Mechanics of deformation and acoustic propagation in

porous media
Maurice A. Biot

To cite this version:

Maurice A. Biot. Mechanics of deformation and acoustic propagation in porous media. Journal of
Applied Physics, American Institute of Physics, 1962, 33 (4), pp.1482-1498. �10.1063/1.1728759�.
�hal-01368725�

HAL Id: hal-01368725

https://hal.archives-ouvertes.fr/hal-01368725
Submitted on 21 Sep 2016

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
 Mechanics of Deformation and Acoustic Propagation in Porous Media
M. A. Biot

Citation: Journal of Applied Physics 33, 1482 (1962); doi: 10.1063/1.1728759

View online: http://dx.doi.org/10.1063/1.1728759
View Table of Contents: http://scitation.aip.org/content/aip/journal/jap/33/4?ver=pdfcov
Published by the AIP Publishing

Articles you may be interested in

Acoustic waves propagation along mechanically free surface of unconsolidated granular porous media
J. Acoust. Soc. Am. 123, 3035 (2008); 10.1121/1.2932694

Acoustic wave propagation through porous media: Theory and experiments

J. Acoust. Soc. Am. 102, 2495 (1997); 10.1121/1.420304

Acoustic wave propagation through porous media revisited

J. Acoust. Soc. Am. 100, 2949 (1996); 10.1121/1.417106

Theory of Acoustical Wave Propagation in Porous Media

J. Acoust. Soc. Am. 52, 247 (1972); 10.1121/1.1913086

Generalized Theory of Acoustic Propagation in Porous Dissipative Media

J. Acoust. Soc. Am. 34, 1254 (1962); 10.1121/1.1918315

Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 195.221.196.65 On:
Wed, 21 Sep 2016 10:06:13
 1482 KELSCH, KAMMERER, GOLAND, AND BUHL

like structures seen at high magnification being col- G. J. Dienes of the Solid State Physics Group, and to
lapsed or contracted "bubbles." express their thanks to Dr. D. H. Gurinsky, Head of
ACKNOWLEDGMENTS the Metallurgy Division, for his encouragement and
The authors wish to gratefully acknowledge the participation in this research, and to J. J. Floyd (BNL
stimulating assistance of Dr. G. H. Vineyard and Dr. reactor) for his cooperation.

JOURNAL OF APPLIED PHYSICS VOLUME 33. NUMBER 4 APRIL. 1962

Mechanics of Deformation and Acoustic Propagation in Porous Media

M. A. BlOT·
Shell Development Company, Houston, Texas
(Received August 18, 1961)

A unified treatment of the mechanics of deformation and acoustic propagation in porous media is
presented, and some new results and generalizations are derived. The writer's earlier theory of deformation
of porous media derived from general principles of nonequilibrium thermodynamics is applied. The fluid-solid
medium is treated as a complex physical-chemical system with resultant relaxation and viscoelastic
properties of a very general nature. Specific relaxation models are discussed, and the general applicability
of a correspondence principle is further emphasized. The theory of acoustic propagation is extended to
include anisotropic media, solid dissipation, and other relaxation effects. Some typical examples of sources
of dissipation other than fluid viscosity are considered.

1. INTRODUCTION the mechanics of porous media are formally the same

for an elastic or viscoelastic system, provided that the
T HE purpose of the present paper is to reformulate
in a more systematic manner and in a somewhat
more general context the linear mechanics of fluid-
elastic coefficients are replaced by the corresponding
operators.
saturated porous media and also to present some new Equations for acoustic propagation in the elastic
results and developments with particular emphasis on isotropic porous solid containing a viscous fluid were
viscoelastic properties and relaxation effects. established by adding suitable inertia terms in the
The theory finds numerous applications in a diversity original theory, and the propagation of three kinds of
of fields, including geophysics, seismology, civil engi- body waves was discussed in detail. 5 Two other publica-
neering, and acoustics. tions have dealt with general solutions and stress
A linear theory of deformation of a porous elastic functions in consolidation problems 6 and with a discus-
solid containing a viscous fluid was developed by this sion of the physical significance and of methods of
writer in 1941 and was applied to problems of consolida- measurement of the elastic coefficients. 7
tion of a foundation under a given load distribution.l,2 Consolidation theories have also been developed by
The medium was assumed to be statistically isotropic. FlorinS and Barenblatt and Krylov. 9 An acoustic
In later years, the theory was extended to include the propagation theory has been initiated by Frenkel/o who
most general case of anisotropy for a porous elastic brought out the existence of two dilatational waves.
solid. 3 The importance of viscoelasticity in consolidation
The theory of deformation in a porous viscoelastic problems of clay has been emphasized by Tan Tjong
medium was developed on the basis of the thermo- Kiel 1 and demonstrated in test results by Geuze and
dynamics of irreversible processes. 4 The results included Tan Tjong Kie.l 2 Special aspects of the consolidation
general anisotropy. It was shown that, on the basis of 6 M. A. Biot, J. Acoust. Soc. Am. 28, 168-178 (1956)' 28
Onsager's relations, it is possible to extend to a visco- 179-191 (1956). ' ,
6 M. A. Biot, J. AppJ. Mech. 23, 91-96 (1956).
elastic porous medium the principle of correspondence 7 M. A. Biot and D. G. Willis, J. AppJ. Mech 24 594-601
introduced in 1954 by the writer for homogeneous (1957). . ,
solids. The principle states that the equations governing 8 V. A. Florin, Theory of Soil Consolidation, in Russian (Stroy-
izdat, Moscow, 1948).
g G. 1. Barenblatt and A. P. Krylov, Izvest. Akad. Nauk
* Consultant, Shell Development Company. SSSR, Tech. Sci. Div. No.2, 5-13 (1955).
1 M. A. Biot, J. Appl. Phys. 12, 155-164 t1941)' 12 426--430 10 J. Frenkel, J. Phys. (U.S.S.R.) 8, 230 (1944).
(1941). ' , 11 Tan Tjong Kie, "Investigations on the rheological properties
2 M. A. Biot and F. M. Clingan, ]. AppJ. Phys. 12 578-581 of clay" (in Dutch with English synopsis), dissertation Technical
(1941); 13,35-40 (1942). ' University, Delft, Netherlands. '
3 M. A. Biot,]. AppJ. Phys. 26,182-185 (1955). 12 E. C. Geuze and Tan Tjong Kie, The Mechanical Behavior of
4 M. A. Biot, ]. AppJ. Phys. 27, 459-467 (1956). Clays (Academic Press Inc., New York, 1954).

Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 195.221.196.65 On:
Wed, 21 Sep 2016 10:06:13
 DEFORMATION AND ACOUSTIC PROPAGATION 1483

problem which do not agree with the elastic theory can Uij and U which are related to Tij by the equation14
be explained by the introduction of the more general Tij=Uij+OijU,
thermodynamic operators of the viscoelastic theory, 4
as further exemplified in the discussion of Secs. 5; 6, i=j,
O;j=l, (2.1)
and 7 of this paper. i,c j.
O;j=O,
Sections 2 and 3 begin with a general and rigorous If we consider a unit cube of bulk material, the compo-
derivation of the stress-strain relations, which is nents (j;j represent the force applied to the solid part of
valid for the case of an elastic porous medium with the faces, and (j represents the force applied to the fluid
nonuniJorm jJorosity, i.e., for which the porosity varies part of these faces. With p, denoting the fluid pressure,
from point to point. This re-emphasizes the use of the we can write
particular variables and coefficients introduced in the U=- JPI. (2.2)
original paper/,2 and attention is called to the fact that
some of the papers written in the later sequence are Since we arc dealing with a system which is in thermo-
formulated in the context of uniform porosity. dynamic equilibrium, the fluid is at rest and PI is
The derivation of Darcy's law from thermodynamic constant throughout the body.!"
principles, which has previously been briefly outlined,3,4 We can define the strain energy of a porous elastic
is discussed in detail in Sec. 4. medium as the isothermal free energy of the fluid-solid
system. W denotes the strain energy per unit volume.
In Secs. 5 and 6, the writer's previous thermodynamic
For a volume n bounded by a surface S, the variation
theory of viscoelastic properties of porous media is
discussed in more detail and with particular emphasis of the strain energy is equal to the virtual work of the
surface forces 16 :
on the physical significance of the operators.
Section 7 discusses the formulation of the field
equations and of their general solution in a particular
case. !l s
The theory of acoustic propagation developed +Fx/lU",+FyIlUy+F.fJU.)dS. (2.3)
previously for the isotropic elastic medium is extended
to anisotropic media in Sec. 8, and particular attention In the expression JidS and FidS, the components of the
is given to viscoelasticity and solid dissipation in Sec. 9. forces acting on the solid part and the fluid part of an
The term viscoelasticity here encompasses a vast element of surface dS are
range of physical phenomena, involving relaxation, J
which find their origin in physical-chemical, thermo- J.= 'E Uii n;,
(2.4)
elastic, electrical, mechanical, or other processes of the j
complex fluid-solid medium considered as a single Fi= 'E uo,jn;,
system. This generality is provided by the underlying
thermodynamic theory. It is also pointed out that by respectively, where n,
denotes the components of the
outward unit vector normal to the surface. We can
putting the fluid density equal to zero, we can apply
all results to thermoelastic propagation in a nonporous express these forces in terms of Tij and PI by introducing
elastic continuum. This is a consequence of the' iso- relations (2.1) and (2.2). Hence,
morphism between thermo elasticity and the theory j
of porous mediaY For similar reasons, the propagation Ji='E (Tij+fJii!PI)nj,
equations of Sec. 9 are also applicable to a thermovisco- (2.5)
elastic continuum. i
F,= 'E oii!p/nj.
2. STRAIN ENERGY OF A POROUS Introducing these expressions in Eq. (2.3) yields
ELASTIC MEDIUM
The displacement of the solid matrix is designated by
the components u"" U y, u.. The components of the
average fluid-displacement vector are U x, U1/, Uz ' These
components are defined in such a way that the volume - PI f f (nx/lw,,+nyllwy+n.ow.)dS. (2.6)
of fluid displaced through unit areas normal to the s
x, y, z directions are JU x, JU 1/, JU., respectively, where
14 See Eq. (25) of reference 7. In the original papers I we used
J denotes the porosity. the t?tal stress Tii and the fluid pressure PI represent~d by the
We shall use the total stress components of the bulk notations <Tii and <T.
15 Body forces are neglected in the present derivation.
material Tij. In earlier papers, we have used components • 18 The boundary S can be thought of not as a physical termina-
tion of the body but as any closed surface in the body. In this wav
~he surface tension at a physical boundary does not have to b~
13 M. A. Biot, J. App!. Phys. 27,240-253 (1956). mtroduced.

Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 195.221.196.65 On:
Wed, 21 Sep 2016 10:06:13
 1484 M. A. BlOT

From Eqs. (2.6), (2.9), (2.12), and (2.15), we derive

The vector Wi is defined as

or in vector notation
Wi= f(Ui-Ui), (2.7)
f !
U
!OWdn= !!f
!l
(TZ.,oe z+ Ty,fie y + Tzzoe.+ TyzO'Y.,

w=f(U-u). (2.8) + Tz.,o'Yv+T.,yO'Yz+pjOr)dn. (2.17)

Hence,
The vector w represents the flow of the fluid relative to
the solid but measured in terms of volume per unit oW = Txxoe x+ TyyOe y + Tzzoe.+ TyzO'Yz
area of the bulk medium. +Tz.,o'Yy+TzyO'Yz+PjOr. (2.18)
We consider now the surface integrals of Eq. (2.6).
The variable r was introduced by this writer in the
They can be transformed to volume integrals by means
original paper/ where it was designated by the symbol
of Green's theorem. We write
fJ. It was defined by the same general Eq. (2.16) as
valid for nonhomogeneous porosity. The same variable
f f t T,jnjOUidS = f f f t a~}T'jOUi)dn. (2.9) was also used in some later wotk 4- 6 in the context of a
medium with uniform porosity. For uniform porosity,
s u
we can write Eq. (2.16) as
The coordinates x, y, z are designated by Xi. The r= f div(u- U). (2.19)
integrand can be transformed as
This variable is obviously a measure of the amount of
fluid which has flowed in and out of a given element
attached to the solid frame, i.e., it represents the incre-
ment of fluid content.
Because the total stress field is in equilibrium, the The strain energy W must be a function of r and of
stress must satisfy the condition the six strain components defined by Eq. (2.13).
Hence, W is a function of seven variables:
(2.11) (2.20)
Hence we can write oW must be an exact differential. Hence,
ij a Tx:c=aW/aex, Tl/z=aWja'Yx
L -(TijOUi) = T.,.,oe.,+ Tl/yOel/+ Tzzoez+ TyzO'Y., Tyy=iJW/ae y, Tz:c=iJW/iJ'YY
aXj (2.21)
(2.12) Tzz=aW /ae z, Txy=aW/iJ'Yz
We have put
pt=iJw/ar·
e.,=au.,/ax, 'Y.,= (aUI//az) + (auz/ay)
These relations, obtained earlier by different meth-
el/=au,iay, 'YI/= (au./ax) + (aU.,/az) (2.13) ods,l·a.5.7 lead immediately to the formulation of the
e.=au./az, 'Y.= (aU.,/ay)+ (aul//ax). general stress-strain relations in a porous medium. Since
W is the isothermal free energy, the stress-strain rela-

The component, ,.-{i;. !:. t~:} (2.14)

tions (2.21) include phenomena which depend on the
physical chemistry of the fluid-solid system and others
which are expressible by means of thermodynamic vari-
21'" i'Y., e. ables such as interfacial and surface tension effects.
represent the strain tensor of the porous solid.
Similarly, the second surface integral of Eq. (2.6) 3. LINEAR STRESS-STRAIN RELATIONS
can be transformed to a volume integral: In an isotropic medium, the strain energy is a function
of four variables, the three invariants II, 1 2 , Ia of the
- f f (n.,ow",+nyOwl/+n.ow.)dS= f f foran, (2.15) strain components and the fluid content r:
s II W=W(I l ,I2,I a ,n. (3.1)

where We are restricted here to linear relations. Equations

(2.21) are valid for either linear or nonlinear properties.
r= -[(aw.,jax)+ (aw,,/ay) + (aw./az)] It is readily seen that expansion of W to the third degree
=div[j(u- U)], (2.16) leads to quadratic expressions for the stresses with
or eleven elastic constants. The case of nonlinear materials
r=-divw. shall be analyzed in a forthcoming publication. In the

Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 195.221.196.65 On:
Wed, 21 Sep 2016 10:06:13
 DEFORMATION AND ACOUSTIC PROPAGATION 1485

present analysis we shall consider only the linear Another useful form of the equations is obtained by
relations. using the so-called "effective stress," defined as
For a linear material, the strain energy is quadratic,
and we must include only the linear and quadratic (3.11)
invariantsl7 This represents the portion of the total stress which is
1 1 = e.,+ell+e.= e, in excess of the local fluid pressure.
(3.2) With these effective stresses, the relations (3.9)
12= elle.+e.e.,+e.,ell-i ('Y .,2+'Y1I2+'Y.2). become
In this case, it is easier to use the invariant T.,.,' - (1-a)PI= 2,ue.,+Xe
1 2'= -412='Y)+'Y1I2+'Y.2-4elle.-4e.e.,-4e.,ell, (3.3) Tilv' - (1-a)PI= 2,uell +Xe
T••' - (1-a)PI= 2,ue.+Xe (3.12)
instead of 1 2• We derive the quadratic form for W:
Til.' = .u'Y ." T..,' = ,u'YII
2W=He2+,uU-2Cer+Mr2. (3.4)
T.,/=,u'Y., r= (1/M)PI+ae.
The reason for using a negative constant - 2C and a
factor 2 is one of convenience in later equations. For an incompressible fluid and an incompressible
Substituting this expression in the general Eq. (2.21), matrix material, we have shown l •7 thata= 1 andM = 00.
we obtain the stress-strain relations In that case, the fluid pressure does not appear III
relations (3.12).
T"",,=He-2,u(e ll +e.)-Cr In abbreviated notation, Eqs. (3.12) are written
TIIII =He-2,u(e.+e.,)-Cr Til - (1-a)6ijPI= 2,ueij+6ijXe
T.. = He-2,u (e.,+ell)-Cr (3.5) (3.13)
r= (1/M)PI+ae.
TII.=,u'Y." T..,=,u'Y1i The interest in the use of the effective stress compo-
T"71=,u'Yz, PJ=-Ce+Mr. nents TJ lies in the experimental fact that slip and
By putting failure properties of porous and granular media are
H=A c+2,u, C=aM, Xc=A+a2M, (3.6) dependent primarily upon the magnitude of these
components alone. In this connection, a very useful
we see that relations (3.5) become viewpoint was introduced by Hubbert and Rubey,18
who pointed out that the average effective stresses can
T.,.,=2,ue.,+Ace- aMr
readily be determined by conditions of static equilib-
T1I1I = 2,ue1l +A ce-aMr rium of these forces with the total weight and the
T.. = 2,ue.+Ace-aMr (3.7) "buoyancy" associated with the field PI considered
continuous. This provides a simple and practical
procedure for the approximate analysis of failure in
PI= -aMe+Mr. porous media when the distribution of pore pressure is
considered.
In abbreviated notation, they are also written
The above equations and elastic coefficients were
Tij= 2,ue;j+6 ij(X ce-aMr) derived by a different procedure in previous work.
(3.8) The stress-strain relations in the form of Eq. (3.9)
PJ= -aMe+Mr·
were obtained in 1941, and a discussion of the signif-
We can obtain an alternate form of Eq. (3.7) by icance of elastic coefficients such as M and a was
substituting the value of r as a function of PI and e: presented. l In later work/· 8 the equations were re-
T.,.,+aPI=2,ue.,+Ae written in the form of Eqs. (3.7) and (3.12), and
methods of measurement of the elastic coefficients were
TIIII+apJ= 2,uel/+Ae further discussed.
T•• +aPI= 2,ue.+Ae (3.9) For a "closed system," i.e., a system in which the
pores are sealed, we must put r=O. Equation (3.7) then
shows that in this case,
T,,"=,u'Yz, r= (1/M)PI+ae.
(3.14)
In abbreviated notation, they are written
is the bulk modulus for a closed system. On the other
Tij+6ijaPJ= 2,ueij+6ijAe hand, by putting PI=O in Eq. (3.12), we see that for
(3.10) this "open system,"
r= (1/M)PI+ae.
17 A. E. H. Love, A Treatise on the Mathematical Theory of
(3.15)
Elasticity (Dover Publications, New York, 1944), 4th ed., pp. 18 M. King Hubbert and W. Rubey, Bull. Geol. Soc. Am. 70,
43,102. 115-166 (1959).

Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 195.221.196.65 On:
Wed, 21 Sep 2016 10:06:13
 1486 M. A. BlOT

represents a bulk modulus which is the inverse of the TABLE I. Equivalence of symbols.
"jacketed compressibility." This corresponds to a
test where a fluid pressure is applied to a jacketed Reference 1 (1941) Later publications
specimen, while the pore fluid is allowed to escape Q M
freely through a tube. a a
2Gv/(1-v) X and S
From the analogy between a porous medium and a G }J.and N
thermoelastic solid,13 we conclude that the coefficients 8 r
PI and - (,,/f)
he and h correspond to the adiabatic and isothermal "
Lame coefficients for a nonporous medium, respectively.
It is of interest to examine the restrictions on the coeffi-
cients imposed by the nonnegative character of the for the unjacketed test, the coefficient M is given by
strain energy W. We can write expression (3.4) m (3.26)
the form
For some types of materials, the coefficient 'Y can be
2W = K ce2- 2Cer+ Mr 2+tJL[ (e ll -e.)2+ (e.-e,,)2 expressed in terms of the porosity f and the fluid
+ (e,,-e ll )2]+JL('Y,,2+'Y,l+y.2). (3.16) compressibility c as
'Y= f(c-5). (3.27)
By putting e=r=O, we see that we must have
These coefficients for some sandstones have been
JL2::0. (3.17) measured by Fatt. 19 ,20
By putting Different coefficients have been used by this writer
in the past. As an aid to the reader, Table I shows the
and equivalence of various notations used in the earlier
work.
we are left with In some of the publications,3,5,6 stress-strain relations
(3.18) for uniform porosity were also written in the form
Uij= 2Ne;j+5ij (Ae+Qe)
This expression is never negative if (3.28)
u=Qe+Re,
K c2::0, M2::0, KcM -02::0. (3.19) with
e=divU=e- (11 f)r, fu= -PI' (3.29)
Thus, the four conditions (3.17) and (3.19) are both
necessary and sufficient. We note that These eq,uations can be written
KcM -0= (Kc-a2M)M. (3.20) Tij= uij+5;ju= 2Neii+5i;[(A +2Q+ R)e
- (Q+R)(rl f)] (3.30)
From Eqs. (3.14), (3.15), and (3.4), we derive
PI= - (Q+R)(elf)e+ (RI p)r.
Kc- K =Xc-X=a2M. (3.21)
Hence, Comparing these equations with Eq. (3.7), we derive
KcM-C2=KM. (3.22) relations between the two sets of coefficients:
N=JL, A=X+M(a- f)2,
Therefore, the necessary and sufficient conditions for
W to be nonnegative are that Q= f(a- f)M, R= pM. (3.31)
(3.23) The coefficient H of Eq. (3.5) can be written
H = A +2N+2Q+ R= Xc+2JL, (3.32)
The significance and methods of measurement of the
various coefficients have been discussed in detail which is the same as that used in the theory of acoustic
earlier.1,7 propagation. 6
It was shown in reference 7 that a can be measured For a material which is isotropic about the z axis, i.e.,
in two different ways, and that its value lies within transverse isotropic, the stress-strain relations become
the range f5.a5.1. We shall briefly recall some of the T"",=2Ble",+B2(e",+ell)+Bae.+B6r
more important equations relating to the coefficients.
The jacketed compressibility is T7I1I = 2B 1ell + B 2 (e,,+e ll )+ B 3e.+ Bet
(3.24) T•• = B 4e.+B 3 (e,,+e ll )+ B7t (3.33)
Til' = Bs'Y", T.,,=B 6'Y7I' T"II=B 1'Y.
The unjacketed compressibility 5 is
PI=B 6 (e,,+e ll )+ B 7e.+ Bst.
5= (l-a)K. (3.25) ----
19 I. Fatt, Bull. Am. Assoc. Petrol. Geologists 42 1924-1957
(1958). '
With a measurable "coefficient of fluid content" 'Y 20 I. Fatt, J. Appl. Mech. 26, 296-297 (1959).

Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 195.221.196.65 On:
Wed, 21 Sep 2016 10:06:13
 DEFORMATION AND ACOUSTIC PROPAGATION 1487

The eight elastic coefficients are chosen in order to proportional to the rate of entropy production. Per
constitute a symmetric matrix and satisfy at the same unit volume of bulk material, this dissipation function is
time the geometric symmetry.
For orthotropic symmetry, i.e., when the three
D=tTr (rate of entropy production), (4.2)
coordinate planes are planes of elastic symmetry, the where Tr= absolute temperature of the undisturbed
stress-strain relations become system. It can be written as a quadratic form with the
rate of volume flow as variables:
Tu = A lle x+ A 12ell+ A 13e.+ M lr
TII II = A 12ex+ A 22e + A 23e.+ M 2r
ll
2D= 1/(rllwi+r22w,,2+r33w.+ 2r23wyW,
T.. = A 13ex+ A 23ey+ A 33e.+ M 3r (3.34) +2r31w2w,,+2r12wxwII)' (4.3)
The viscosity of the fluid is denoted by 1/. 21
It is important to emphasize here that in applying
Pi = M lex+ M 2ey+ M 3e.+ Mr the thermodynamics of irreversible processes, we start
These equations contain thirteen elastic coefficients. with a thermodynamic system under conditions of
Finally, in the most general case of anisotropy, the equilibrium in the initial state. The initial state in
stress-strain relations are written this case is chosen to be one in which no pressure
gradients or gravity forces are acting on the fluid in
Txx
TYII
Au A12 Al3 Au
A22 A 23 A24
Al6
A 26
Al6
A 26
Mil
M2
ex
ell
the pores. The system is then perturbed by the applica-
tion of a disequilibrium force. This force must be
T.. A33 A34 A36 A36 e. expressed in a form which is conjugate to the volume
Ty. A44 A46 A46 M4 'Yx flow coordinate w. The components of this force are
T.x
Txy
symmetry A66 A66
A66
M'J
M6
M6
'Yy
'Y. (4.4)
PI M r where PI is the mass density of the fluid, and gi the
(3.35) components of the gravity acceleration. In vector form,
The matrix of twenty-eight coefficients Aij, M i , and we can write
M is symmetric about the main diagonal. The stress- (4.5)
strain relations (3.35) can be written in abbreviated
The quantity G is the gravitational potential per unit
notation by the introduction of quadruple indices for
mass. Because we are dealing here with a linear theory,
the coefficients Aij and double indices for the coefficients
we have assumed that the application of gravitational
Mi. We put
forces introduces density changes in the fluid which
(3.36) are "small of the first order." Therefore, the fluid
density PI in Eq. (4.4) can be put equal to the mass
With these coefficients, Eq. (3.35) takes the form density in the initial state. Under these circumstances,
)lV
the principle of superposition is applicable. Hubbert's
Tij= L A,j'"el'v+Mijr analysis22 of Darcy's law introduces a total fluid
(3.37) potential which, under the assumption just stated, can
ij be written
PI= L Mijeij+Mr. (4.6)
The various stress-strain relations discussed above for With this potential, we can write Eq. (4.4) in the form
anisotropic media were derived earlier in equivalent
form in the context of uniform porosity.3 X= -PI grad</>. (4.7)
Applying a general procedure used earlier by this
4. DARCY'S LAW AND ITS THERMODYNAMIC writer, we find that Onsager's principle in this case is
FOUNDATION equivalent to the relation
In the preceding sections, we have dealt with equilib-
(aDjaw x, aD/awII, aDjaw.) = X= -PI grad</>. (4.8)
rium phenomena or thermostatics. We shall now
examine an entirely different aspect of the problem-the In matrix form,
mechanics of flow through porous media. This brings
into play the thermodynamics of irreversible processes
and the Onsager relations. The rate of flow of the fluid - (Plj1/)[~::~;J=[;~~
a</>/az
;~: ;:~J [~:J.
r31 r23 r33 W.
(4.9)
is defined by the time derivative of the volume flow ----
vector: 21 Attention is called to the general character of the dissipation
function [Eq. (4.3)]. It is valid for slip flow or other more complex
awjat= (wx,w",w.). (4.1) interfacial effects and does not require that '1 be introduced
explicitly.
It is possible to write a dissipation function which is 22 M. King Hubbert, J. Geol. 48, 785-944 (1940).

Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 195.221.196.65 On:
Wed, 21 Sep 2016 10:06:13
 1488 M. A. BlOT

This form of the generalized Darcy's law has previously We need two constants {31 and {32 to define the first-
been derived 4 for the particular case f = const. order change of permeability in an initially isotropic
The symmetric matrix medium. The above equation need not be restricted to
the first order if {31 and ~2 are not considered as constants
but as functions of the volume change e:
(4.10)
(4.19)
represents a flow resistivity, whereas its inverse In this way, it seems possible to express a considerable
variety of porosity dependence by a close analysis of
the change of geometry of the pores.
(4.11) An equation such as (4.18) for anisotropic media can
easily be written, as previously shown. 4 In the case of
also symmetric, represents a "permeability matrix." transverse isotropy, for instance, such relations in~olve
Introducing the latter, we can write Eq. (4.9) as six coefficients.

5. THERMODYNAMICS OF VISCOELASTIC
(4.12) BEHAVIOR-THE CORRESPONDENCE
PRINCIPLE
General stress-strain relations for isotropic and
For the particular case of an isotropic medium,
anisotropic viscoelastic media have previously been
kn=k22=k33=k derived by this writer from the thermodynamics of
(4.13) irreversible processes.24 They were expressed in a form
k 12 =k 31 =k 2S =O,
which brings out the complete isomorphism between
and Eq. (4.12) becomes theories of elasticity and viscoelasticity. It follows from
this property that equations valid for the linear theory
ow!iJt= -k(j)f!Tf) gradcp, (4.14)
of elasticity (with linear boundary conditions and time
which is Darcy's law in the form expressed by Hubbert. 23 independent constraints) can immediately be extended
It can also be written to viscoelasticity by the substitution of time operators
for the elastic coefficients. In order to emphasize the
generality of this isomorphism, we have referred to it
as the correspondence principle and developed in more
The quantity k is the usual "coefficient of permeability"
detail its applications to various areas, such as the
of the medium. Clearly, the symmetric matrix [k.J
theory of plates, wave propagation, and dynamics. 25 ,26
represents a generalization of this coefficient.
Certain general theorems were also derived by combin-
For the case of isotropy, the dissipation function
ing the correspondence principle and variational
is given by
methods in operational form.
(4.16) The validity of the correspondence principle for
It is of interest to consider the possible relationship viscoelastic porous media is self-evident in this writer's
between the permeability and the deformation of the formulation of viscoelasticity for porous media. 4 This
porous medium. This aspect has been discussed earlier 4 formulation was based on the thermodynamics of
in connection with viscoelastic properties and in the irreversible processes.
less general context of homogeneous porosity. If the system is initially in thermodynamic equilib-
The porosity matrix represents a tensor analogous rium, we consider that the strain components, the
to a stress. If we start with a medium initially isotropic, stresses, and the change in fluid pressure represent
the permeability after deformation will be small deviations from that state of equilibrium. In a
great many cases, such deviations will be governed by
kij=k{)ii+t:.kij. (4.17) linear laws, and the Onsager reciprocity relations will
be valid.
The permeability increments to the first order will be
Several important points should be stressed regarding
related to the strain components by relations analogous
to the stress-strain relations as in an isotropic medium: the type of phenomena considered in this departure
from equilibrium in order to clarify what is meant here
t:.k ll = 2the"'+~2e by "viscoelasticity."
t:.k 22 = 2~le"+~2e M. A. Biot, J. Appl. Phys. 25, 1385-1391 (1954).
(4.18) 24

t:.kS3 = 2f31eZ+~~ 25 M. A. Biot, in Proceedings Fourth Mid-Western Conference

on Solid Mechanics, Purdue University, 1955, pp. 94-108.
t:.k23=~1'Yz, t:.k31=~1'Y", t:.k 12 ={31'Y•. 26 M. A. Biot, Deformation and Flow of Solids, IUTAM Col.
---- loquium, Madrid, 1955, edited by R. Gramme! (Springer-Verlag,
Berlin, 1956), pp. 251-263.
>3 Equation (80) of Hubbert's paper in reference 22.

Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 195.221.196.65 On:
Wed, 21 Sep 2016 10:06:13
 DEFORMATION AND ACOUSTIC PROPAGATION 1489

Weare considering only the local effect of the fluid elastic dissipation and that occurring in a homogeneous
pressure. The flow induced by the pressure gradient is solid. The latter is also included in the general thermo-
treated as a different phenomenon. This point was dynamic theory and depends essentially on the strain
discussed in Sec. 4 in connection with the thermo- gradient. The formulation of this case is quite different
dynamic derivation of Darcy's law and does not and was developed earlier.13 The viscoelastic effects
require further elaboration at present. might also result solely from certain physical properties
A second point is that, in formulating the viscoelastic of the fluid itself independent of any interaction with the
properties, we have assumed the hidden inertia forces to solid. Such is the case, for instance, for the propagation
be negligible. This excludes, for example, the inertia of sound in water containing solution of certain salts.
effects due to the motion of small particles representing Equilibrium concentrations of the various molecular
hidden coordinates. This assumption, however, is not species in solution are sometimes sensitive to fluid
essential, and we shall indicate below what modifica- pressure with an associated time lag and relaxation
tions must be introduced when such hidden inertia effect. These are but a few examples illustrating the
forces are taken into account. enormous range and variety of phenomena included in
A third point to be stressed refers to the extreme the present theory.
generality of the phenomena which are encompassed The viscoelastic and relaxation properties which we
by the term viscoelasticity. Its meaning in the present have just discussed are obtained by replacing the elastic
context far exceeds the narrow concepts of the purely coefficients by operators. Applying this correspondence
mechanical models usually associated with the word. principle to Eq. (3.5) for the isotropic medium, we
The derivation of the stress-strain relations from derive
thermodynamics is purely phenomenological. The two- Tzx =H*e-2J.1*(el/+ez)-c*r
phase fluid-solid aggregate is considered as a single
TYI/ = H*e- 2J.1* (ez+e x) - C*r
thermodynamic system. This is in contrast with. the
procedure of dealing with the "dry" solid and the Tzz =H*e-2J.1*(e x+el/)-C*r (5.1)
fluid as two separate entities, each with its own proper- TI/z = p,*'Yx, Tzx=p,*'YI/' TXI/=J.I*'Yz
ties. Such artificial separation is incorrect because of
Pf=-C*e+M*r.
the important role played by the surface forces at the
fluid-solid interface in the pores. In the case of gels, The operators are of the form
the interfacial surface tension contributes significantly
to the over-all rigidity, as pointed out many years ago '" H(r)
by this writer.l Because of the large area of contact of
fluid and solid, such interfacial effects should play an
important role in porous media. In general, they are
H*=p
1 --dr+H+pH'
o p+r

the result of certain equilibrium configurations of ions

and molecules which involve electrical fields and
physical-chemical interactions. Such configurations are (5.2)
represented by "hidden coordinates." When disturbed
from equilibrium, they evolve toward a new state
with a certain time delay or relaxation time. There may
'" M(r)
be a finite number or a continuous distribution of such
relaxation times, as represented by a relaxation spec-
trum. Such effects are included in the thermodynamic
M*=p
1 --dr+M+pM'.
o p+r

treatment of viscoelastic behavior developed by the The operator p is the time differential
writer. 4 Other effects involved here are exemplified by
the behavior of a crystal in equilibrium with its solution. p=d/dt. (5.3)
Under stress, this equilibrium is disturbed. Some areas The operational Eq. (5.1) can be considered as relating
of the crystal enter into solution, and precipitation Laplace transforms of stress and strain variable. For
occurs on others. The rate of deformations will depend instance, if these Laplace transforms are
not only on the stress but also on the rate of diffusion
in the solvent, giving rise to a relaxation spectrum.
Another type of phenomenon included here is the
thermoelastic relaxation. This is due to differential
(5.4)
temperatures arising in the solid and the fluid in the
pores when stress is applied. Because of the thermal
conductivity, such temperature differences tend to
even out, but with a certain time lag. This gives rise to
a thermoelastic relaxation spectrum. Attention is etc., the first Eq. (5.1) can be written
called to the difference between this type of thermo- £Txz=H*£e-2p,*£(e ll +e.)+C*£t (5.5)
Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 195.221.196.65 On:
Wed, 21 Sep 2016 10:06:13
 1490 M. A. BlOT

or tation. We shall consider, for example, the sixth of

Tu = H*e- 2p.* (ey+ e.) +C*r, (5.6) Eq. (5.1)
(6.1)
by simply dropping the symbol ..e.
lf the variables are harmonic functions of time, we which represents a response to pure shear. This relation
can write Txx and Txxe iwt , e",=exe iwt , etc., where Tn, ex, does not involve the pore pressure. It would govern,
etc., are complex amplitudes. Equation (5.1) then for example, the relation between torque and twist in a
represents the relations between the complex ampli- material such as clay. We assume the operator to be
tudes, when we put
P=iw (5.7) p.*=7Jrp/(p+r). (6.2)
in the operators. This shows that putting p=O yields In this case, we can write Eq. (6.1) as
stress-strain relations for very slow deformations,
'Y.= (1/7J r+ 1/7JP)T xv' (6.3)
whereas p= 00 corresponds to very fast deformations.
It is of interest to point out a difference between A constant stress of unit value applied at t=O is
the case of homogeneous and porous media with represented by
isotropic symmetry. As we have pointed out,27 the o t<O
use of the correspondence principle for homogeneous Txy=l(t)= {1 t>O. (6.4)
isotropic media is a consequence of the geometric
symmetry. On the other hand, for isotropic porous By the rules of the operational calculus, the correspond-
media, the symmetry of the operational matrix of ing strain as a function of time is
Eq. (5.1) is a consequence of Onsager's relations. Hence,
the correspondence principle for isotropic porous media 'Y.= (1/7Jr+t/7J) 1(t). (6.5)
invokes the laws of the thermodynamics of irreversible The quantity 7Jr is equivalent to an elastic shear
processes. The variable r in the operators represents the modulus, whereas 71 represents a viscosity. The time
relaxation constants of the hidden degrees of freedom. dependence of 'Y. is represented in Fig. 1. The response
We have shown from thermodynamics that they are is that of a so-called Maxwell material represented by a
real and positive,l·4 This, however, assumes that inertia mechanical model of a spring and dashpot in series
forces are not significant in the hidden degrees of (Fig. 2).
freedom. For extreme ranges of frequency, or in some Another type of behavior is illustrated by the operator
exceptional cases, this assumption may not be justified.
For example, a fluid may contain small air bubbles, p.*=ap', (6.6)
and resonance may occur with their natural frequency with O<s<1. We derive
of oscillation in the ultrasonic range. In such a case,
the hidden degrees of freedom represented by these air (6.7)
bubbles give rise to complex conjugate values for r.
For anisotropic media, the stress-strain relations are and for a constant stress (6.3) applied at t=O,
formally identical with those discussed in Sec. 3 for the 'Y.= [t'/ ar (1 +s)Jl (t). (6.8)
elastic case, except for the replacement of the elastic
coefficients by the operators. This was previously The typical time dependence is shown in Fig. 3 (curve
discussed in more detail. 4 By the correspondence a). It is easily shown that the operator P' is a particular
principle, the stress-strain relations (3.37) become case of the more general expression (5.2). From the
I'V

Tij=L Ai/!'>e!,>+Mi/r
(5.8)
1'j

Pt=L Mi/eij+M*r,
The operators A i /!'>, Mi/, and M* are of the same form
as Eq. (5.2), and they are governed by the same
symmetry relations (3.36) as the elastic case.
y
6. PHYSICAL SIGNIFICANCE OF THE OPERATORS
The physical significance of the operators can be
illustrated by some simple examples which bring out
the flexibility and generality of the operational represen-
27 M. A. Biot, Proceedings of the Third U. S. National Congress
on Applied Mechanics (American Society of Mechanical Engi- t
neers, New York, 1958), pp. 1-18. FIG. 1. Creep law represented by Eq. (6.5).

Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 195.221.196.65 On:
Wed, 21 Sep 2016 10:06:13
 DEFORMATION AND ACOUSTIC PROPAGATION 1491

neously throughout the body and thereby to eliminate

the additional time lag arising from the seepage of the
fluid flowing through the boundary. In principle, the
fluid must be thought of as being fed directly or
generated inside the pores. Another possibility is to
FIG. 2. Dashpot and spring in series, representing the operator
consider a test specimen which is small enough so that
(6.2) (Maxwell element). the time lag due to seepage is negligible relative to the
time constants of relations (6.13); whereas, on the
known definite integra12 8 other hand, its size is still sufficient in relation to the
pores for the statistical behavior to be valid.
This unjacketed test experiment was discussed
(6.9) earlier for the case of an elastic matrix, 7 and the
following relations were considered:
we derive
sins1I"
J.I*=ap 8 =a--
1 00
P
---r1J-1dr. (6.10)
e= -aPi
(6.14)
11" 0 p+r r='YPh
This corresponds to a spectral distribution29 where a is the unjacketed compressibility, and "1 is a
coefficient of fluid content. For a viscoelastic matrix,
sms1I" the coefficients aand "1 must be replaced by operators
J.I(r)=a-rs--l. (6.11) a* and "1*:
11"
e= -a*Pi
(6.15)
The sum of the Eqs. (6.5) and (6.8) yields a time- r='Y*pi'
dependent creep law of a very general nature which
covers both the short-range fast primary creep and the In some problems, it may be justified to neglect the
long-range steady-state secondary flow (Fig. 3, curve b). time lags of the operators a* and "1* and to assume that
We shall now turn our attention to the creep laws the relations (6.14) for the elastic matrix are applicable.
which involve the pore pressure. We consider the case We have also pointed out that if the matrix material
of isotropic stresses: is constituted by elements of an elastic isotropic solid,
the coefficient "1 can be expressed by relation (3.27), i.e.,
(6.12)
"1= j(e-a), (6.16)
We can then write
which introduces the fluid compressibility e and the
r= (H*-4J.1*/3)e-C*r porosity f. In this case, arepresents the compressibility
(6.13) of the matrix material itself. If we desire to take into
Pi= -C*e+M*r,
account the time lags in the compressibility of both
Under these conditions, we can perform a number of fluid and solid, we write
"thought experiments." For instance, we assume that a
stress r= -Pi and a fluid pressure Pi are applied "1*= f(e*-a*). (6.17)
suddenly at time zero, and we evaluate the values of e
and r as a function of time. This corresponds to the (b)
so-called unjacketed test previously discussed, when a
uniform hydrostatic pressure Pi is applied throughout
the solid matrix and the fluid in the pores. In the present
case of viscoelasticity, it is a thought experiment
because the fluid pressure is assumed to appear instanta-
28 Ch. J. de la Vallee Poussin, Cours d'Analyse Injinitesimale
(Gauthiers Villars, Paris, 1925), Vol. II, p. 75. (0)
29 As pointed out in an earlier paper,27 this is a particular case
of the problem represented by the integral equation

10(" p+r!(r)dr=F(p).
P

Its solu,tion was given by Fuoss and Kirkwood 80 and further

discussed by Gross.31 The solution is
f(r)= (1/27rri) lim [F(r+iE)-F(r-iE)].
r->O

80 R. M. Fuoss and J. G. Kirkwood, J. Am. Chem Soc 63

385-394 (1941). . . ,
al B. Gross, Mathematical Structure of the Theories of Viscoelasti- FIG. 3. (a) Creep law represented by Eg. (6.8); (b) Creep law
cities (Hermann & Cie, Paris, 1953). represented by the sum of Eqs. (6.5) and (6.8).

Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 195.221.196.65 On:
Wed, 21 Sep 2016 10:06:13
 1492 M. A. BlOT

From Eqs. (3.6), (3.25), and (3.26),

a*=1-bK*, (6.23)
M*= 1/('Y+b-b2K*), (6.24)
C*=a*M*= (1-bK*)/('Y+O-02K*). (6.25)

Also from Eqs. (3.6) and (3.15),

X*=K*-2J.L*/3
H*=X*+2J.L*+a*2M*. (6.26)
FIG. 4. Spring dashpot mo~el r.epre~enting the interaction ?f Hence,
elasticity of the solid and flUId vIscosity around areas of gram H*=4.u*/3+ (K*+'Y-O)/[K*('Y+O)-02]. (6.27)
contact.
The operators in this case can be expressed in tenus
As noted above even in the fluid alone, time lags of only two operators /10* and K*. They can be reduced
may occur betwe~n pressure and ,,:olume. A P?ss.ible to the spectral form [Eq. (5.2)J by application of the
cause of such effects is the relaxatIOn of eqUIlIbnum Fuoss-Kirkwood method or by expansion in partial
concentrations of various chemical species in solution fractions.
in the fluid. This phenomenon, sometimes referred to as We have already pointed out that the viscoelasticity
bulk viscosity, will be reflected in the time. constant operators contain properties which are representative of
contained in the operator c*. The effect of aIr bubbles solid viscosity, but which depend also on fluid-solid
on damping in the fluid can also be included in the interaction. In such cases, the properties of the solid
operator c* (see Sec. 5). We should also point out an matrix cannot be considered separately from those of
earlier remark that the resonance of such bubbles at a the fluid. One such type of interaction which is purely
certain frequency can be taken into account by includ- mechanical in nature is illustrated by Fig. 4. When two
ing the inertia forces of the hidden coordinates. In this elastic bodies are in contact and are surrounded by a
case, the correspondence is still valid, but the form of viscous fluid, a force applied in a direction normal to
the operators is different. Similar considerations apply the area of contact will tend to squeeze the flow away
to the inclusion of bulk viscosity and relaxation in the from this area. Because of fluid viscosity, the fluid will
solid matrix material and can be expressed by a suitable not move away instantaneously. A time delay, which is
operator. exemplified by the equivalent spring dashpot model of
Another thought experiment which can be considered Fig. 4, will be involved. This model corresponds to an
corresponds to the jacketed compressibility. In this operator of the form
case, the pore pressure is assumed to be zero, and Eq.
(6.13) is written (6.28)
7'= (H*-4J.L*/3)e-C*r Actually, of course, a more accurate evaluation of the
(6.18)
0= -C*e+M*r, interaction of fluid displacement and elasticity would
yield a somewhat more complex model or a more
Eliminating r, we find sophisticated operator with a spectral distribution. Such
7'=K*e, (6.19)
effect may be more pronounced for grains containing
with cracks or separated by narrow gaps (Fig. 5).
K*=H*-4J.L*/3- (C*2/M*). (6.20)
We have previously discussed this type of visco-
The jacketed compressibility operator is elasticity in connection with the effect of wall sponginess
and micropores. 4 A similar mechanism can also be
K*=l/K*. (6.21) responsible for the viscoelasticity represented by the
The stress T is applied to a jacketed specimen, while the operator J.L* for pure shear and can be evaluated in
pore pressure is maintained at zero thr~ughout br some similar fashion.
device which connects the pores dIrectly WIth an
outside region of zero pressure. 7. FIELD EQUATIONS FOR DEFORMATION AND
By the correspondence principle, it is possible to STRESS DISTRIBUTION
express the operators in terms of. the coeffi~ients If we include the gravity force pg. (p=mass density
'Y *, b*, and K* with the formulas whIch are valId for of bulk material), the total stress field must satisfy
the elastic case. the equilibrium equations
In particular, it will often be justified to approximate
the operators 'Y* and b* by elastic coefficients:
(7.1)
'Y*='Y, b*=b. (6.22)

Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 195.221.196.65 On:
Wed, 21 Sep 2016 10:06:13
 DEFORMATION AND ACOUSTIC PROPAGATION 1493

case, Eqs. (7.7) become

,uV'2u+ (,u+hc) grade-aM gradr=O
(7.8)
aw/at= (k/1/)Ma grade- (kM/1/) gradr.
These equations can be written in other equivalent
forms previously derived1 ,6 by the application of the
divergence operator to the second Eq. (7.8). We derive
J.LV'2 U +(J.L+h c) grade-aM gradr=O
(7.9)
arjat= (kM/1/),'V 2r- (kMa/1/)V'2e.
Application of the divergence operator to the first
Eq. (7.9) yields
FIG. 5. Example of general grain geometry with viscoelastic
behavior of the type idealized in Fig. 4. (2J.L+h c)V2e-aMV'2r=0. (7.10)
Hence, we can further transform Eqs. (7.9) to
and Darcy's law for the isotropic medium is written J.LV'2U + (J.L+h c) grade-aM gradr=O
(7.11)
aw/at= - (k/1/) gradpf+Pf gradG. (7.2) BNBt= (k/1/)Mc'V2t,
Under the assumption of linearity, as pointed out in with a constant Me defined as
Sec. 4, the principle of superposition is applicable. Mc=M(2,u+h c-a2M)/(2J.L+h c) (7.12)
Hence, the general solution that we are seeking is the or
superposition of a particular solution due to gravity M c= M (2J.L+h)/ (2J.L+hc)' (7.12)
forces alone and of another due to other causes. Under
these restrictive assumptions, we can neglect the General solutions of these equations were also derived
gravity forces in the present theory and thereby isolate and discussed earlier. 6 We shall derive them hereafter
that part of the problem dealing with deformations in somewhat simpler form.
resulting from causes other than gravity. A convenient form of the general solution can be
Equations (7.1) and (7.2) are therefore simplified to obtained with Eq. (7.8). The second equation indicates
that w must be a gradient. We can write without loss
of generality
(7.3) w=gradcp
and (7.13)
and
u=ul-[aM/(2,u+h c)] gradcp.
aw/at= - (k/'TJ )gradpl' (7.4)
Substituting these expressions into Eq. (7.8), we obtain
For an elastic medium with bulk isotropy, the
stress-strain relations (3.10) are J.LV'2 U1 +(J.L+h c) gradel=O

Tij= 2,ueii+6,;(hce-aMr) acp kMc kMa)

grad ( ----V'2cp---el =0.
(7.14)
(7.5)
PI= -aMe+Mr. at 1/ 'TJ
We have put
In these equations, we have defined rand e as
(7.15)
r= -divw and e=divu, (7.6) The first of these equations is the classical form of
where u is the solid displacement. Lame's equations of the theory of elasticity. The
general Papkovich-Boussinesq solution of these equa-
Substituting the values of Tij and PI [Eq. (7.S)J
into the Eqs. (7.3) and (7.4), we derive tions is well known:
U1 = grad(,po+r· \h)-[2(2J.L+h c)/CJ.L+h c)]th, (7.16)
ij a a
2 L: -(J.Lei;)+-(X,e-aMt)=O where the scalar ,po and the vector tlt1 are general
ax; ax, (7.7) solutions of Laplace's equation
Bw/Bt= (k/1/) grad(aMe-Mr)=O. (7.17)
The vector r is
These are six equations'for the six components of the r= (x,y,z). (7.18)
unknown vector fields u and w.
We shall examine the particular case where the Note that we can write
coefficients J.L, "Xc, a, M, and k/'TJ are constants. In this el= -[2.u./CJ.L+hc)J div\h, Vel=O. (7.19)

Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 195.221.196.65 On:
Wed, 21 Sep 2016 10:06:13
 1494 M. A. BlOT

We consider now the second Eq. (7.14). It implies that In deriving these equations from relations (7.9), we
have taken into account the identities
a<{J kMc kMa
__ -'V'2<{J=--el+C(t), (7.20) 'A.*+J.I.*=H*-J.l.* and a*M*=C*. (7.29)
at ~ ~
As shown in earlier work,4 the diffusion Eq. (7.22) is
where C(t) is a function of time independent of the converted into a generalized operational form
coordinates. We put (7.30)

t[kMa ] where the operator is

<(J=lf+
J -~-el+C(t) dt. (7.21)
M.*= (1/H*) (H*M*-C*2). (7.31)

Substituting in Eq. (7.20), we find The general solution corresponding to Eqs. (7.23) and
(7.26) have also been derived in operational form.4
(7.22) In the notation of the present paper, the general
solutions (7.23) and (7.26) for the viscoelastic case
The general solution for w becomes assume the operational form
2k C*J.I.* 1
(7.23) w= gradlf-- grad divth
~ H*-J.l.* P
(7.32)
Applying the divergence operator in this expression, 2H* C*
we derive u= grad(lfo+r· th) - - - t h - - gradlf,
H*-J.f.* H*
(7.24)
The solution for u is and .I is given by the same Eq. (7.24).
The functions lfO and the vector ~l satisfy Laplace's
equation, whereas If is a solution of the generalized
diffusion equation
(7.33)
Because 'V'2el = 0, the time integral satisfies Laplace's
equation. This term can therefore be absorbed in the Note that lfO and ~l are generally functions not only of
function lfO contained in the expression for UI, Eq. (7.16). the coordinates x, y, Z, but also of the operator p. This
Hence, we can write the general solution will generally be introduced by the boundary conditions
which are also operational relations.
The boundary conditions are easily introduced in
the solution of specific problems. Total stresses or
pore pressures can be specified at certain boundaries.
The condition that a boundary be impervious is
In these expressions, If satisfies the diffusion Eq. (7.22),
introduced by putting the normal component of w at
whereas lfO and lfl are solutions of Laplace's equation.
that boundary equal to zero. Also, conditions at the
For anisotropic media, the field equations are
interface of solids of different properties are expressed
by requiring that the total stresses, fluid pressure, and
solid displacement be continuous at this boundary;
whereas for w, the condition of continuity applies only
(7.27) to the normal component. Consolidation problems
based on the above results for elastic and viscoelastic
media have been treated by this writer1.4 and others.32

For materials with viscoelastic properties, we have 8. ACOUSTIC PROPAGATION IN ISOTROPIC

shown that the field equations are obtained from the AND ANISOTROPIC MEDIA
correspondence principle. 4 Equations (7.27) are im- Equations for acoustic propagation in a porous
mediately extended to viscoelastic media by replacing elastic isotropic solid containing a viscous fluid have
the elastic coefficients by operators. For example, been developed by this writer. 5 They were obtained
Eq. (7.9) for the isotropic medium with uniform by adding the inertia terms to the consolidation theory.
properties becomes A detailed discussion was given of the propagation of
body waves. These equations in slightly different form
J.I.*'V'2u+(H*-J.l.*) grade-C* gradt=O
(7.28) 32 G. Paria, J. Math. and Phys. 36, 338-346 (1958); G. Paria,
(~/k) (at/at) = M*'V'2r-C*'V'2e. Bull. Calcutta Math. Soc. 50, 71-76, 169-179 (1958).

Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 195.221.196.65 On:
Wed, 21 Sep 2016 10:06:13
 DEFORMATION AND ACOUSTIC PROPAGATION 1495

are discussed below and extended to the anisotropic The notation Wj designates aWj/at. The coefficients aij
medium with an elastic matrix. depend on the coordinates in the pores and the pore
Their further extension to a medium with viscoelastic geometry.
and solid dissipation properties is outlined in the next The kinetic energy of a unit volume of bulk material
section. is given by
Attention is called to the immediate applicability
of the results presented here to acoustic propagation in
a thermoelastic continuum. When the fluid density PI
is put equal to zero, the equations become identical to
T=!Pl(Ux2+Uy2+UZ2)+!PI III
!l
[(u,,+v,y

those of thermoelasticity, and w plays the role of the

entropy displacement vector. This follows from the
analogy between thermoelasticity and the properties of
porous media derived earlier,13 In this expression, PI represents the mass density of the
The following six equations in vectorial form 5 were fluid, and PI the mass of solid in the unit volume of
obtained for the six components of the displacements bulk material.
u and V: The volume integral is extended to the total volume
Q of fluid in the pores. We consider the term
NV 2u+grad[(A +N)e+QEJ
iJ2 a
=-(PllU+P 12 V)+b-(u- V)
iW at
!PI III
!l
(u,,+v,,)2dQ
(8.1)
a2
grad(Qe+RE)=-(P12U+p22V)-b-(u- V).
at 2 at
a
=!PI III !l
(u"2+2u,,v,,+V,,2)dQ. (8.5)

The density parameters Pll, P12, and P22 were discussed We can write
in the quoted paper. 5 The coefficient b is
b= ('I)/k)P, (8.2) "2PI
1 III' U" 2dO-
u - "2P2U"
1 • 2, (8.6)
where k is the permeability coefficient of Eq. (4.14), !l
'I)is the fluid viscosity, and f is the porosity. The where
elastic coefficients N, A, Q, R can be expressed in terms (8.7)
of JI., A, a, M by the relations (3.31). Methods of
measurement were analyzed in reference 7. represents the mass of fluid per unit volume. Also,
Expression (8.2) for b is valid for the low-frequency
range, where the flow in the pores is of the Poiseuille
type. For higher frequencies, a correction factor is
applied to the viscosity, replacing it by 'l)F, where F is
PI III
!l
u"v"dQ= PIU" III
!l
v"dQ= p/u"w". (8.8)

a complex function of the frequency which has been Hence,

evaluated. 5 Strictly speaking, a similar correction must
apply to the density parameters Pll, P12, and P22 to take
into account the departure of the microvelocity field
from Poiseuille flow as the frequency increases.
~2PI 111(' U" + V" )2dO_
1 . 2+ . .
.'-"2P2U" PIU"W"
!l
Equations (8.1) have recently been applied to an
analysis of the effect of a discontinuity surface on the
propagation. 33 •34
Using the variable w instead of V, we shall discuss
+!PI IIIo
v,,2dQ. (8.9)

here a system of equations which are equivalent to

Eq. (8.1) for the case of uniform porosity. Expression (8.4) now becomes
We shall first consider the low-frequency range. In
this case, the components of the relative microvelocity
field in the pores are determined by the vector w:
V",= allw",+a12WI/+a13w. +!PI fff (v,,2+ vll+ v.2)dQ, (8.10)
Vg = a21w",+a22wlI+a23w• (8.3) !l

v.= a31W",+a32WII+a33W•. where

33 J. Geertsma and D. C. Smit, Geophysics 26, 169-181 (1961). (8.11)
H. Deresiewicz, Bull. Seismol. Soc. Am. 50, 599-607 (1960)'
34
51,51-59 (1961). ' is the total mass of bulk material per unit volume.

Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 195.221.196.65 On:
Wed, 21 Sep 2016 10:06:13
 1496 M. A. BlOT

From relations (8.3), we derive momentum, and the second can be interpreted as
expressing the dynamics of relative motion of the fluid

PI fff (v,,?+V y2+v. 2)dQ= t mi,wiwh (8.12)

in a frame of reference moving with the solid. As stated
in Sec. 7, this is a legitimate procedure under the
assumption that the principle of superposition is
!l

with applicable. Extension of the theory under less restrictive

mij=PI ff f ct akiakj)dQ. (8.13)

assumption will be carried out in a later paper.
For the isotropic medium, we substitute expressions
(8.16) and (4.16) for T and D and the stress-strain
II
relations (3.7). Equations (8.23) become
Note the reciprocal property j a a a2
(8.14) 2 L -(J,te,J+-(Ace-aMr}=-(pUi+PfW,)
a~ a~ ap
For a medium with statistical isotropy of the micro- (8.24)
velocity field, the coefficients mij reduce t0 35
mij=mOij (8.15)
and
For constant values of the parameters, these equations
T=!p(U,,?+Uy2+uz2)+P/(U.,W.,+UyWy+u.wz) can be written
+!m(w.,2+wi+w}).(8.16) a 2

J,tV'2 U +(J,t+A c) grade-aM grad~=-(pu+Plw)

In order to compare with the mass parameters pu, at 2
(8.2S)
Pl2, and P22 used in previous work, 5 we express the a2 1J aw
kinetic energy III terms of the variables u and U. grad(aMe-M~)=-(plu+mw)+- - .
Substi tu ting at 2 k at
Wi= f(Ui-ui). (8.17)
If we multiply the second equation by f and then
In the value (8.16) of T, we obtain subtract it from the first, we obtain Eq. (8.1), in which
T= !pu(u.,2+Uy2+U.2)+P12(U.,U .,+uyU y+u.Uz ) the coefficients and the mass parameters are given by
expressions (3.31) and (8.19) derived above. Putting
+!P22(U.,2+Ui+Ul), (8.18)
with (8.26)
pu=p-2pfi+mj2, P22=mp, P12=pfi-mp. (8.19) and using the constant H = Ac+ 2J,t, we obtain the equa-
We put tions of propagation of dilatational waves:
(8.20)
V'2(Hcf>l+aMcf>2) = (a 2jat2) (pcf>1+p!cf>2)
where P. is the density of the solid matrix. The quantities V'2(aMcf>1+Mcf>2) (8.27)
Pl and P2 are the masses of solid and fluid, respectively,
= (a 2jat2) (p!cf>1+mcf>2) + ('1j k) (acf>2jat).
per unit volume of bulk material. Also writing
Pa=mf-pfi=mf-P2, (8.21) When the Laplacian operator is applied, these equations
can also be written in the form
we see that the mass parameters become
V'2(He-aM~)= (a2jat2)(pe-P/~)
PU=Pl+Pa, P22=P2+Pa, P12= -Pa' (8.22)
V'2( -aMe+M~) (8.28)
This coincides with the relations in the quoted paper.5 = (a 2jat2) ( -Ple+m~)+ (1Jjk)(a~jat).
If we now look at the forces applied to a unit volume
of bulk material and consider Ui, Wi as generalized They can also be written with the constant C=aM. A
coordinates, we can apply Lagrange's equations. They condition of "dynamic compatibility" for which a
are written wave propagation is possible without relative motion
between fluid and solid has previously been derived. 5
j aT;j _ d (aT) We can readily obtain it in equivalent and simpler
L---- , form by putting tP2=0 in Eq. (8.27):
aXj dt au;
These are the general dynamical equations when the (aMjH) = (pljp). (8.29)
gravity forces are neglected. The first of these equations By putting
could have been derived directly from the linear (8.30)
36 Equation (8.15) is also valid for cubic symmetry. we derive the equations of propagation, of rotational

Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 195.221.196.65 On:
Wed, 21 Sep 2016 10:06:13
 DEFORMATION AND ACOUSTIC PROPAGATION 1497

waves The function F has been evaluated numerically.5 A

,uV'2tb= (a 2/at2) (ptb+Pfth), similar approximation can be introduced for the case
(8.31) of anisotropy by the replacement of rij by
- (T// k) (at¥.2/at) = (a 2/at 2) (Pft¥.1+mt¥.2).
(8.36)
The propagation of the dilatational and rotational
waves has also been previously analyzed in detail. 5 As a further refinement, the mass coefficient m for the
Equations of propagation in anisotropic media are case of isotropy can be replaced by a complex quantity
immediately derived from the results established above.
From Eqs. (8.3) and (8.10), we write for the kinetic (8.37)
energy the more general expression [see Eq. (8.13)J
and for the anisotropic medium the tensor mij can be
T= !p( u,,?+Uy2 +Uz2)+Pf(uxwx+UyWy+uzwz) replaced by
(8.38)
(8.32)
The nature of the operators ri/ and mil will be
The dissipation function D in this general case is given discussed in more detail in a forthcoming publication.
by Eq. (4.3):
iJ 9. WAVE PROPAGATION WITH INTERNAL
D= h L: ri/wiWj. (8.33) DISSIPATION IN THE SOLID

We must also use for the stress components Tij and Pf The incorporation of internal solid dissipation in the
the general .stress-strain relations (3.37). Introducing theory of wave propagation can be accomplished
expressions (8.32), (8.33), and (3.37) into the dynamical without further development by the replacement of
Eq. (8.23), we derive the elastic coefficients by suitable operators.
As shown in an earlier paper4 and in the more detailed
discussion of Sec. 6, this procedure, because of its
thermodynamic foundation, actually takes into account
a wide variety of dissipative effects which are not
restricted to the solid alone, but which are the result of
(8.34) complex interaction between fluid and solid of mechan-
ical, electrical, chemical, or thermoelastic origin.
For the case of isotropy, when we introduce operators
in Eq. (8.24), we find that these equations become

These six equations for the unknown vector components j a a a2

2 L: -(,u*eij)+-(>'c*e-c*n=-(pU.+PfWi)
Ui and Wi govern the propagation of waves in the general aXj aXi at 2
case of anisotropy. The various cases of higher sym- (9.1)
metry, such as orthotropy and transverse isotropy, are
easily derived as particular cases by the introduction
of the various stress-strain relations discussed in Sec. 3.
We shall now consider the higher frequency range. The operator >./ is
As the frequency increases, a boundary layer develops
where the microvelocities are out of phase. At higher (9.2)
frequency, this boundary layer becomes very thin. The
In these equations, we have also replaced m by m* and
viscous forces are then confined to this layer, and the
T/ by T/* in order to introduce the frequency dependence
microvelocity field in the major portion of the fluid is
determined by potential flow. Similar considerations of these coefficients in the higher frequency range, in
apply to the viscous forces, as shown by earlier analysis accordance with the discussion of the previous section.
Similarly, for anisotropic media, the propagation Eqs.
on simple models. 5 The friction force of the fluid on the
(8.34) are written
solid becomes out of phase with the relative rate of
flow and exhibits a frequency dependence represented
by a complex quantity.
There are several ways of approximating these
effects. One approximation used by the writer for the
case of isotropic media is the replacement of the
a ij
(9.3)
--(L: M,/eij+M*r)
viscosity T/ by ax;
(8.35)
where F is a complex function of the frequency p=iw.

Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 195.221.196.65 On:
Wed, 21 Sep 2016 10:06:13
 1498 M. A. BlOT

Again, we have introduced the frequency dependent The variation of w' with frequency is very slow.
tensor ri/ as valid for the higher frequency range. Therefore, the imaginary part of ,u* represents a damp-
In a further refinement, mij is also replaced by mi/. ing which varies very little within a relatively large
Equations for dilatational waves in the isotropic range of frequency.
medium of uniform properties are Another type of operator which exhibits the same
property was proposed by this writer in an earlier
V2(H*e-C*r) = (iJ2jap) (pe-Plr)
V2( -C*r+M*r) (9.4)
paper.27 We write
'" p ,ul
, = (iJ2/at2)( -PI+m*r)+ ('I]*/k) (iJr/at).
The physical significance of these equations can be
,u*=
f
c
- - -dr+,u.
p+r r
(9.10)

illustrated by various types of dissipation. Consider, This amounts to introducing for ,u* m the general
for instance, a purely elastic solid. In the discussion expression (5.2) a relaxation spectrum
of Sec. 6, we have shown that, in this case, a dissipation
can occur because of the presence of the fluid in minute
(r)=J,ul/r for r>f (9.11)
cracks or in their regions adjacent to the areas of ,u LO for O<r< f.
contact between the grains (Figs. 4 and 5). As we have
seen [Eq. (6.27)], this effect is represented by the use of Performing the integration and putting p=iw yields
an operator of the type
(9.5)
,u*= ,ul[1r/2- tan-1 (f/w)]i
+,u 1 log(1+w2/e 2)l+,u. (9.12)
or more generally
(9.6) When E is sufficiently small, the imaginary part becomes,
in effect, frequency-independent over a large range.
for the jacketed compressibility. The coefficient KO
Operators such as (9.7) and (9.10) can therefore be
represents the elastic compressibility due to the elastic
used to represent some of the typical features of
grains, and Kr](P) is the dissipative term corresponding
internal friction in solids. Operators other than ,u*
to the viscoelastic effects associated with the squeezing
which may involve solid friction properties can, of
of the fluid in the small, cracklike volumes surrounding
course, be represented by similar expressions.
the areas of contact. This operator K* is introduced in
the expressions (6.24), (6.25), and (6.26), which in We should remember that the operational equations
turn are used in the propagation Eq. (9.4). remain valid if the operators are of a form more general
As another example, we shall take the case of internal than Eq. (5.2) and incorporate terms which involve
dissipation in the solid itself. Such a case can be the inertia and resonance effects of hidden degrees of
represented by the operator freedom, as previously discussed for the particular
example of air bubbles.
(9.7) Attention is also called to the applicability of the
wi th 0 <s < 1 (already considered above). We shall propagation Eq. (9.3) to the problem of thermovisco-
assume that s is small and put P=iw. Then, elasticity in a continuum, provided that we identify w
with the entropy displacement and put the fluid
,u*=aw'[cos(s1r/2)+i sin (s1r/2)]. (9.8) density PI equal to zero. We are referring here to the
Since s is small, we can write approximately dynamics of a homogeneous solid (without pores) in
which viscoelastic and thermoelastic dissipation occur
,u*=aw'[1 + (i1r/2)s]. (9.9) simul taneously.

Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 195.221.196.65 On:
Wed, 21 Sep 2016 10:06:13