-
RESERVOIR ANALYSIS USING INTERMEDIATE
FREQUENCY EXCITATION
a dissertation
submitted to the department of petroleum engineering
and the committee on graduate studies
of stanford university
in partial fulfillment of the requirements
for the degree of
doctor of philosophy
By
Yan Pan
August 1999
-
c© Copyright by Yan Pan 1999All Rights Reserved
ii
-
I certify that I have read this thesis and that in my opin-
ion it is fully adequate, in scope and in quality, as a
dissertation for the degree of Doctor of Philosophy.
Dr. Roland N. Horne(Principal Advisor)
I certify that I have read this thesis and that in my opin-
ion it is fully adequate, in scope and in quality, as a
dissertation for the degree of Doctor of Philosophy.
Dr. Khalid Aziz
I certify that I have read this thesis and that in my opin-
ion it is fully adequate, in scope and in quality, as a
dissertation for the degree of Doctor of Philosophy.
Dr. Thomas A. Hewett
Approved for the University Committee on Graduate
Studies:
iii
-
Acknowledgements
I would like to thank my research advisor Professor Roland N.
Horne for his constant
support, encouragement and guidance during the course of this
study. Professor
Horne is always available to answer my questions, or to help me
in finding the right
path to the answers. When I felt depressed because of some
unsuccessful strugglings,
Professor Horne kept his belief in my ability of solving
problems and encouraged me
to overcome the difficulties eventually. I could not have
finished this research project
without Professor Horne’s guidance and encouragement.
Sincere thanks are due to my friends, Jing Wan, Suwat
Athichanagorn, Erik Skjetne,
Hongkai Zhao for generously providing valuable ideas and
techniques for this research.
I would like to express my special appreciation to my boyfriend,
Jorge Landa, for
his love, kindness, patience and constant spiritual and academic
support. I always
benefit from his generosity of intelligence and experiences.
I would like to thank my family for their love and support. I am
lucky to be able to
stay in a warm and comfortable home with my family while I am
pursuing the degree
of PhD. My gratitude is endless to my parents, Zhengpu Pan and
Ruyuan Zhang,
and to my brother Lei Pan.
Financial support from the members of the SUPRI-D Research
Consortium on In-
novation in Well Testing, the Department of Petroleum
Engineering, and Chevron
Scholarship Program is gratefully acknowledged.
iv
-
Abstract
This study investigated mathematical models to describe the
motion of fluids in
porous media, and applied these models to harmonic well testing
at intermediate
frequencies. The purpose was to examine the possibility of
obtaining more informa-
tion about reservoirs than can be usually derived in
conventional well testing (low
frequency excitation) and seismic data processing (high
frequency excitation). The
problem of fluid flow in the pores or small channels of a
periodic elastic solid matrix
was studied at pore scale, and the homogenization technique was
applied to predict
the macroscopic behavior of reservoirs.
The theoretical analysis and the numerical results show that the
responses of a porous
medium to harmonic perturbations depend on the parameters of the
pore structure,
the properties of fluid and solid and the frequencies of the
excitation signals. The
effective parameters, such as dynamic permeability and porosity,
are also functions of
perturbation frequencies. From the investigation of the coupling
effects of fluid and
solid motions, it seems that the elastic solid vibration has
positive impact on fluid
flow under harmonic perturbations of intermediate frequencies,
which may provide a
potential new technique for stimulation of oil production.
Based on the homogenization study, five separate characteristic
macroscopic model
were identified according to the relation between a length scale
parameter and a prop-
erty contrast number. These five models can be used to interpret
the corresponding
responses of a reservoir. It is possible to infer the effective
parameters of porous
media, such as porosity and fracture density, by analyzing the
diphasic macroscopic
v
-
response to cyclic excitation at various frequencies.
vi
-
Contents
Acknowledgements iv
Abstract v
1 Introduction 1
1.1 Statement of the Problem . . . . . . . . . . . . . . . . . .
. . . . . . 1
1.2 Literature Review . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . 3
1.3 Outline of Approach . . . . . . . . . . . . . . . . . . . .
. . . . . . . 11
2 Homogenization Techniques 16
2.1 Acoustics of an Empty Porous Medium . . . . . . . . . . . .
. . . . . 18
2.1.1 Local Description . . . . . . . . . . . . . . . . . . . .
. . . . . 18
2.1.2 Macroscopic Description . . . . . . . . . . . . . . . . .
. . . . 20
2.2 A Prior Estimate for Saturated Porous Media . . . . . . . .
. . . . . 23
2.3 Local Description of Saturated Porous Media . . . . . . . .
. . . . . . 25
2.4 Acoustics of a Fluid in a Rigid Porous Medium . . . . . . .
. . . . . 27
2.5 Diphasic Macroscopic Behavior . . . . . . . . . . . . . . .
. . . . . . 29
2.6 Monophasic Elastic Macroscopic Behavior . . . . . . . . . .
. . . . . 36
2.7 Monophasic Viscoelastic Macroscopic Behavior . . . . . . . .
. . . . . 37
3 Mathematical Model 41
3.1 Pore-Scale Problem . . . . . . . . . . . . . . . . . . . . .
. . . . . . . 41
3.1.1 Motion of Elastic Solid . . . . . . . . . . . . . . . . .
. . . . . 43
3.1.2 Fluid Flow in Elastic Channel . . . . . . . . . . . . . .
. . . . 44
vii
-
3.2 Macroscopic Behavior . . . . . . . . . . . . . . . . . . . .
. . . . . . . 52
3.2.1 Acoustics of Empty Channels . . . . . . . . . . . . . . .
. . . 54
3.2.2 Acoustics of a Fluid in a Rigid Channel . . . . . . . . .
. . . . 59
3.2.3 Diphasic Macroscopic Behavior . . . . . . . . . . . . . .
. . . 63
3.2.4 Monophasic Elastic Behavior . . . . . . . . . . . . . . .
. . . . 70
3.2.5 Monophasic Viscoelastic Behavior . . . . . . . . . . . . .
. . . 75
3.3 Relation to Existing Theories . . . . . . . . . . . . . . .
. . . . . . . 75
3.3.1 Darcy’s Law . . . . . . . . . . . . . . . . . . . . . . .
. . . . . 75
3.3.2 Telegrapher’s Equation . . . . . . . . . . . . . . . . . .
. . . . 76
3.3.3 Biot’s Theory . . . . . . . . . . . . . . . . . . . . . .
. . . . . 78
4 Numerical Methods 82
4.1 Program Structure . . . . . . . . . . . . . . . . . . . . .
. . . . . . . 82
4.2 Finite Element Method for Elasticity . . . . . . . . . . . .
. . . . . . 82
4.2.1 Strong Form . . . . . . . . . . . . . . . . . . . . . . .
. . . . . 82
4.2.2 Variational Form . . . . . . . . . . . . . . . . . . . . .
. . . . 84
4.2.3 Galerkin Formulation . . . . . . . . . . . . . . . . . . .
. . . . 85
4.2.4 Matrix Equations . . . . . . . . . . . . . . . . . . . . .
. . . . 86
4.2.5 Bossak’s Algorithm for Structural Dynamics . . . . . . . .
. . 88
4.2.6 Bilinear Quadrilateral Element . . . . . . . . . . . . . .
. . . 90
4.2.7 Element Matrices and Force . . . . . . . . . . . . . . . .
. . . 93
4.2.8 Gaussian Quadrature Numerical Integration . . . . . . . .
. . 95
4.3 Finite Difference Method for Fluid Flow . . . . . . . . . .
. . . . . . 97
4.3.1 Discretized Governing Equations . . . . . . . . . . . . .
. . . 97
4.3.2 Treatment of Velocity-Pressure Linkage . . . . . . . . . .
. . . 99
4.3.3 Procedure . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . 100
5 Analysis of Results 102
5.1 Hagen-Poiseuille Flow in Elastic Channels . . . . . . . . .
. . . . . . 102
5.1.1 Effect of Initial Porosity . . . . . . . . . . . . . . . .
. . . . . 103
5.1.2 Effect of Reynolds Number . . . . . . . . . . . . . . . .
. . . . 106
5.1.3 Effect of P-Wave Velocity . . . . . . . . . . . . . . . .
. . . . 106
viii
-
5.1.4 Effect of Poisson’s Ratio . . . . . . . . . . . . . . . .
. . . . . 108
5.1.5 Effect of Density Ratio . . . . . . . . . . . . . . . . .
. . . . . 108
5.1.6 Effect of Channel Tortuosity . . . . . . . . . . . . . . .
. . . . 109
5.2 Harmonic Motion of Fluid in Rigid Channels . . . . . . . . .
. . . . . 112
5.3 Harmonic Motion of Fluid in Elastic Channels . . . . . . . .
. . . . . 115
5.3.1 Effects of Frequency on Fluid Motion . . . . . . . . . . .
. . . 126
5.3.2 Effect of Initial Porosity on Fluid Motion . . . . . . . .
. . . . 126
5.3.3 Resonant Behavior . . . . . . . . . . . . . . . . . . . .
. . . . 127
6 Applications 132
6.1 Selection of Macroscopic Models . . . . . . . . . . . . . .
. . . . . . . 132
6.2 Inference of Fracture Density From Pulse Decay Test . . . .
. . . . . 134
6.3 Inference of Porosity From Harmonic Test . . . . . . . . . .
. . . . . 135
6.4 Optimal Frequency Range for Stimulation of Oil Production .
. . . . 137
6.5 Benefit to Environment . . . . . . . . . . . . . . . . . . .
. . . . . . . 140
6.6 Attenuation of Signals . . . . . . . . . . . . . . . . . . .
. . . . . . . 140
7 Conclusions 142
Nomenclature 144
References 147
Appendix 156
A Computer Programs 157
A.1 General Instructions . . . . . . . . . . . . . . . . . . . .
. . . . . . . 157
A.2 Files in makefile . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . 158
A.3 Data Files . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . 159
A.3.1 Input Data File . . . . . . . . . . . . . . . . . . . . .
. . . . . 159
A.3.2 Output Data Files . . . . . . . . . . . . . . . . . . . .
. . . . 161
A.4 .gps Files for Plots . . . . . . . . . . . . . . . . . . . .
. . . . . . . . 161
ix
-
List of Tables
4.1 Nodal Coordinates in ξ-space . . . . . . . . . . . . . . . .
. . . . . . 91
4.2 Integration Point Indices . . . . . . . . . . . . . . . . .
. . . . . . . . 96
4.3 Variable, Parameter and Source Term . . . . . . . . . . . .
. . . . . . 98
4.4 Source Terms . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . 99
5.1 Natural Frequencies of Solid Matrix . . . . . . . . . . . .
. . . . . . . 129
6.1 Selection of Macroscopic Models . . . . . . . . . . . . . .
. . . . . . . 133
x
-
List of Figures
1.1 A Periodic Porous Medium . . . . . . . . . . . . . . . . . .
. . . . . . 12
1.2 Homogenization Procedure . . . . . . . . . . . . . . . . . .
. . . . . . 14
2.1 Periodic Unit Cell of the Porous Media . . . . . . . . . . .
. . . . . . 17
3.1 Pore-Scale Problem . . . . . . . . . . . . . . . . . . . . .
. . . . . . . 42
3.2 One-Dimensional Model . . . . . . . . . . . . . . . . . . .
. . . . . . 45
3.3 Coordinate System for Macroscopic Behavior Analysis . . . .
. . . . . 57
4.1 Program Structure . . . . . . . . . . . . . . . . . . . . .
. . . . . . . 83
4.2 Solid Problem Domain . . . . . . . . . . . . . . . . . . . .
. . . . . . 84
4.3 Bilinear Quadrilateral Element Domain . . . . . . . . . . .
. . . . . . 90
4.4 Block-Centered Grid . . . . . . . . . . . . . . . . . . . .
. . . . . . . 98
5.1 Hagen-Poiseuille Flow φc = 0.25 - Example 1 . . . . . . . .
. . . . . . 104
5.2 Hagen-Poiseuille Flow φc = 0.60 - Example 2 . . . . . . . .
. . . . . . 105
5.3 Effect of Elasticity at Different Initial Porosity . . . . .
. . . . . . . . 106
5.4 Effect of Reynolds Number . . . . . . . . . . . . . . . . .
. . . . . . . 107
5.5 Effect of P-Wave Velocity of Solid . . . . . . . . . . . . .
. . . . . . . 107
5.6 Effect of Poisson’s Ratio . . . . . . . . . . . . . . . . .
. . . . . . . . 108
5.7 Effect of Density Ratio . . . . . . . . . . . . . . . . . .
. . . . . . . . 109
5.8 Hagen-Poiseuille Flow g(x) = 0.1 sin(2πx/xl) - Example 3 . .
. . . . . 110
5.9 Effect of Channel Tortuosity . . . . . . . . . . . . . . . .
. . . . . . . 111
5.10 Response of a Fluid in a Rigid Channel to Harmonic
Perturbation . . 113
5.11 Phase Map of Fluid Motion in a Rigid Channel . . . . . . .
. . . . . 113
xi
-
5.12 Effects of Perturbation Signal Frequency - Model II . . . .
. . . . . . 114
5.13 Harmonic Motion of Fluid in Elastic Channels - Example 4 .
. . . . . 116
5.14 Normalized Maps of Solid Displacements - Example 4 . . . .
. . . . . 117
5.15 Harmonic Motion of Fluid in Elastic Channels - Example 5 .
. . . . . 118
5.16 Phase Maps - Example 5 . . . . . . . . . . . . . . . . . .
. . . . . . . 120
5.17 Normalized Maps of Solid Displacements - Example 5 . . . .
. . . . . 120
5.18 Harmonic Motion of Fluid in Elastic Channels - Example 6 .
. . . . . 121
5.19 Phase Maps - Example 6 . . . . . . . . . . . . . . . . . .
. . . . . . . 123
5.20 Normalized Maps of Solid Displacements - Example 6 . . . .
. . . . . 123
5.21 Harmonic Motion of Fluid in Elastic Channels - Example 7 .
. . . . . 124
5.22 Phase Maps - Example 7 . . . . . . . . . . . . . . . . . .
. . . . . . . 125
5.23 Normalized Maps of Solid Displacements - Example 7 . . . .
. . . . . 125
5.24 Effects of Perturbation Frequency on Fluid Motion in
Elastic Channels 126
5.25 Effect of Initial Porosity on Fluid Motion in Elastic
Channels . . . . . 127
5.26 Effects of Perturbation Frequency on Fluid and Solid
Motions . . . . 130
5.27 Effects of Perturbation Frequency on Fluid and Solid
Motions . . . . 131
6.1 Setup of Pulse Decay Test . . . . . . . . . . . . . . . . .
. . . . . . . 134
6.2 Pressure Transient Responses in Pulse Decay Test . . . . . .
. . . . . 136
6.3 Relation Between Vibration Frequency and Fracture Density .
. . . . 137
6.4 Optimal Frequencies for Stimulation . . . . . . . . . . . .
. . . . . . 138
xii
-
Chapter 1
Introduction
1.1 Statement of the Problem
In petroleum engineering and hydrology, well testing is used to
investigate the prop-
erties (permeability, porosity, etc.) of reservoirs. Normally
the transient change of
pressure in the porous medium during conventional well testing
is described by the
diffusion equation. This equation neglects the inertial effects
on the fluid flow which
in some cases are important. On the other hand, seismic data
used in geophysics pro-
vide other types of information about the rocks, such as
compression and shear wave
velocities, reflection and transmission coefficients, dry and
saturated elastic moduli,
etc.. The underlying mechanism is the acoustic propagation of
high frequency signals,
which can be described by the wave equation. In fact, the motion
of fluids and solids
in porous media is a complicated nonlinear process which
involves the two-phase cou-
pling of fluid and solid, which may be approximated as a
diffusion process or a wave
propagation only under certain conditions. If a new test could
be designed to measure
signals propagating at an intermediate frequency higher than the
pressure pulse yet
lower than the high frequency seismic signals, it may be
possible to determine addi-
tional reservoir information beyond that usually derived from
well testing or seismic
data processing.
The effects of fluid and solid two-phase coupling can be
observed in many natural
1
-
CHAPTER 1. INTRODUCTION 2
phenomena. One of the early observations of periodic tidal
effects on fluid accumula-
tions in porous strata in the earth can be dated back more than
one hundred years.
This phenomenon is a result of the relative displacement of the
particles composing
a celestial body caused by the gravitational attraction between
the sun, moon and
the earth. In petroleum reservoirs, observations accumulated
during the last 40 years
suggest that seismic waves generated from earthquakes and
cultural noise may alter
water and oil production. It has also been observed in some
laboratory measure-
ments and field applications that imposing harmonic signals into
cores or reservoirs
sometimes may induce higher fluid flow rate. The mechanism of
these phenomena is
still unclear and the observations themselves are not
unambiguous. This study will
investigate this problem and seek a theoretical explanation of
such a phenomenon. If
true, this mechanism may provide a potential new technique for
enhanced oil recovery.
Fluid flow in the reservoir is a two-scale problem. The small
scale involves fluid
flow in the pores and small channels. The larger scale
macroscopic behavior of the
reservoir can be observed in well testing and seismic data
processing. The macro-
scopic response to harmonic signals depends on the pore
structure, the fluid and solid
properties and the signal frequency. It is not appropriate to
apply Darcy’s law to
both scales without first analyzing the individual problem. For
each particular case,
a corresponding macroscopic model has to be chosen to describe
the large scale be-
havior of the reservoir according to the scale of two
characteristic parameters: one
permeability related length-scale ratio and one property
contrast number. Special
attention has to be paid to a finely heterogeneous medium to
determine whether an
equivalent macroscopic description exists. These mathematical
treatments not only
clarify on which assumptions a macroscopic law is based, but
also allow development
of techniques that can be applied to similar and related
problems. The homogeniza-
tion approach provides better understanding of the influence of
the structure of the
porous media and of the concepts of effective permeability and
other macroscopic
properties. Therefore this study uses the homogenization
procedure to study the
two-scale problem and apply the results to reservoir
characterization.
-
CHAPTER 1. INTRODUCTION 3
1.2 Literature Review
The modification of Darcy’s law to include momentum balance for
steady single phase
flow through porous media has been discussed for many years.
There have been sev-
eral different approaches to derive more accurate equations to
describe the motion of
fluids in porous media. Forchheimer (1901) suggested adding
terms of higher order
in the velocity. Klinkenberg (1941) demonstrated that the
permeability coefficient in
Darcy’s law depends on the absolute pressure or, alternatively,
on the density field.
Oroveanu and Pascal (1959) noted that the time derivative of the
momentum density
must be included in the equations of motion, their differential
equation for pressure
is commonly known as the Telegrapher’s equation. Slattery (1963)
presented an ex-
tension of Darcy’s law to unsteady state flow in anisotropic
media. Slattery (1966)
discussed local volume-averaging of the equations of continuity
and of momentum
balance. Foster, McMillen and Odeh (1967) proposed complete
equations of average
linear momentum balance for a single-phase fluid in an
incompressible homogeneous,
porous medium, and demonstrated that for transient flow of
compressible Newto-
nian fluid, the space-time description of the pressure is
determined with the lowest
approximation by the Telegrapher’s equation. Foster, McMillen
and Wallick (1968)
performed an experimental study with a perfect gas in a porous
medium, and derived
an approximate nonlinear description of the space-time behavior
of a small propa-
gating pressure pulse. Odeh and McMillen (1972) did theoretical
research on the
diffusion equation and performed experimental work on pulse
propagation in linear
cores saturated with air. Due to the technical limitation at
that time, these ideas
were not explored in any detail, and no reservoir testing
procedure was proposed.
The theory of acoustic propagation in fluid-saturated porous
media was developed
by Biot (1956, 1962). Since then the model has been extended and
applied widely in
geophysics. Bonnet (1987) reviewed some extensions of Biot
equations and derived
the complete basic singular solution in the frequency domain for
dynamic poroelastic-
ity problems in analogy with thermoelasticity. Depollier, Allard
and Lauriks (1988)
compared the equations of continuity of the models traditionally
used to predict the
-
CHAPTER 1. INTRODUCTION 4
acoustical properties of sound-absorbing materials to simplified
expressions obtained
from the Biot model, and pointed out the inconsistencies in
traditional models. Nor-
ris (1992) explored the analogy between the equations of static
poroelasticity and
the equations of thermoelasticity including entropy, and
discussed the method of de-
termining the effective parameters in an inhomogeneous
poroelastic medium using
known results from the literature on the effective thermal
expansion coefficient and
the effective heat capacity of a disordered thermoelastic
continuum. Zimmerman and
Stern (1994) derived the pressure-solid displacement form of the
harmonic equations
of motion for a poroelastic solid from the form of equations
originally proposed by
Biot. The analytical solutions for several basic problems were
presented. Dai, Vafidis
and Kanasewich (1995) considered Biot’s hyperbolic system and
developed a particle
velocity-stress, finite-difference method for the simulation of
wave propagation in two-
dimensional heterogeneous poroelastic media. Abousleiman, Bai
and Roegiers (1995)
studied the effects of fluid flow coupled with displacements in
and around pressurized
tunnels in fractured formations using Biot’s theory. Chen,
Harstad and Teufel (1996)
followed Biot’s theory and extended the conventional fluid-flow
dual-porosity for-
mulations to a coupled fluid-flow-geomechanics model.
Interpolation of pore volume
compressibilities and the associated effective stress laws were
identified to be the most
critical coupling considerations. Abousleiman et al. (1996)
addressed the phenomena
of mechanical creep and deformation in rock formations coupled
with the hydraulic
effects of fluid flow. The theory was based on Biot’s
poroelasticity, generalized to
encompass viscoelastic effects through the correspondence
principle. Atkinson and
Craster (1996) studied a porous elastic material with an
isotropic elastic response,
but highly anisotropic permeability. It was shown that for a
crack with impermeable
faces and oriented to be parallel to the fluid flow direction,
the highly anisotropic
results were good approximations to those found for a material
with isotropic per-
meability. Chen (1996) presented a purely poroelastodynamic
Boundary Element
Method formulation in the Laplace domain, and solved the
transient response of a
long cylindrical cavity in a poroelastic medium under different
boundary conditions
and the transient response of a fluid-saturated porous
half-space to a sudden strip
load on the surface. De Campos and Neto (1996) implemented the
direct Boundary
-
CHAPTER 1. INTRODUCTION 5
Element Method to analyze the poroelastic effects on two linear
quasistatic elasticity
problems, namely the infinite plate with a circular hole and the
thick walled cylinder.
Cui et al (1996) investigated the classical Mandel’s problem,
which demonstrated
the existence of a nonmonotonic pressure response for a
saturated porous medium
subject to constant external loading, with an extension to
transversely isotropic case,
and explored the problem of an inclined borehole using the
finite element equations
for incrementally nonlinear anisotropic poroelasticity. Ochs,
Chen and Teufel (1997)
examined the flow-induced rock-stress responses to the transient
fluid-pressure for
a well with a stationary vertical fracture within the framework
of Biot’s theory of
poroelasticity.
Investigation of the diffusion model and other related models by
using field data and
experimental measurements is important. Some field testing and
laboratory work
using pulse pressure or harmonic signals have been done to
estimate the permeabili-
ties and other properties of the porous media. It appears that
harmonic testing may
be superior to conventional well testing in some cases.
Crosnier, Fras and Jouanna
(1985) applied harmonic techniques to a physical laboratory
model simulating frac-
tured media, and proposed a mathematical method to determine the
parameters, such
as the number of sets of cracks and the different thickness of
each set, from the spec-
tral signatures. The comparison between theory and laboratory
tests showed a good
fit in the medium range of frequencies. Saeedi and Standen
(1987) designed a layer
pulse test and implemented and analyzed it in a pinnacle reef in
Alberta. Charlaix
et al. (1988a, 1991) did a series of experimental studies on
hydrodynamic dispersion
in networks of capillaries. Charlaix et al. (1988b, 1992) used
harmonic techniques in
experiments to study the dynamics of fluid flow in capillaries.
The dynamic perme-
ability was measured at a frequency range of 0.1 Hz to 1 kHz.
Gilicz (1991) applied
the pulse-decay technique for radial cores. Kamath, Boyer and
Nakagawa (1992)
developed analytical methods and an experimental framework for
obtaining and in-
terpreting the time dependence of pressure response to an
initial disturbance, and
demonstrated that the effective permeability calculated from an
interference pressure
transient test could be a function of the direction of the
pressure disturbance, and
-
CHAPTER 1. INTRODUCTION 6
could differ significantly from the effective steady state
value. Rigord, Caristan and
Hulin (1993) presented an experimental study and a model of the
diffusion of sinu-
soidal pressure waves through porous media, and showed that
measurements of the
hydraulic admittance in the sine wave mode allow us to probe the
structure of porous
samples with an adjustable investigation depth depending on the
frequency. Rigord,
Charlaix and Petit (1996) measured the dynamic viscosity and
shear modulus on
suspension of non-Brownian hard spheres subjected to an
oscillating flow in a tube
at the frequency range of 0.1 Hz to 80 Hz. Rashidi et al. (1996)
studied chemical
flow and transport in an experimental porous medium. The
microscopic medium
geometry, velocity and concentration field, dispersive solute
fluxes and reasonable es-
timates of a representative elementary volume (REV) for the
porous medium were
obtained. Chin and Proett (1997) invented the reservoir
description tester (RDT)
and provided a method to determine geologic formation properties
using phase shift
periodic pressure pulse testing. The spherical transient flow
model used to evaluate
the test results was further discussed by Skinner et al. (1997)
and Proett and Chin
(1998). Most of these laboratory studies were based on
traditional theories of fluid
flow in porous media. Few experiments were performed to
investigate approaches
other than the diffusion process.
Harmonic techniques have also been used in numerical studies. Ni
(1988) analyzed the
one-dimensional nonlinear oscillations of an ideal gas in a pipe
due to a periodically
varying pressure perturbation. Ganiev, Petrov and Ukrainskii
(1990) examined the
steady-state oscillations of a simple model of a well face
region, which is a straight
circular cylindrical channel filled with fluid. Oscillations
were caused by harmonic
pressure perturbation at the end of the channel connected with
the well face. The
model was based on Biot’s theory. The results showed that the
principal resonant
frequency only depends weakly on parameters of the porous solid
and strongly on
channel length. Rosa and Horne (1991) formulated a mathematical
model for the
pressure behavior due to sinusoidal flow rates in reservoirs
with continuously varying
radial permeability distributions. The solution was obtained by
the application of
-
CHAPTER 1. INTRODUCTION 7
a regular perturbation analysis and a correlation between the
radius of cyclic influ-
ence and the frequency of a sinusoidal flow rate. Only pressure
diffusion behavior
was considered. Siginer and Valenzuela-Rendón (1995)
investigated the unsteady
nonviscometric motion of a memory integral fluid of order three
in a rigid, circular,
straight tube. The parametric studies of the longitudinal
oscillatory velocity field
and the change in mass transport were presented for a range of
liquids and driving
conditions. The results showed that elasticity of the fluid
tends to increase the lon-
gitudinal steady velocity if driven by longitudinal waves, but
elasticity of the fluid
tends to decrease the enhancement in the case of transversal
waves. Bernabé (1997)
calculated the dynamic permeability of heterogeneous networks of
cracks, tubes and
spheres by simulating the harmonic flow of an interstitial fluid
for a wide range of
frequencies. The frequency dependence of the real and imaginary
parts of the per-
meability showed a transition from viscous macroscopic flow at
low frequencies to
inertial flow at high frequencies. Buschmann (1997) developed a
two-stage numerical
method to represent harmonic response functions from creep and
stress-relaxation
tests for linear viscoelastic materials.
The analysis of reservoir responses to earth tides is one of the
early studies of fluid
and solid two-phase coupling effects on fluid flow in porous
media. Arditty and
Ramey (1978) derived an expression for the pressure induced at
the borehole by a
periodic tidal stress. Hemala and Balnaves (1986) provided an
overview of analyt-
ical interpretation of tidal effects from a petroleum
engineering point of view and
proposed the application of the phase shift of ocean tide
effects to estimate reservoir
heterogeneities. Dean et al. (1991) introduced a method to
monitor compaction and
compressibility changes in offshore chalk reservoirs by
measuring formation pressure
variations caused by the sea tide. Pinilla et al. (1997)
presented a model to describe
the oceanic tidal effects on an infinite reservoir by coupling
geomechanic principles
with fluid flow equations in a deformable porous media. At the
microscopic scale or
pore scale, the two-phase coupling effects are often studied by
analyzing the fluid flow
in deformable tubes or small channels. Padmanabhan and Pedley
(1987) analyzed the
-
CHAPTER 1. INTRODUCTION 8
steady streaming generated in an infinite elliptical tube
containing a viscous, incom-
pressible fluid when the boundary oscillates in such a way that
the area and ellipticity
of the cross-section vary with time but remain independent of
the longitudinal coor-
dinate. Ganiev et al. (1986) analyzed the linear stability of
viscous incompressible
flow in a circular viscoelastic tube and demonstrated that the
Poiseuille flow in an
elastic tube with respect to infinitesimal axisymmetric
perturbations can be unstable.
Ganiev et al. (1988a, 1988b) further considered disturbances
with different azimuthal
wave numbers and small arbitrary three-dimensional perturbations
and analyzed the
flow stability over a broad interval of values of the elasticity
parameter. Ganiev et
al. (1989) studied the wave mechanism of the fluid motion
acceleration in capillar-
ies and porous media and observed that small-scale pulsation of
the pressure and
velocity may lead to the appearance of unidirectional flows with
velocities substan-
tially exceeding the filtration velocities. Ganiev and
Ukrainskii (1992) investigated
the bifurcation problem near the neutral curves constructed for
Poiseuille flow in a
compliant pipe. The bifurcated self-excited oscillation modes
were determined, and
the effects of Reynolds number, the compliance and internal
viscosity of the pipe ma-
terial was analyzed. These studies assumed that the motions of
the tube walls follow
the couple-stress theory of thin shells. In biomechanics, a
similar phenomenon has
been studied by analyzing blood flow in elastic arteries. Hughes
and Lubliner (1973)
developed a theory of one-dimensional flow through distensible
tubes with perme-
able walls. Reuderink et al. (1989) evaluated the accuracy of
nonlinear and linear
one-dimensional models in describing pulse wave propagation in a
uniform cylindrical
viscoelastic tube at 1 Hz. The calculations were compared with
experimental results.
Dutta et al. (1992) simulated numerically the oscillatory and
pulsatile flows of New-
tonian fluids in straight elastic tubes, and indicated that the
flow field and associated
wall shear stress are extremely sensitive to the phase angle
between oscillatory pres-
sure and flow waves. Wang and Tarbell (1995) further considered
the nonlinear effects
on the amplitude of the wall shear rate, on the amplitude of the
pressure gradient,
and on the mean velocity profile, and developed a perturbation
solution. There are
also studies on how the flow of body fluid influences the
mechanical properties of
bones. Lim and Hong (1994) obtained a uniaxial poroelastic model
to investigate the
-
CHAPTER 1. INTRODUCTION 9
mechanics of trabecular bone. Pore pressure and total stress in
response to the given
strain input were predicted to vary as a function of intrinsic
properties of trabecular
bone, such as permeability and Skempton’s coefficient. These
results suggest that the
body fluid in the bone may play a significant role in enhancing
the the mechanical
properties of trabecular bone. Mak et al. (1997) developed a
finite element model to
study the contributions from various hierarchical flow channels
in bone. Cortical bone
was modeled as a fully hydrated biphasic poroelastic material
with a superposing net-
work of one-dimensional channels radiating from the Haversian
canals. The model
was subjected to stress fields simulating uniform compression
and pure bending. The
effects of the interfacial permeability and the solid content
within the channels on the
drag forces in the channels were assessed. In solid mechanics,
the effects of fluid mo-
tion to solid structure have been studied by analyzing the
vibrations of fluid-saturated
poroelastic structures. Li et al. (1996, 1997) solved the
problems of transverse vibra-
tions of a poroelastic beam with axial fluid diffusion and of a
poroelastic plate with
fluid diffusion in the in-plane directions based on Biot’s
theory and Darcy’s law. The
results showed that the resonant frequencies of the
fluid-saturated structures increase
as the influence of fluid increases, and the amplitude response
is significantly reduced
as the fluid influence increases. In petroleum engineering, the
effect of elastic solid
motion on the permeability of saturated rock has been confirmed
in numerous labo-
ratory experiments. Many observations in the field also show
that seismic waves may
alter water and oil production. Beresnev (1994) reviewed the
methods and results of
elastic-wave stimulation of oil production. Nycal Royalty
Corporation (1997) investi-
gated the application of intermediate frequency waves for
stimulation of immobile oil
stocks. The mechanism of this phenomenon needs further testing
and understanding.
Homogenization techniques have been applied to acoustic studies
concerning poroelas-
ticity, the mechanics of porous elastic solids with fluid-filled
pores. These approaches
start with the detailed microstructure of the pores, the
linearized equations of elastic-
ity, and the linearized equations of fluid dynamics. Burridge
and Keller (1981) derived
equations which govern the linear macroscopic mechanical
behavior of a porous elas-
tic solid saturated with a compressible viscous fluid. Auriault,
Borne and Chambon
-
CHAPTER 1. INTRODUCTION 10
(1985) obtained a formulation similar to Biot’s results, and
presented some proper-
ties of the generalized Darcy coefficient and an experimental
checking. Kazi-Aoual,
Bonnet and Jouanna (1988) applied homogenization techniques to
study Green’s func-
tions in an infinite, transversely isotropic, saturated,
poroelastic medium for harmonic
displacements. Zobnin, Kudryavtsev and Parton (1988) developed
an equation de-
scribing the motion of a viscous fluid in a rigid porous medium
of periodic structure.
The procedure for averaging linearized hydrodynamic equations
with small viscosity
coefficients was used. Auriault (1991) proved that the
homogenization process is the
appropriate method to determine whether an equivalent
macroscopic description ex-
ists for a finely heterogeneous medium submitted to some
excitation. Auriault and
Boutin (1992, 1993, 1994) investigated the macroscopic
quasistatic description of a
deformable porous medium with a double porosity constituted by
pores and fractures.
It appeared that the macroscopic description is sensitive to the
ratios between the
characteristic lengths of the pores, the fractures and the
macroscopic medium. Ra-
soloarijaona and Auriault (1994) studied the nonlinear seepage
of a fluid through a
porous medium by using the homogenization theory for very small
Reynolds numbers
and performed an experimental and a numerical checking of the
results. Norris and
Grinfeld (1995) derived explicit motion equations for a medium
consisting of alter-
nating solid and fluid layers. The wave solutions were discussed
and compared with
other studies. Auriault and Lewandowska (1996) developed a
macroscopic model for
pollutant transport in a heterogeneous medium, including
diffusion, advection and
adsorption, and discussed the issue of the existence of
nonhomogenizable problems.
Lee et al. (1996) computed numerically the permeability and
dispersivities of solute
and heat for a periodic porous medium with geometry consisting
of a cubic array of
uniform Wigner-Seitz shape grains. Lenoach (1998) studied wave
propagation in a
random poroelastic medium and derived an upscaled
one-dimensional representation
of a reservoir with a heterogeneous fluid distribution
consisting of several pore fluids
with very different constitutive parameters. Some researchers
studied in detail the
microscopic problems for periodic structures. Saeger et al.
(1995) solved numerically
the Stokes equation system and Ohm’s law for fluid in periodic
bicontinuous porous
media of simple cubic, body-centered cubic and face-centered
cubic symmetry. The
-
CHAPTER 1. INTRODUCTION 11
dependence of Darcy permeability and conductivity on porosity
was investigated.
Skjetne (1995) presented analytical and numerical studies on
high-velocity flow in
spatially periodic media. The results showed that the most
pronounced inertial ef-
fect is the formation of narrow straight flow tubes impacting
with the pore walls at
obstructions. Lee and Mei (1997) derived nonlinear effective
poroelastic equations
at the macroscale applying the method of homogenization. The
matrix displacement
corresponding to the global strain was assumed to be comparable
to the granular
size. The constitutive coefficients of the nonlinear terms were
analyzed. The result
suggests that the linear effective equations may be adequate
even for practical prob-
lems involving moderate deformation or loading, as long as the
microscale geometry
is isotropic in the statistical sense.
The mathematical models introduced previously provide some
possible approaches
to improve the description of fluid flow in porous media. The
idea of applying har-
monic signals in well testing or as a possible new method for
enhanced oil recovery
appears promising, but needs yet to be fully understood and
further to be imple-
mented in reservoir characterization and production control.
This study focuses on
exploring the theoretical models generated from the
homogenization process in har-
monic testing to provide accurate description and
characterization of oil, gas and
groundwater reservoirs.
1.3 Outline of Approach
This study applies homogenization techniques to investigate the
two-scale problems
in reservoirs due to harmonic perturbations. The objectives
are:
• To study the sensitivity of pore structure parameters and the
properties of fluidand solid to harmonic signal frequencies. This
provides a method to choose
appropriate frequency ranges for diagnosis of different
properties.
• To investigate the mechanism of acceleration of fluid flow due
to harmonic per-turbation in pore geometry. If we know the
compatible porous media conditions
-
CHAPTER 1. INTRODUCTION 12
and the effective frequency range, a harmonic stimulation could
be performed
to increase production in reservoirs.
• To analyze different macroscopic behavior of porous media
corresponding tocertain parameter scales. It is important to apply
the appropriate macroscopic
model to interpret the reservoir responses in well testing,
seismic data processing
and other kinds of measurements.
• To obtain effective macroscopic parameters for the purposes of
reservoir charac-terization and large scale simulation, which may
lead to the prediction of future
production.
A porous medium with periodic pore structure (Figure 1.1) is
investigated in this
study. The ratio of the period to the overall size of the porous
medium, denoted by
�, is the small parameter of asymptotic analysis in the
procedure because the pore
size is usually much smaller than the characteristic length of
the reservoir. The wave
length λ of the excitation signal is associated with the
characteristic length L at the
macroscopic scale.
����������������������������������������������������������������
����������������������������������������������������������������
���������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������
���������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������
λ ~ L
L
l
~ l / Lε
P
S
Figure 1.1: A Periodic Porous Medium
This approach follows the procedure of homogenization (Figure
1.2). The input of
the system includes the unit cell geometry, two length scales,
the excitation signal
frequency, the pressure gradient and the basic properties of the
fluid and the solid in
the microscopic structure. The output will be the macroscopic
behavior of the porous
medium which includes the distributions of solid displacement,
fluid relative velocity
-
CHAPTER 1. INTRODUCTION 13
and pore pressure.
The principal steps are as follows:
1. For a given microscopic geometry, solve the periodic unit
cell problem under
harmonic perturbation to obtain the solid displacement field,
the fluid velocity
field and the pressure distribution in the unit cell.
2. Perform volume averaging in the unit cell to calculate all
the effective coefficients
that may appear in the macroscopic equations.
3. Study the sensitivity of pore scale parameters and the
effective macroscopic
coefficients to the excitation frequencies.
4. According to the specific physical problems, evaluate the
scale parameter � based
on the periodic unit cell geometry and the macroscopic scale,
and calculate the
value of the contrast number C = µω/a, where µ is fluid
viscosity, ω is signal
frequency, and a is the scale of pore pressure and solid
stress.
5. By asymptotic analysis, determine whether an equivalent
macroscopic descrip-
tion exists for the relation between � and C evaluated in the
previous step. If
an intrinsic description is not possible, using an average
theorem will lead to
a large-scale description depending directly upon the external
boundary condi-
tions of the medium.
6. In homogenizable situations, the macroscopic equations
independent of external
boundary conditions can be obtained. There are different
macroscopic models
corresponding to the scale of the contrast number.
7. Solve the macroscopic equations to describe the large-scale
behavior of the
porous medium.
The review of literature reveals some important and interesting
areas that were poorly
developed and hence are targets of this study. These areas
include:
-
CHAPTER 1. INTRODUCTION 14
, ,
K(ω)
fluid and solid propertiesperturbation: ω
Evaluation of
dp,
l, Lpore geometry,
Description ofMacroscopic Behavior
Model II
StudySensitivity
ModelMacroscopicSelection of
YesCoefficients
EffectiveCalculation of
Exists ?Model
MacroscopicIntrinsic
ProblemMacroscopic
ParticularSolving
NoAveragingVolume
Asymptotic Analysis
Cell ProblemSolving Unit
C, εScale Parameters
Figure 1.2: Homogenization Procedure
-
CHAPTER 1. INTRODUCTION 15
• At the microscopic scale, the fluid flow in channels of smooth
trajectory (notnecessarily straight) and of smooth width variation.
The channel walls (the
solid part) are elastic and are of random thickness.
• Study of sensitivity of porous medium properties to harmonic
signal frequencies.
• The mechanism of alternation of fluid flow rate due to
harmonic perturbation.
• Diagnosis of effective parameters from the porous medium
responses to har-monic excitation.
-
Chapter 2
Homogenization Techniques
The homogenization theory was first developed in mathematics to
deal with partial
differential equations with rapidly oscillating coefficients.
Many papers have been
devoted to the derivation of Darcy’s law by homogenization using
asymptotic ex-
pansions (Sanchez-Palencia 1980, Lions 1981). In recent years,
because of the rapid
mathematical development in this area, many basic phenomena are
now far better
understood. Several nonstandard models were developed including
Darcy’s law with
memory and nonlinear Darcy’s law. A survey on the method of
homogenization ap-
plied to problems in porous media was presented in the book
edited by Hornung
(1997). This chapter focuses on the derivation of some linear
macroscopic laws for
poroelastic porous media following the steps contributed by
Auriault in Hornung’s
book, but in much more detail. The mathematical convergence
proofs can be found
in the references, and will not be reproduced here. The purpose
of this chapter is to
introduce the general ideas of homogenization to people not very
familiar with the
techniques. Our nonlinear approach will be discussed in Chapter
3.
Assumptions
To derive the macroscopic quasistatic behavior, classically, the
inertial terms are
neglected. The material of the porous matrix is assumed elastic
and the saturating
fluid is assumed viscous, Newtonian and incompressible. The
local pore structure
16
-
CHAPTER 2. HOMOGENIZATION TECHNIQUES 17
is assumed periodic with a period Y of characteristic length l
(Figure 1.1). The
characteristic length L at the macroscopic scale is assumed to
be associated with the
excitation wave length λ: L = λ/2π. The two characteristic
lengths introduce two
dimensionless space variables:
y =X
l
x =X
L
X is the physical space variable. We denote the solid part of
the period S, the pores
P , and the interface Γ (Figure 2.1). An important small
parameter of scale separation
is defined as:
� =l
L� 1
and
x = �y
Different types of macroscopic behavior of the saturated porous
medium are studied
in relation to the value of a dimensionless number as a function
of the small parameter
� in the following sections.
ΓS
P
Figure 2.1: Periodic Unit Cell of the Porous Media
-
CHAPTER 2. HOMOGENIZATION TECHNIQUES 18
2.1 Acoustics of an Empty Porous Medium
For a porous medium filled with a very compressible and low
viscosity fluid, the fluid
can be ignored. This represents the most basic problem to be
used in all the following
investigations.
2.1.1 Local Description
The medium satisfies the Navier equation in the solid part.
∑j
∂σsij∂Xj
= ρs∂2usi∂t2
in S (2.1)
and the stress-strain relation is:
σsij = aijklekl(us) in S (2.2)
On the solid surface Γ, the normal stress is zero due to the
pore emptiness.
σsijνj = 0 on Γ (2.3)
νj is the unit normal vector of the surface pointing outside of
the solid.
The stress tensor can be resolved into normal components σii and
shear components
σij . For shear components, the first suffix denotes the
direction of the stress, and the
second suffix denotes the normal to the plane on which the
component is acting. For
normal stresses, only one suffix is needed since the direction
of component is the same
as that of the normal to the surface. The elastic tensor a
represents the stiffness of
the material. It is a periodic function of y, which
satisfies:
aijkl = ajikl = aijlk = aklij (2.4)
and the ellipticity condition:
aijkleijekl ≥ γeijeij , γ > 0. (2.5)
-
CHAPTER 2. HOMOGENIZATION TECHNIQUES 19
The strain tensor is related to the solid displacement by:
eij(u) =1
2(∂ui∂Xj
+∂uj∂Xi
) (2.6)
Equation 2.1 introduces a dimensionless number P :
P =|ρs ∂2usi∂t2 ||∂σsij∂Xj
| (2.7)
We consider a wave of pulsation ω with a wave length at the
order of the macroscopic
scale L, and we use the pore scale l as the characteristic
length to make the equations
dimensionless. We assume that all the components of elastic
tensor a are of the same
order of magnitude a, and the strain tensor e is small as the
order of O(l/l) = O(1),
then,
|∂σsij∂Xj
| ∼ al
(2.8)
|ρs∂2usi∂t2
| ∼ ρω2l (2.9)
We notice that the propagation velocity of a compression wave
(P-wave) can be
written as:
vp =
√a
ρ∼ ωL = ω λ
2π(2.10)
then,
Pl = O(ρω2l2
a) = O(
l2
L2) = O(�2) (2.11)
We define the dimensionless variables as:
yj =Xjl, u∗si =
usil, σ∗sij =
σsija, ω∗ =
ωL√a/ρ
, ρ∗s =ρsρ
(2.12)
then,
∂σsij∂Xj
=∂σ∗sij∂yj
(a
l)
ρs∂2usi∂t2
= −ω2ρsusi = −ω∗2ρ∗su∗si(al
L2)
-
CHAPTER 2. HOMOGENIZATION TECHNIQUES 20
Finally, the dimensionless form of the local description is
derived. The superscript *
is omitted without confusion. The subscript y shows operators
with respect to the
space variable y.
∑j
∂σsij∂yj
= −�2ω2ρsusi in S (2.13)
σsij = aijkleykl(us) in S (2.14)
σsijνj = 0 on Γ (2.15)
2.1.2 Macroscopic Description
We consider two-scale problems and we write the unknown us in
the form:
usi(x, y) = u(0)i (x, y) + �u
(1)i + �
2u(2)i (x, y) + · · · (2.16)
with x = �y. We substitute this expansion into Equations 2.13 -
2.15, and use the
following differentiation rule:
d
dy=
∂
∂y+ �
∂
∂x(2.17)
We look at the left hand side and the right hand side of
Equation 2.13 respectively.
d
dyj(∂uk∂yl
) = (∂
∂yj+ �
∂
∂xj)[∂u
(0)k
∂yl+ �(
∂u(1)k
∂yl+∂u
(0)k
∂xl) + �2(
∂u(2)k
∂yl+∂u
(1)k
∂xl) + · · ·]
=∂2u
(0)k
∂yj∂yl+ �[
∂
∂yj(∂u
(1)k
∂yl+∂u
(0)k
∂xl) +
∂2u(0)k
∂xj∂yl]
+ �2[∂
∂yj(∂u
(2)k
∂yl+∂u
(1)k
∂xl) +
∂
∂xj(∂u
(1)k
∂yl+∂u
(0)k
∂xl)] +O(�3)
dekldyj
=∂e
(0)ykl
∂yj+ �[
∂
∂yj(e
(1)ykl + e
(0)xkl) +
∂e(0)ykl
∂xj]
+ �2[∂
∂yj(e
(2)ykl + e
(1)xkl) +
∂
∂xj(e
(1)ykl + e
(0)xkl)] +O(�
3)
then,
L.H.S. =∂
∂yj(aijkle
(0)ykl) + �{
∂
∂yj[aijkl(e
(1)ykl + e
(0)xkl)] +
∂
∂xj(aijkle
(0)ykl)}
-
CHAPTER 2. HOMOGENIZATION TECHNIQUES 21
+ �2{ ∂∂yj
[aijkl(e(2)ykl + e
(1)xkl)] +
∂
∂xj[aijkl(e
(1)ykl + e
(0)xkl)]}+O(�3) (2.18)
R.H.S. = −�2ρsω2∞∑n=0
(�nu(n)i ) = −ρsω2
∞∑n=0
(�n+2u(n)i ) (2.19)
The boundary-value problem for the solid displacement at the
order of �0 becomes:
∑j
∂
∂yj[aijkleykl(u
(0))] = 0 in S (2.20)
aijkleykl(u(0))νj = 0 on Γ (2.21)
Because a and u(0) are Y periodic in y, the solution is an
arbitrary function of x,
independent of y.
u(0) = u(0)(x) (2.22)
The boundary-value problem for the solid displacement at the
order of �1 becomes:
∑j
∂
∂yj[aijkl(eykl(u
(1)) + exkl(u(0))] = 0 in S (2.23)
aijkl[eykl(u(1)) + exkl(u
(0))]νj = 0 on Γ (2.24)
Since u(1) is Y periodic in y, it can be expressed as a linear
vectorial function of
exkl(u(0)) to an arbitrary vector ū
(1)i (x) independent of y.
u(1)i = ξ
khi (y)exkh(u
(0)) + ū(1)i (x) (2.25)
where the third order tensor ξlmi (y) is taken of zero average
to insure the uniqueness.∫S
ξ dY = 0 (2.26)
Finally, the boundary-value problems for u(2) is obtained from
the order of �2.
∑j
(∂σ
(1)sij
∂yj+∂σ
(0)sij
∂xj) = −ρsω2u(0)i in S (2.27)
σ(1)sijνj = 0 on Γ (2.28)
-
CHAPTER 2. HOMOGENIZATION TECHNIQUES 22
where σ(0)s and σ(1)s are the zero- and first-order terms
respectively in the asymptotic
expansion of σs.
σ(0)sij = aijkh(eykh(u
(1)) + exkh(u(0))) (2.29)
σ(1)sij = aijkh(eykh(u
(2)) + exkh(u(1))) (2.30)
Equation 2.27 is a balance equation for the periodic quantity
σ(1)s with the source
terms −∂σsij/∂xj and −ρsω2u(0)i . The compatibility condition
for the existence ofu(2) can be obtained by integrating Equation
2.27 in S and using the boundary
condition 2.28 and the periodicity of σ(1).
∑j
∂〈σ(0)sij 〉s∂xj
= −ρeffs ω2u(0)i (2.31)
and ρeffs = 〈ρs〉s
The volume average is defined by:
〈•〉s = |Y |−1∫S• dY (2.32)
From Equation 2.25, we can get:
eylm(u(1)) = eylm(ξ
kh)exkh(u(0))
σ(0)sij = aijkhexkh(u
(0)) + aijlmeylm(ξkh)exkh(u
(0)) (2.33)
To obtain the macroscopic description, define the total stress
by:
σT =
σs in S0 in P (2.34)
and 〈σTij〉s = 〈σ(0)ij 〉s = aeffijkhexkh(u(0)) (2.35)with
aeffijkh = 〈aijkh + aijlmeylm(ξkh)〉s (2.36)
then,
∑j
∂〈σTij〉s∂xj
= −ρeffs ω2usi (2.37)
-
CHAPTER 2. HOMOGENIZATION TECHNIQUES 23
where usi = u(0)si .
Finally, by returning Equation 2.37 to dimensional quantities,
the macroscopic de-
scription (xj = Xj/L, u ∼ L) at the first order of approximation
is obtained. Wewill refer to this as Model I.
∑j
∂〈σTij〉s∂Xj
= ρeffs∂2usi∂t2
+Or(�) (2.38)
and 〈σTij〉s = aeffijkhekh(us) +Or(�) (2.39)It can be shown that
aeff is an elastic tensor. Therefore, the set of Equations 2.38
and 2.39 is a classical Navier description. The symbolOr(�)
shows that the description
is an approximation at the relative order �.
2.2 A Prior Estimate for Saturated Porous Media
To anticipate the different macroscopic description of the
acoustics of saturated porous
media, consider the volume balance of such media at the
macroscopic scale assuming
that the material constituting the porous matrix is
incompressible. Then,
∂θ
∂t= −∂vri
∂Xi(2.40)
where vr is the fluid velocity relative to the matrix and θ is
the volume change
percentage of the matrix. This introduces the Strouhal
number:
Sh =|∂θ∂t|
|∂vri∂Xi
| (2.41)
Obviously Sh = O(1) corresponds to diphasic behavior with
distinguishable macro-
scopic displacements of the matrix and the fluid. For Sh ≤ O(�)
the matrix dis-placement is negligible compared to the fluid
displacement. When Sh ≥ O(�−1) thefluid has a relative small
velocity compared to the matrix velocity. The macroscopic
behavior is monophasic.
Xj ∼ L, u ∼ L, ∂u∂t
∼ ωL (2.42)
|∂θ∂t| ∼ 1
L3∂u3
∂t∼ uω
L, |∂vri
∂Xi| ∼ vr
L(2.43)
-
CHAPTER 2. HOMOGENIZATION TECHNIQUES 24
then,
Sh = O(uω
vr) (2.44)
This expression of the Strouhal number can be generalized as
follows. The displace-
ment in the solid is related to the solid stress by:
u = O(σL
a) (2.45)
The relative fluid velocity is governed by a Darcy-like
relation,
vri = −kij ∂p∂Xj
(2.46)
The filtration tensor is of the order:
k = O(l2
µ) (2.47)
and µ is the fluid viscosity. From relations 2.46 and 2.47,
vr = O(l2p
µL) (2.48)
Then the Strouhal number becomes:
Sh = O(L2
l2σ
p
µω
a) (2.49)
Assuming that the stress in the matrix is at the same order as
the fluid pressure at
the macroscopic scale, the Strouhal number is simplified to:
Sh = O(�−2C), where C =µω
a. (2.50)
C is the property contrast number. Actually it is the ratio of
the characteristic
diffusion velocity in the fluid and the wave propagation
velocity in the solid.
C =µω
a=
(µ/ρω)ω2
a/ρ=
(δω)2
v2p=v2Dv2p
(2.51)
where δ =√
µρω
is the viscous skin length. This characteristic length
represents the
diffusion of the wave imposed on the fluid by an oscillating
impermeable boundary
wall.
From the physical definition of the Strouhal number, the
following cases will be in-
vestigated.
-
CHAPTER 2. HOMOGENIZATION TECHNIQUES 25
• Model II, C = O(�3): acoustics of a fluid in a rigid porous
matrix
• Model III, C = O(�2): diphasic macroscopic behavior
• Model IV, C = O(�): monophasic elastic macroscopic
behavior
• Model V, C = O(1): monophasic viscoelastic macroscopic
behavior
As the fluid viscosity or the perturbation frequency increases,
the scale of C increases,
and the macroscopic behavior of a saturated porous medium
changes from fluid mo-
tion dominated to elastic solid motion dominated.
2.3 Local Description of Saturated Porous Media
When the matrix is not rigid (Models III, IV and V), the solid
part satisfies the Navier
equation and the the fluid part satisfies the Navier-Stokes
equation. The interface
between the fluid and solid satisfies the continuity equations
of the normal stress and
the displacement.
∑j
∂σsij∂Xj
= ρs∂2usi∂t2
in S (2.52)
σsij = aijklekl(us) in S (2.53)∑j
∂σpij∂Xj
= ρp(∂vi∂t
+∑j
(vj∂
∂Xj)vi) in P (2.54)
σpij = 2µDij − pIij in P (2.55)∑i
∂vi∂Xi
= 0 in P (2.56)
usi = upi on Γ (2.57)
σsijνj = σpijνj on Γ (2.58)
For compressible Newtonian fluid, the tensor Dij is defined
as:
2µD = µ(∇v + (∇v)T ) + η(∇ · v)I (2.59)∇ · (2µD) = µ∇2v + (η +
µ)∇(∇ · v) (2.60)
-
CHAPTER 2. HOMOGENIZATION TECHNIQUES 26
where µ is the fluid shear viscosity coefficient and η is the
fluid bulk viscosity coeffi-
cient. For incompressible fluid (∇ · v = 0), the tensor
becomes:
Dij =1
2(∂vi∂Xj
+∂vj∂Xi
) (2.61)
Here we consider the fluid and solid movements at constant
pulsation. To neglect
the nonlinear terms in Equation 2.54, we assume that the
amplitude of the pertur-
bation is sufficiently small. For simplicity, we restrict the
problems to the following
assumptions:
• The displacements and densities are of the same order of
magnitude in the solidand the fluid parts respectively.
|us| ∼ |up|, ρs ∼ ρp (2.62)
• At the macroscopic scale (estimated with the characteristic
length L), The pres-sure in the fluid and the stress in the solid
are of the same order of magnitude.
Σ = |σsp|, ΣL = au
pL= O(1), and Σl =
au
pl= O(�−1) (2.63)
As shown in Section 2.1.1, the acoustic behavior of the matrix
at the pore scale
(X ∼ l) is governed by the ratio:
Pl =|ρsω2usi||∂σsij∂Xj
| = O(ρω2l2
a) = O(�2) (2.64)
Therefore, for the fluid at the pore scale,
Tl =|ρp ∂vi∂t || ∂p∂Xi| = O(
ρω2u
p/l) = O(
ρω2ul
au/L) = O(�) (2.65)
From the above estimates, there is only one free dimensionless
number that concerns
the acoustic behavior of the fluid.
Ql =|µ ∂2vi
∂Xj∂Xj|
| ∂p∂Xi| = O(
µωu
pl) = O(�−1
µω
a) (2.66)
The value of Ql is directly related to the property contrast
number C = µω/a. Let
Ql = O(�−r), then following the discussion in Section 2.2 leads
to the evaluation of
Ql for macroscopic Models III, IV and V.
-
CHAPTER 2. HOMOGENIZATION TECHNIQUES 27
• Model III, r = −1: diphasic macroscopic behavior
• Model IV, r = 0: monophasic macroscopic elastic behavior
• Model V, r = 1: monophasic macroscopic viscoelastic
behaviorDefine the following dimensionless variables:
yj =Xjl, ω∗ = ωL√
a/ρ, (2.67)
ρ∗s =ρsρ, u∗si =
usil, σ∗sij =
σsija
ρ∗p =ρpρ, v∗i =
vi
�√a/ρ, σ∗pij =
σpija�
p∗ =p
a�, µ∗ = µ
�−rl√aρ , k∗ =
k
µ
�−r√aρ
l
Finally, the dimensionless equations for pore scale local
description of a saturated
porous medium are obtained (* is omitted).
∑j
∂σsij∂yj
= −�2ρsω2usi in S (2.68)
σsij = aijkleykl(us) in S (2.69)∑j
∂σpij∂yj
= �ρpiωvi in P (2.70)
σpij = 2µ�−rDyij(v)− pIij in P (2.71)∑
i
∂vi∂yi
= 0 in P (2.72)
usi = upi on Γ (2.73)
(σsij − �σpij)νj = 0 on Γ (2.74)
2.4 Acoustics of a Fluid in a Rigid Porous Medium
When the matrix can be considered rigid (Model II, r = −1), the
fluid is in motionrelative to the matrix. The dimensionless
equations of fluid motion are those of Model
III, with the interface continuity condition 2.73 reduced to the
adherence condition.
∑j
∂σpij∂yj
= �ρpiωvi in P (2.75)
-
CHAPTER 2. HOMOGENIZATION TECHNIQUES 28
σpij = 2µ�Dyij(v)− pIij in P (2.76)∑i
∂vi∂yi
= 0 in P (2.77)
vi = 0 on Γ (2.78)
Introduce the asymptotic expansions for p and v in the form:
p(x, y) = p(0)(x, y) + �p(1)(x, y) + �2p(2)(x, y) + · · ·
(2.79)vi(x, y) = v
(0)i (x, y) + �v
(1)i (x, y) + �
2v(2)i (x, y) + · · · (2.80)
with x = �y. The p(i)’s and the v(i)’s are Y -periodic in y.
From Equations 2.75
and 2.76, we obtain:
∑j
{µ�[ ∂∂yj
(∂vi∂yj
+∂vj∂yi
)]} − ∂p∂yi
= �ρpiωvi (2.81)
Noticing that d/dy = ∂/∂y + �∂/∂x, Equation 2.81 becomes:
∑j
{ µ�[ ∂∂yj
(∂v
(0)i
∂yj+∂v
(0)j
∂yi) + · · ·]} − [∂p
(0)
∂yi+ �(
∂p(1)
∂yi+∂p(0)
∂xi) + · · ·]
= ρpiω�(v(0)i + �v
(1)i + · · ·) (2.82)
From Equation 2.82 at the order of �0, we obtain:
∂p(0)
∂yi= 0 (2.83)
then p(0) = p(0)(x) (2.84)
Equation 2.82 at the order of �1 and Equations 2.77 - 2.78 at
the order of �0 are:
∑j
µ∂2v
(0)i
∂y2j− ∂p
(0)
∂xi− ∂p
(1)
∂yi= ρpiωv
(0)i in P (2.85)
∑i
∂v(0)i
∂yi= 0 in P (2.86)
v(0)i = 0 on Γ (2.87)
Because of the linearity of Equations 2.85 - 2.87, v(0) can be
written in the form:
v(0)i = −kij
∂p(0)
∂xj(2.88)
-
CHAPTER 2. HOMOGENIZATION TECHNIQUES 29
where the tensor kij(x, y, ω) depends on ω and is
complex-valued. Finally from Equa-
tions 2.77 and 2.78 at the order of �1, we obtain:
∑i
(∂v
(1)i
∂yi+∂v
(0)i
∂xi) = 0 in P (2.89)
v(1)i = 0 on Γ (2.90)
Performing a volume average on P with respect to the variable y
on Equation 2.89
and applying the divergence theorem and the boundary condition
2.90, we can obtain
the compatibility condition.
∑i
∂〈v(0)i 〉p∂xi
= 0 (2.91)
〈v(0)i 〉p = −keffij∂p(0)
∂xj(2.92)
and keffij = 〈kij〉p (2.93)
where 〈•〉p is the volume average on P .
〈•〉p = |Y |−1∫P• dY (2.94)
Equation 2.92 represents a monochromatic seepage law with a
complex-valued and
ω-dependent filtration tensor keff . Returning to physical
quantities, Model II is
expressed by:
∑i
∂〈vi〉p∂xi
= Or(�) (2.95)
〈vi〉p = −keffij∂p
∂xj+Or(�) (2.96)
and keffij = 〈kij〉p (2.97)
2.5 Diphasic Macroscopic Behavior
The diphasic macroscopic behavior is described by Equations 2.68
- 2.74 with r = −1.We search for the solutions of the unknowns us,
v and p in the asymtotic expansion
-
CHAPTER 2. HOMOGENIZATION TECHNIQUES 30
form:
usi(x, y) = u(0)i (x, y) + �u
(1)i (x, y) + �
2u(2)i (x, y) + · · · (2.98)
vi(x, y) = v(0)i (x, y) + �v
(1)i (x, y) + �
2v(2)i (x, y) + · · · (2.99)
p(x, y) = p(0)(x, y) + �p(1)(x, y) + �2p(2)(x, y) + · · ·
(2.100)with x = �y. u(i),v(i) and p(i) are Y -periodic in y.
Introducing expansions 2.98 - 2.100
into Equations 2.68 - 2.74, and applying the relation d/dy =
∂/∂y+�∂/∂x, we obtain:
∑j
{ ∂∂yj
(aijkle(0)ykl) + �[
∂
∂yj(aijkl(e
(1)ykl + e
(0)xkl)) +
∂
∂xj(aijkle
(0)ykl)]
+ �2[∂
∂yj(aijkl(e
(2)ykl + e
(1)xkl)) +
∂
∂xj(aijkl(e
(1)ykl + e
(0)xkl))] + · · ·}
= −ρsω2�2(u(0)i + �u(1)i + · · ·) in S (2.101)∑j
{ µ�−r[ ∂∂yj
(∂v
(0)i
∂yj+∂v
(0)j
∂yi)
+ �(∂
∂yj((∂v
(1)i
∂yj+∂v
(1)j
∂yi) + (
∂v(0)i
∂xj+∂v
(0)j
∂xi)) +
∂
∂xj(∂v
(0)i
∂yj+∂v
(0)j
∂yi))
+ �2(∂
∂yj((∂v
(2)i
∂yj+∂v
(2)j
∂yi) + (
∂v(1)i
∂xj+∂v
(1)j
∂xi))
+∂
∂xj((∂v
(1)i
∂yj+∂v
(1)j
∂yi) + (
∂v(0)i
∂xj+∂v
(0)j
∂xi))) + · · ·] }
− [∂p(0)
∂yi+ �(
∂p(1)
∂yi+∂p(0)
∂xi) + · · ·]
= ρpiω�(v(0)i + �v
(1)i + · · ·) in P (2.102)∑
i
[∂v
(0)i
∂yi+ �(
∂v(1)i
∂yi+∂v
(0)i
∂xi) + · · ·] = 0 in P (2.103)
iω ( u(0)si + �u
(1)si + · · ·) = v(0)i + �v(1)i + · · · on Γ (2.104)
{aijkle(0)ykl + �[aijkl(e(1)ykl + e(0)xkl)] + �2[aijkl(e(2)ykl +
e(1)xkl)] + · · ·}νj
= {µ�1−r[(∂v(0)i
∂yj+∂v
(0)j
∂yi) + �((
∂v(1)i
∂yj+∂v
(1)j
∂yi) + (
∂v(0)i
∂xj+∂v
(0)j
∂xi))
+ �2((∂v
(2)i
∂yj+∂v
(2)j
∂yi) + (
∂v(1)i
∂xj+∂v
(1)j
∂xi)) + · · ·]
− �(p(0) + �p(1) + · · ·)}νi on Γ (2.105)
-
CHAPTER 2. HOMOGENIZATION TECHNIQUES 31
Equations 2.101 - 2.105 are used to solve the boundary-value
problems for Models
III, IV and V with different values of r. Here we consider the
case when r = −1corresponding to the diphasic behavior of the
matrix and the fluid.
From the equations at the order of �0, we can obtain u(0) and
p(0) similarly to those
in Sections 2.1 and 2.4 respectively.
u(0) = u(0)(x) (2.106)
p(0) = p(0)(x) (2.107)
From Equations 2.101 and 2.105 at the order of �1, the
boundary-value problem for
u(1) is obtained.
∑j
∂∂yj
[aijkh(eykh(u(1)) + exkh(u
(0)))] = 0 in S (2.108)
aijkh(eykh(u(1)) + exkh(u
(0)))νj = −p(0)νi on Γ (2.109)where u(1) is Y periodic in y.
When p(0) = 0, this set of equations is identical to
Equations 2.23 - 2.24 for empty pores. Because of the linearity,
u(1) can be expressed
in the form:
u(1)i = ξ
lmi exlm(u
(0))− ηip(0) + ū(1)i (x) (2.110)where ξ(y) is a third order
tensor introduced in Section 2.1 and η(y) is a vector. Both
are zero averaged. ∫S
ξ dY = 0,∫S
η dY = 0 (2.111)
From Equation 2.110 we can derive that:
eylm(u(1)) = eylm(ξ
kh)exkh(u(0))− eylm(η)p(0)
σ(1)sij = aijkh(eykh(u
(1)) + exkh(u(0))) (2.112)
= (aijkh + aijlmeylm(ξkh))exkh(u
(0))− aijkheykh(η)p(0)
The problem for p(1) and v(0) can be obtained from Equation
2.102 at the order of �1
and Equations 2.103 and 2.104 at the order of �0.
∑j
µ∂2v
(0)i
∂y2j− ∂p
(0)
∂xi− ∂p
(1)
∂yi= ρpiωv
(0)i in P (2.113)
-
CHAPTER 2. HOMOGENIZATION TECHNIQUES 32
∑i
∂v(0)i
∂yi= 0 in P (2.114)
v(0)i = iωu
(0)i on Γ (2.115)
v(0)i and p
(1) are Y -periodic with respect to y. To solve this problem,
set wi =
v(0)i − u̇(0)i , then,
∑j
µ∂2wi∂y2j
− ∂p(0)
∂xi− ∂p
(1)
∂yi= ρpiωwi − ρpω2u(0)i in P (2.116)
∑i
∂wi∂yi
= 0 in P (2.117)
wi = 0 on Γ (2.118)
wi and p(1) are Y -periodic in y. This is the linear
boundary-value problem in Sec-
tion 2.4 with ∂p(0)
∂xireplaced by ∂p
(0)
∂xi− ρpω2u(0)i . Therefore wi can be expressed as:
wi = v(0)i − iωu(0)i = −kij(
∂p(0)
∂xj− ρpω2u(0)i ) (2.119)
where k is defined in Section 2.4 as the complex-valued
permeability tensor. Inte-
grating Equation 2.119 over P , we obtain:
∫Pv
(0)i dY − iωu(0)i |YP | = −
∫PkijdY · (∂p
(0)
∂xj− ρpω2u(0)i ) (2.120)
Further averaging Equation 2.120 in the whole domain (|Y | = |YP
|+ |YS|), we obtain:
〈v(0)i 〉p − φiωu(0)i = −keffij (∂p(0)
∂xj− ρpω2u(0)i ) (2.121)
where 〈v(0)i 〉p = |Y |−1∫Pv
(0)i dY (2.122)
keffij = 〈kij〉p = |Y |−1∫PkijdY (2.123)
and φ = |Yp|/|Y | is the porosity.
Two compatibility conditions are required for the existence of
solutions of this prob-
lem. One compatibility condition is obtained from Equations
2.101, 2.105 at the order
-
CHAPTER 2. HOMOGENIZATION TECHNIQUES 33
of �2 and Equation 2.102 at the order of �1.
∑j
(∂σ
(2)sij
∂yj+∂σ
(1)sij
∂xj) = −ρsω2u(0)i in S (2.124)
∑j
∂σ(1)pij
∂yj− ∂p
(0)
∂xi= ρpiωv
(0)i in P (2.125)
σ(2)sijνj = σ
(1)pijνj on Γ (2.126)
where
σ(2)sij = aijkh(eykh(u
(2)) + exkh(u(1))) (2.127)
σ(1)sij = aijkh(eykh(u
(1)) + exkh(u(0))) (2.128)
σ(1)pij = µ(
∂v(0)i
∂yj+∂v
(0)j
∂yi)− p(1)Iij (2.129)
The total stress can be defined in the whole domain as:
σT =
σs in Sσp in P (2.130)
Integrating Equation 2.124 over S, Equation 2.125 over P , and
applying the diver-
gence theorem together with the periodicity condition, we
obtain:∮Γsσ(2)s · dΓ +∇x · (
∫Sσ(1)s dY ) = −ω2u(0)
∫SρsdY (2.131)∮
Γpσ(1)p · dΓ−∇xp(0)|YP | = iωρp
∫P
v(0)dY (2.132)
Noticing that dΓs = −dΓp and applying the interface boundary
condition 2.126, weobtain:
∇x · (∫Sσ(1)s dY )−∇xp(0)|YP | = −ω2u(0)
∫SρsdY + iωρp
∫P
v(0)dY (2.133)
Define the volume average in the whole domain as:
〈•〉s = |Y |−1∫S•dY and 〈•〉p = |Y |−1
∫P•dY (2.134)
Then we derive:
∇x · 〈σ(1)s 〉s − φ∇xp(0) = −〈ρs〉sω2u(0) + iωρp〈v(0)〉p
(2.135)
-
CHAPTER 2. HOMOGENIZATION TECHNIQUES 34
where φ = |Yp|/|Y | is the porosity. Taking into account the
solid stress-strain rela-tion 2.112, Equation 2.135 can be
rewritten as:
∑j
∂〈σT (0)ij 〉∂xj
= −ρeffs ω2u(0)i + ρpiω〈v(0)i 〉p (2.136)
〈σT (0)ij 〉 = aeffijkhexkh(u(0))− αeffij p(0) (2.137)
where the effective coefficients are defined by:
ρeffs = 〈ρs〉s (2.138)aeffijkh = 〈aijkh + aijlmeylm(ξkh)〉s
(2.139)αeffij = φIij + 〈aijkheykh(η)〉s = αeffji (2.140)
When the medium is isotropic, it can be verified that the value
of the effective tensor
αij has the following relation.
φ < α < 1. (2.141)
The second compatibility condition can be obtained from the
incompressibility, Equa-
tion 2.103 at the order of �1.
∑i
(∂v
(1)i
∂yi+∂v
(0)i
∂xi) = 0 (2.142)
Integrating Equation 2.142 over P and applying the divergence
theorem with the
periodicity and the interface continuity condition 2.104, we
obtain:
∇x ·∫P
v(0)dY = −∮Γp
v(1) · dΓ = iω∮Γs
u(1) · dΓ = iω∫S∇y · u(1)dY (2.143)
Performing a volume averaging on Equation 2.143 in the whole
domain Y , we have,
∇x · 〈v(0)〉p = iω〈∇y · u(1)〉s (2.144)
From Equation 2.110 we know that
∇y · u(1) = (∇y · ξ) · ex(u(0))− (∇y · η)p(0) (2.145)
-
CHAPTER 2. HOMOGENIZATION TECHNIQUES 35
Then Equation 2.144 becomes:
∇x · 〈v(0)〉p = iω〈∇y · ξ〉s · ex(u(0))− iω〈∇y · η〉sp(0)
(2.146)
Adding the term (−φiω∇x · u(0)) at both sides of Equation 2.146
and noticing that
∇x · u = I · (∇xu + (∇xu)T )/2 = I · ex(u) (2.147)
We derive the second compatibility equation.
∑l
∂
∂xl(〈v(0)l 〉p − φiωu(0)l ) = −γeffij iωexij(u(0))− βeff iωp(0)
(2.148)
where γeff is a effective tensor and βeff is a positive scalar
effective coefficient.
γeffij = φIij − 〈∑l
∂ξijl∂yl
〉s (2.149)
βeff = 〈∑l
∂ηl∂yl
〉s (2.150)
The compatibility relations 2.136 and 2.148 represent the
macroscopic behavior of the
acoustics of a saturated porous medium at the first order
approximation. Returning
to physical quantities, the macroscopic Model III is expressed
as:
∑j
∂〈σTij〉∂Xj
= ρeffs∂2usi∂t2
+ ρp∂〈vi〉p∂t
+Or(�) (2.151)
〈σTij〉 = aeffijkheXkh(us)− αeffij p+Or(�) (2.152)∑l
∂
∂Xl(〈vl〉p − φ∂usl
∂t) = −γeffij
∂eXij(us)
∂t− βeff ∂p
∂t+Or(�) (2.153)
〈vi〉p − φ∂usi∂t
= −keffij (∂p
∂Xj+ ρp
∂2usi∂t2
) +Or(�) (2.154)
where aeff ,αeff ,γeff and βeff are elastic effective
coefficients. The permeability
keff is complex-valued and ω-dependent. The model introduces two
macroscopic
displacement fields. It can be shown that this model has a
structure similar to Biot’s
theory.
-
CHAPTER 2. HOMOGENIZATION TECHNIQUES 36
2.6 Monophasic Elastic Macroscopic Behavior
When the fluid viscosity is bigger or the perturbation frequency
is higher, and the
contrast number C = µω/a is of order � (r = 0), the porous
medium is dominated by
monophasic behavior. The fluid stress tensor becomes:
σpij = 2µDyij(v)− pIij (2.155)
From Equation 2.101 and the interface condition 2.105 at the
order of �0, we can still
obtain that
u(0) = u(0)(x) (2.156)
The fluid boundary-value problem for v(0) is obtained from
Equations 2.102, 2.103
and 2.104 at the order of �0.
∑j
µ∂2v
(0)i
∂y2j− ∂p
(0)
∂yi= 0 in P (2.157)
∑i
∂v(0)i
∂yi= 0 in P (2.158)
v(0)i − iωu(0)i = 0 on Γ (2.159)
Obviously, the solution is:
v(0)i = iωu
(0)i (x) (2.160)
p(0) = p(0)(x) (2.161)
Similar to Section 2.5, u(1)i is obtained from Equation 2.101
and the interface condi-
tion 2.105 at the order of �1.
u(1)i = ξ
lmi exlm(u
(0))− ηip(0) + ū(1)i (x) (2.162)
The compatibility relation for the total momentum balance is
also unchanged.
∑j
∂〈σT (0)ij 〉∂xj
= −ρeffs ω2u(0)i + ρpiω〈v(0)i 〉p (2.163)
= −ρeffs ω2u(0)i − φρpω2u(0)i〈σT (0)ij 〉 = aeffijkhexkh(u(0))−
αeffij p(0) (2.164)
-
CHAPTER 2. HOMOGENIZATION TECHNIQUES 37
Using v(0)i = iωu
(0)i and integrating the compressibility equation 2.103 at the
order of
�1 over P , the second compatibility relation can be
obtained.
φ∇x · v(0) = φiω∇x · u(0) = iω〈∇y · u(1)〉s (2.165)φI · ex(u(0))
= 〈∇y · ξ〉s · ex(u(0))− 〈∇y · η〉sp(0) (2.166)
Then,
γeffij exij(u(0)) + βeffp(0) = 0 (2.167)
where the effective coefficients γeff and βeff are the same as
defined in Equa-
tions 2.149 and 2.150. Eliminating p(0) between Equations 2.164
and 2.167 leads
to the first order macroscopic description.
∑j
∂〈σT (0)ij 〉∂xj
= −(ρeffs + φρp)ω2u(0)i (2.168)
〈σT (0)ij 〉 = a∗effijkh exkh(u(0)) (2.169)
where a∗eff is an elastic tensor defined by:
a∗effijkh = aeffijkh + α
effij γ
effkh (β
eff)−1 (2.170)
Returning to the physical quantities, the macroscopic Model IV
can be expressed as:
∑j
∂〈σTij〉∂Xj
= (ρeffs + φρp)∂2usi∂t2
+Or(�) (2.171)
〈σTij〉 = a∗effijkh eXkh(us) +Or(�) (2.172)
At the macroscopic scale, the porous medium behaves like a
monophasic elastic
medium with the elastic tensor a∗eff and the density (ρeffs +
φρp).
2.7 Monophasic Viscoelastic Macroscopic Behav-
ior
For certain high viscosity fluid, the contrast number C = µω/a =
O(1) with r = 1,
there is no contrast between the mechanical properties of the
constituents in the pore
-
CHAPTER 2. HOMOGENIZATION TECHNIQUES 38
system. The fluid stress tensor then can be written as:
σpij = 2µ�−1Dyij(v)− pIij (2.173)
In this case, it is convenient to define the total displacement
in the porous medium
by:
u =
us in Sup in P (2.174)
The first boundary-value problem for u(0) is obtained from
Equation 2.102 at the
order of �−1 and Equations 2.101, 2.103, 2.104 and 2.105 at the
order of �0.∑j
∂
∂yj(aijkleykl(u
(0))) = 0 in S (2.175)
∑j
µiω∂2u
(0)i
∂y2j= 0 in P (2.176)
∑i
∂u(0)i
∂yi= 0 in P (2.177)
u(0)si = u
(0)pi on Γ (2.178)
aijkleykl(u(0)))νj − 2µiωeyij(u(0))νj = 0 on Γ (2.179)
where u(0) is Y -periodic with respect to y. The solution can be
expressed as:
u(0) = u(0)(x) (2.180)
The boundary-value problem for u(1) and p(0) is obtained from
Equation 2.102 at the
order of �0 and Equations 2.101, 2.103, 2.104 and 2.105 at the
order of �1.∑j
∂
∂yj[aijkl(eykl(u
(1)) + exkl(u(0)))] = 0 in S (2.181)
∑j
µiω∂2u
(1)i
∂y2j− ∂p
(0)
∂yi= 0 in P (2.182)
∑i
(∂u
(1)i
∂yi+∂u
(0)i
∂xi) = 0 in P (2.183)
u(1)si = u
(1)pi on Γ (2.184)
aijkl[eykl(u(1)) + exkl(u
(0))]νj
= [2µiω(eyij(u(1)) + exij(u
(0)))− p(0)Iij]νj on Γ (2.185)
-
CHAPTER 2. HOMOGENIZATION TECHNIQUES 39
From the linearity of the equations, the solutions are derived
in the following form.
u(1)i = χ
lmi exlm(u
(0)) + ū(1)i (x) (2.186)
p(0) = ζlmexlm(u(0)) (2.187)
Here ū(1)i (x) is an arbitrary vector independent of y. The
tensors χ(y, ω) and ζ(y, ω)
are complex-valued and depend on y and ω. Due to the stress
continuity on Γ at the
order of �0, there is no additive constant in the expression of
p(0).
The only compatibility relation is obtained from Equation 2.102
at the order of �1
and Equations 2.101 and 2.105 at the order of �2.
∑j
(∂σ
(2)sij
∂yj+∂σ
(1)sij
∂xj) = −ρsω2u(0)i in S (2.188)
∑j
(∂σ
(1)pij
∂yj+∂σ
(0)pij
∂xj) = −ρpω2u(0)i in P (2.189)
σ(2)sijνj = σ
(1)pijνj on Γ (2.190)
where the solid stress tensors are the same as defined in
Equations 2.127 and 2.128,
and the zero and first order terms in the asymptotic expansion
of the fluid stress
tensors are:
σ(1)pij = 2µiω[eyij(u
(2)) + exij(u(1))]− p(1)Iij (2.191)
σ(0)pij = 2µiω[eyij(u
(1)) + exij(u(0))]− p(0)Iij (2.192)
By integrating Equation 2.188 over S and Equation 2.189 P
respectively, and applying
the divergence theorem with the stress continuity condition
2.190 at the interface, we
obtain:
∇x · 〈σ(1)s 〉s +∇x · 〈σ(0)p 〉p = −〈ρs〉sω2u(0) − φρpω2u(0)
(2.193)Substituting the solutions of u
(1)i and p
(0) into stress-strain relations 2.128 and 2.192,
and then into Equation 2.193, the compatibility relation can be
obtained as:
∑j
∂〈σT (0)ij 〉∂xj
= −(ρeffs + φρp)ω2u(0)i (2.194)
〈σT (0)ij 〉 = a∗∗effijkh exkh(u(0)) (2.195)
-
CHAPTER 2. HOMOGENIZATION TECHNIQUES 40
where the effective coefficient is given by:
a∗∗effijkh = 〈aijkh + aijlmeylm(χkh)〉s + 〈2µiω(eyij(χkh) +
IijIkh)− ζijIkh〉p (2.196)
Returning to physical quantities, the monochromatic macroscopic
model is expressed
in the form:
∑j
∂〈σTij〉∂Xj
= −(ρeffs + φρp)ω2ui +Or(�) (2.197)
〈σTij〉 = a∗∗effijkh eXkh(u) +Or(�) (2.198)
The fourth-order tensor a∗∗eff is complex-valued and depends on
ω. The Model V
describes a monophasic viscoelastic medium at constant
pulsation.