic.690140112.
Slattery, J. C., 1981, Momentum, energy, and mass transfer in continua:
Robert E. Krieger Publishing.
Stahl, S., 1997, Introductory modern algebra: John Wiley & Sons.
Tuncay, K., 1995, Wave propagation in single- and double-porosity deform-
able porous media saturated by multiphase fluids: Ph.D. thesis, Texas
A&M University.
Tuncay, K., and M. Y. Corapcioglu, 1996, Body waves in poroelastic media
saturated by two immiscible fluids: Journal of Geophysical Research,
101, no. B11, 2514925159, doi: 10.1029/96JB02297.
Tuncay, K., and M. Y. Corapcioglu, 1997, Wave propagation in poroelastic
media saturated by two fluids: Journal of Applied Mechanics, 64, no. 2,
313320, doi: 10.1115/1.2787309.
van Genuchten, M. T., 1980, Aclosed-formequationfor predictingthe hydrau-
lic conductivity of unsaturated soils: Soil Science Society of America Journal,
44, no. 5, 892898, doi: 10.2136/sssaj1980.03615995004400050002x.
Vasco, D. W., 2008, Modeling quasi-static poroelastic propagation using an
asymptotic approach: Geophysical Journal International, 173, 1119
1135, doi: 10.1111/gji.2008.173.issue-3.
Vasco, D. W., 2009, Modelling broad-band poroelastic propagation using an
asymptotic approach: Geophysical Journal International, 179, 299318,
doi: 10.1111/gji.2009.179.issue-1.
Vasco, D. W., 2011, On the propagation of a coupled saturation and pressure
front: Water Resources Research, 47, W03526, doi: 10.1029/
2010WR009740.
Vasco, D. W., and S. E. Minkoff, 2009, Modelling flow in a pressure-sen-
sitive, heterogeneous medium: Geophysical Journal International, 179,
972989, doi: 10.1111/gji.2009.179.issue-2.
Vidale, J., 1988, Finite-difference calculation of travel times: Bulletin of the
Seismological Society of America, 78, 20622076.
Wang, R., and H.-J. Kumpel, 2003, Poroelasticity: Efficient modeling of
strongly coupled, slow deformation processes in a multilayered half-
space: Geophysics, 68, 705717, doi: 10.1190/1.1567241.
Whitham, G. B., 1974, Linear and nonlinear waves: John Wiley and Sons.
Wyckoff, R. D., and H. G. Botset, 1936, The flow of gas-liquid mixtures
through unconsolidated sands: Physics, 7, no. 9, 325345, doi: 10
.1063/1.1745402.

Vascoand Minkoff

Uploaded by

Copyright:

Available Formats

Vascoand Minkoff

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

Vascoand Minkoff

Uploaded by

Copyright:

Available Formats

On the propagation of a disturbance in a heterogeneous, deformable,

porous medium saturated with two fluid phases

is the associated eigenvalue. Invoking similar arguments

are two orthogonal vectors lying in the plane perpendicular

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.