Geophys. J. Int.

(2002) 151, 944–949

RESEARCH NOTE

The hydroelectric problem of porous rocks: thermodynamic

approach and introduction of a percolation threshold

André Revil
CNRS-CEREGE, Department of Hydrogeophysics and Porous Media, BP 80, Cedex 4, 13545 Aix-en-Provence, France. E-mail: [email protected]

Downloaded from https://academic.oup.com/gji/article-abstract/151/3/944/623281 by guest on 08 November 2018

Accepted 2002 June 11. Received 2002 May 16; in original form 2002 January 7

SUMMARY
An electrical field is produced in response to groundwater flow in porous materials such as
soils and permeable rocks. This electrical field is due to the relative displacement between the
charged mineral grains and the pore water, which drags the excess of electrical charge located
in the close vicinity of the pore water/mineral interface in the so-called electrical double layer.
In this note, I take the hydroelectric problem back to its thermodynamic roots by showing
how the hydroelectric equations can be derived from the Gibbs-Duhem equation. In addition,
I suggest that the introduction of a percolation porosity may improve the description of the
material properties of granular porous materials entering the coupled hydroelectric problem at
the macro-scale. Comparison between the proposed model and a set of laboratory data available
from the literature are in agreement for a reasonable choice of specific parameters.
Key words: electrical transport, percolation, permeability, porous medium, streaming
potential.

this electric field is high enough to be recorded at the ground sur-

1 INTRODUCTION
face using non-polarisable electrodes (e.g. Revil et al. 2002). This
The study of the groundwater flow in the subsoil is a very difficult ‘hydro-electric’ conversion represents therefore an appealing signal
problem due to the very high range covered by the hydraulic proper- to geophysicists to study the pattern of groundwater flow in the sub-
ties of porous soils and rocks. For example, permeability of sand clay surface of the Earth (Birch 1998; Ishido & Pritchett 1999; Trique
mixtures varies over 12 orders of magnitudes (e.g. Revil & Cathles et al. 1999). The development of new methods to invert the electri-
1999). There is clearly a lack of methods available to determine cal field at the ground surface in order to determine the hydraulic
the spatial variations of the key hydraulic parameters or to visualize source at depth (that Revil et al. in press, termed ‘electrography’)
the pattern of groundwater flow in the subsurface of the Earth. The could represent a breakthrough in this direction.
classical use of piezometers to monitor the water head has several In this paper, I first discuss the thermodynamic roots of the elec-
drawbacks including (1) the fact that the hydraulic head is obtained trokinetic coupling in the context of ‘generalized’ linear thermody-
only at discrete locations, (2) the existence of hydraulic perturba- namics. I then discuss the origin of the equality between the two
tions associated with the use of piezometers, and (3) the cost and coupling terms arising into the hydroelectric problem. In Sections 3
time associated with the installation of a set of piezometers. There to 6, I show that the introduction of a percolation threshold in the
is also a lack of efficient geophysical tools available to measure at transport properties (electrical conductivity, permeability, and elec-
distance (without perturbation) the pattern of fluid pressure distri- trokinetic coupling coefficient) improves the predictive capabilities
bution in the subsoil associated for example with a pumping test in of the equations developed on the basis of a differential effective
a borehole. medium model, which, by construction, does not possess such a
A new method could arise from the study of the electrical field percolation threshold.
generated by the flow of the groundwater itself. This electrical field is
due to the relative displacement between the charged mineral grains
2 THERMODYNAMIC BACKGROUND
and the pore water, which drags the excess of electrical charge lo-
cated in the close vicinity of the pore water/mineral interface into The system under consideration corresponds to a water-saturated
the so-called electrical double layer (e.g. Revil & Leroy 2001). This porous plug (e.g. a cylindrical jacketed sample like use for perme-
phenomenon produces a net current density, which serves as source ability measurements). If equilibrium is perturbed by an increase of
term in the Maxwell equations. This effect is known as the streaming pore fluid pressure at one of its boundary, the porous body behaves
potential and is one of the so-called electrokinetic effects. In addition as an open system exchanging energy, entropy, and matter with its

944 
C 2002 RAS
 Hydroelectric coupling 945

surrounding environment. One of the pillars of ‘generalized’ linear tion equation for entropy. In isothermal conditions, the dissipation
thermodynamics is the existence of a local equilibrium assump- function of the system D ≡ T  ≥ 0 is given by:
tion, which is characterized by the use of the classic Gibbs-Duhem
equation. Indeed Prigogine (1949) showed that the Gibbs-Duhem D = −j(+) ∇ µ̃(+) − j(−) ∇ µ̃(−) − jw ∇µw . (11)
equation remains valid in the vicinity of equilibrium to the Dissipation is zero at thermodynamic equilibrium. In linear ther-
first-order of perturbation of the state variables. Neglecting defor- modynamics, the choice of thermodynamic forces and fluxes is
mation of the porous plug (e.g. poro-elastic effects) and assuming arbitrary to a certain extent as long as the product of any pair of
that the pore water is composed by a multicomponent (N species) conjugated flows and thermodynamic forces have the dimension of
electrolyte (N-1 ionic species plus the solvent), this yields: an entropy-production rate and the sum of all products leave the en-

N tropy production rate invariant. In addition needless to say that the
du = T ds + µ̃i dn i , (1) material fluxes and scalar potentials are not uniquely defined and
i=1 only the divergence of a flux and the gradient of a scalar potential
where µ̃i represents the electrochemical potential of the species have a physical meaning. This usually means that for experimen-
i, n i represents the concentration (in mol m−3 ) of species i per unit

Downloaded from https://academic.oup.com/gji/article-abstract/151/3/944/623281 by guest on 08 November 2018

tal measurements, the choice of a reference state and a Lagrangian
volume of the porous plug, s and u represents the entropy and internal frame of reference represents crucial points.
energy per unit volume of the porous plug, and T is the temperature Rather than using the ionic and solvent fluxes, geophysicists usu-
(in K). For a 1:1 electrolyte saturating the pore space (e.g. NaCl or ally use the electrical current density j (in A m−2 ) and Darcy filtration
KCl), this yields: velocity u (in m s−1 ) as independent fluxes in the hydroelectric prob-
ds du dn (+) dn (−) dn w lem (e.g. Ishido & Pritchett 1999). The Darcy velocity is here defined
T = − µ̃(+) − µ̃(−) − µw , (2) as the volume flow of solvent flowing per unit surface area and per
dt dt dt dt dt
unit time. In absence of macroscopic ionic concentration gradients,
where t is time (in s), d/dt represents the substantial time derivative,
this yields u = w jw and j = e(j(+) − j(−) ), where w is the molar
n (±) and n w represent the concentrations of salt ions and water per
volume of the solvent. The electrochemical potential of the ions and
unit volume of the porous plug, respectively, µw represents the total
the chemical potential for the water phase are µ̃(±) = (±)eψ and
potential of water, and µ̃(±) = µ(±) ± eψ, represents the electro-
µw = w ( p−ρ f gz), respectively, where g is the gravity acceleration
chemical potentials of the salt ions (µ(±) represents the chemical
(in m s−2 ) z is a vertical coordinate measured positively downward,
potentials, e is the elementary charge, 1.19 × 10−19 C, the sign ±1
and where I have neglected the concentration dependent component
depends on the sign of the charge carried by the ion, and ψ is the
of the chemical potential of the pore water and therefore the osmotic
electric potential in V). The equations of continuity are:
pressure. Under these assumptions, the dissipation function is now
du given by,
= −∇ · w, (3)
dt
D = −j∇ψ − u(∇ p − ρ f g). (12)
dn (±)
= −∇ · j(±) , (4) where g = gz, z being the unit vector along the vertical axis and
dt
directed downward. The two terms of eq. (12) correspond to the
dn w Joule and viscous dissipations of energy.
= −∇ · jw , (5)
dt It is known, at least empirically, that irreversible flows are linear
where w represents the internal energy flux exchanged by the porous functions of the thermodynamic forces, as expressed by the phe-
plug with its environment, and j(±) and jw represent the flux of ions nomenological laws, which are introduced ad hoc to conform to
and solvent (in mol m2 s−1 ), respectively. Eqs (2) to (5) yield, experimental results. For example, Darcy’s law expresses the ob-
served fact that hydraulic flux is a linear function of the pore water
ds 1 1  
= − ∇ ·w+ µ̃(+) ∇ · j(+) + µ̃(−) ∇ · j(−) + µw ∇ · jw , pressure gradient. Ohm’s law expresses that the electrical current is
dt T T proportional to the electrical field. Also included in this type of linear
(6) relationships are the laws for such cross-phenomena as streaming
   
ds 1   1 potential (in which an electrical field is produced in response to the
+∇· w − µ̃(+) j(+) − µ̃(−) j(−) − µw jw = w∇ flow of pore water through a porous material) and electro-osmosis
dt T T
      (in which a flow of pore water is produced in response to the ap-
µ̃(+) µ̃(−) µw
− j(+) ∇ − j(−) ∇ − jw ∇ , (7) plication of an electrical field to a water-saturated porous material).
T T T In the vicinity of equilibrium, it is therefore safe to assume that all
ds the fluxes are linearly dependent on all the forces operative in the
+ ∇ · j S = , (8) system:
dt
1   j = −L 11 ∇ψ − L 12 (∇ p − ρ f g), (13)
jS = w − µ̃(+) j(+) − µ̃(−) j(−) − µw jw , (9)
T
        u = −L 21 ∇ψ − L 22 (∇ p − ρ f g). (14)
1 µ̃(+) µ̃(−) µw
 = w∇ − j(+) ∇ − j∇ − jw ∇ , The condition D ≥ 0 (as both T and  are positive) and the use of
T T T T
eqs (12) to (14) yields immediately L 12 = L 21 (Onsager recipro-
(10)
cal law) (we note  = L 12 = L 21 this coupling term, expressed
where j S is the entropy flux vector (in Walt m−2 ) and  represents the in m2 V−1 s−1 ), L ii ≥ 0 (straight permeability coefficients defi-
rate of inner entropy production (in Walt m−2 ) in the porous plug nite positive), and 2 ≤ L 11 L 22 . Consequently, the open system is
(positive definite since entropy can only be created in the system characterized by a set of 3 independent coefficients, L 11 = σ the
during irreversible transformations). Eq. (8) represents a conserva- electrical conductivity (in S m−1 ) of the porous plug, L 22 = k/η f


C 2002 RAS, GJI, 151, 944–949
946 A. Revil

the ratio between the intrinsic permeability k (in m2 ) and the dy- ξ = σ S /σ f is the ratio between surface to pore fluid electrical con-
namic shear viscosity of the pore water η f (in Pa s), and . The ductivity (dimensionless), σ S represents the surface conductivity (in
conservation equations for electrical charge is obtained combining S m−1 ), and t(+) is the Hittorf number of the cations in the electrolyte
eq. (4) with (dimensionless), which represents the fraction of electrical current
  transported by the cations in the pore water (t(+) = 0.38 for NaCl
j = e j(+) − j(−) , and 0.51 for KCl). Surface conductivity corresponds to an electrical
conduction mechanism located in the close vicinity of the pore wa-
∇ · j = −dρ/dt, (15) ter/mineral interface in the so-called electrical double layer coating
where ρ = e(n (+) − n (−) ) represents the free charge density per unit the mineral water interface (e.g. Revil & Leroy 2001). In eq. (20),
volume of the porous rock. Eq. (5) and u = w jw (and n w w = φ χd ≡ (ε f kb T /2e2 I )1/2 is the Debye screening length (approximately
for a water saturated porous medium) yields: half the thickness of the electrical double layer coating the pore wa-
ter/mineral interface), kb is the Boltzmann constant, I is the ionic
∇ · (ρ f u) = −d(ρ f φ)/dt, (16) strength in mol L−1 (∼salinity) of the pore water, ε f is the dielectric
constant of water ∼(80 × 8.84) ×10−12 F m−1 , and ζ is the so-called

Downloaded from https://academic.oup.com/gji/article-abstract/151/3/944/623281 by guest on 08 November 2018

the conservation equation for the mass of the pore water. Now that
ζ -potential, a key-property of the electrical double layer (e.g. Lorne
the form of the macroscopic equations has been established, we dis-
et al. 1999; Pengra et al. 1999; Revil & Leroy 2001).
cuss the relationship between the material properties entering these
The correction term (1 – 2 χd /) (valid as long as χd  /2,
equations and the microstructural parameters of granular porous
Pride 1994) accounts for the finite thickness of the electrical dou-
media.
ble layer by comparison with the characteristic length  controlling
transport in the connected pore space (see Kostek et al. 1992 and
3 MATERIAL PROPERTIES Wildenschild et al. 2000 for a description of the characteristic length
, which corresponds to an effective pore throat radius controlling
Eqs (13) to (15) imply that the flow of pore water generates an
the transport properties through the connected porosity). This length
electrical field called the streaming potential. The strength of this
scale is not a geometrical parameter and therefore cannot be mea-
electrical field is determined by the streaming potential coupling
sured directly. However, it is a very convenient parameter to use in
coefficient C (in V Pa−1 ):
  the purpose to unify transport properties of porous materials (Kostek
∂  et al. 1992). In (21), d is the mean diameter of the grains and α is a nu-
C≡ =− , (17)
∂ p j=0 σ merical constant depending on the pore space topology (Section 4).
For granular materials composed by uniform grains of diameter d,
where  represents the electrical potential difference produced in the formation factor F and the -parameter are given by (e.g. Revil
response to a pore water pressure drop p imposed at the end faces & Cathles 1999):
of the jacketed cylindrical sample. In order to obtain the material
properties of a granular porous medium with a random distribution F = φ −m , (22)
of grains of the same size, different methods can be applied and
d
combined. The volume averaging procedure of the local equations = , (23)
2m(F − 1)
can be applied to a wide class of porous media (e.g. Pride 1994), and
therefore to granular media as a specific case. For the electrical con- where m represents a grain shape parameter related to the shape
ductivity, a differential effective medium approach is a very efficient distribution of the grains (Sen et al. 1981). Eqs (22) and (23) have
method to obtain the effective electrical conductivity as a function actually the same status as they can both be derived using a differ-
of the porosity (e.g. Sen et al. 1981). For granular porous media, the ential effective medium approach applied to the determination of
three independent coefficients entering into eqs (13) and (14) are the electrical conductivity of a representative elementary volume of
related to two fundamental textural parameters characterizing the a granular porous medium (Revil & Cathles 1999). Eq. (22) should
porous material F and  by not be mislead with the Archie’s law, which, in the original paper by
σf Archie (1942), applies only to the porosity/formation factor correla-
σ = H(ξ ), (18) tion of a set of samples coming from the same formation reservoir.
F
In the high porosity limit (φ → 1) corresponding to a dilute sus-
1 pensions of spheres, m = 1.5 (e.g. Sen et al. 1981), (22) and (23)
H(ξ ) ≡ 1 − t(+) + Fξ + (t(+) − ξ )
 2  tend to
 2  
ξ 1 4F 3 1−φ
× 1 − + 1− ξ + ξ , (19) F = φ −3/2 → 1 + , (24)
t(+) t(+) t(+) 2 φ

  2φ d
εf ζ 2χd → . (25)
≈− 1− , (20) 9(1 − φ)
ηf F 
Eq. (24) represents the well-known Hashin-Shtrikman lower bound
d2 and (25) is similar to eq. (24c) of Kostek et al. (1992). The Hashin-
k= , (21)
α F(F − 1)2 Shtrikman lower bound is known to be fairly accurate in the limit of
dilute suspensions of spherical particles whatever its geometry. At
where eqs (18) and (19) result from the application of a differen-
high fractional connected porosity (φ  φ p , say φ > 0.40) F can
tial effective medium approach (Revil et al. 1998), eq. (20) results
be replaced by its asymptotic limit and (21) reaches a well-known
from a volume-averaging approach (Pride 1994), and eq. (21) from
asymptotic limit:
a combination of both methods (Revil & Cathles 1999). In (18)  
and (19), σ f represents the electrical conductivity of the pore wa- 8 d 2φ3
k≈ , (26)
ter (in S m−1 ), F is the electrical formation factor (dimensionless), 27α (1 − φ)2


C 2002 RAS, GJI, 151, 944–949
 Hydroelectric coupling 947

which corresponds to the Kozeny-Carman relationship with α ≈ 53

(e.g. Carman 1937). In the high porosity limit, the Kozeny-Carman
relationship is a fairly good indicator of the permeability. How-
ever, the Kozeny-Carman relationship fails to predict accurately the
permeability at porosity below 0.30 (Revil & Cathles 1999, their
Fig. 5).
One of the purposes of this note is to improve the predictive
capability of (18) to (21) at porosity below 0.30. This can be done
by introducing a percolation porosity in the electrical formation
factor/porosity relationship and therefore in the transport properties
described by eqs (18) to (21):
F = (φ − φ p )−m , (27)
where φ p represents the percolation porosity. Eq. (27) can be seen

Downloaded from https://academic.oup.com/gji/article-abstract/151/3/944/623281 by guest on 08 November 2018

as an attempt to force a percolation threshold into an equation ob-
tained from the use of differential effective medium theory, which
excludes, by construction, any percolation threshold. The percola-
tion porosity corresponds to the minimum value of the connected
porosity at which connected paths through the representative ele-
mentary volume exist. The value of φ p depends on the porosity
reduction mechanism, e.g. mineral precipitation, pressure solution.
For example, if the diameter of the grains growths uniformly during
the porosity reduction process, the percolation porosity is expected
to be in the range ∼0.02–0.05 (for an sc-packing of spheres of
same diameter, φ p ≈ 0.035, Roberts & Schwartz (1985); Schwartz
& Kimminau (1987). Near percolation of course, any type of mod-
eling based on the notion of elementary representative volume is
in principle not valid as heterogeneities have the same size as
the volume investigated. Using percolation concepts, Guéguen &
Palciauskas (1992) obtained a formation factor/porosity relation-
ship F ∼ (φ − φ p )−2 φ −1 quite similar to eq. (27). I show below
that the introduction of a percolation threshold in eqs (18) to (21)
improves clearly the predictive capabilities of these equations down
to very low porosities. Figure 1. (a) Electrical conductivity of a crushed Fontainebleau sandstone
versus pore water conductivity (data from Lorne et al. 1999, grain size
4 ELECTRICAL CONDUCTIVITY 80 µm, pH 5.7, electrolyte is KCl and therefore t(+) = 0.51). The electrical
formation factor and the surface conductivity are determined from a best fit of
I test here the validity of the electrical conductivity/formation fac- the electrical conductivity model described in the main text and materialized
tor relationship using the data obtained by Lorne et al. (1999, their by the plain line. (b) Electrical formation factor versus porosity. Comparison
Fig. 5) (crushed Fontainebleau sandstone with a mean grain diam- between the predictions of the modified differential effective medium theory
eter ∼80 µm). Eqs (18) and (19) provide an excellent fit of the with a percolation threshold, the classical effective medium theory (DEM),
experimental data from Lorne et al. (1999) in the entire salinity and the Hashin-Shtrikman lower bound (the filled black circles corresponds
to the experimental data from Johnson et al. 1982).
range above the iso-conductivity point characterized by the condi-
tion σ = σ f (see Fig. 1a). The specific surface conductivity S (in
S) represents the excess electrical conductivity associated with the tion factor/porosity relationship improves clearly the ability of the
excess of ions located in the electrical double layer (Revil & Leroy porosity/formation factor to predict the F versus φ trend at poros-
2001). The specific surface conductivity is related to σ S and the ity below 0.25. The data reported in Fig. 1(b) yields a percolation
mean grain diameter d of the porous aggregate by σS = 4 S /d porosity ∼0.025 in agreement with that determined for porous rocks
(e.g. Revil & Leroy 2001). Using d = 80 µm and the surface con- by Roberts & Schwartz (1985) and Schwartz & Kimminau (1987)
ductivity determined in Fig. 1(a) using eqs (18) and (19), I obtain with unimodal and very narrow grain size distributions.
S = (2.1 ± 0.4) × 10−9 S. This value is in the same range with that
determined independently from the double layer model of Revil &
5 PERMEABILITY
Leroy (2001), which is (1.6 ± 0.3) × 10−9 S at pH 7.
For granular porous materials, formation factor versus porosity As explained in Section 3, there is a relationship between hydraulic
data are shown in Fig. 1(b). For an assemblage of perfect spheres, the and electrical transport properties. If we note σS the equivalent con-
differential effective medium theory predicts a porosity/formation ductivity of a grain coated with the electrical double layer, a sim-
factor relationship, eq. (22), with m = 1.5 (Sen et al. 1981). In ple arithmetic average of the electrical conductivity of the porous
Fig. 1(b), we compare three estimates to evaluate the formation medium yields σ = φ σ f + (1 − φ) σS . In real porous materials, the
factor/porosity relationship. They are eq. (22) with m = 1.5, the porosity should be replaced by the inverse of the electrical formation
Hashin-Shtrikman lower bound, eq. (24), and the modified poros- factor, which represents by definition the efficient surface area for
ity/formation factor relationship with a percolation porosity, eq. (27) the electromigration of the ions through the connected pore space.
with m = 1.5. The inclusion of a percolation threshold in the forma- This yields:


C 2002 RAS, GJI, 151, 944–949
948 A. Revil

1
σ = [σ f + (1 − F)σ S ]. (28)
F
According to volume averaging arguments, the electrical conduc-
tivity of a porous material σ in the high salinity limit is also given
by (e.g. Pride 1994),
 
1 2
σ = σ f + S + · · · , (29)
F 
where the specific surface conductivity S is related to σS by
σS = 2 S /R where R = 2 d is the radius of the grains. Com-
parison between eqs (28) and (29) yields  = d/[2(F − 1)], which
result is slightly different from eq. (23). A relationship between the
permeability and the length scale  is given by k = 2 /a F with

Downloaded from https://academic.oup.com/gji/article-abstract/151/3/944/623281 by guest on 08 November 2018

usually 4 ≤ a ≤ 8 (Pride 1994). Incorporating  = d/[2(F − 1)]
in this relationship yields eq. (21) with α = 4 a.
In Figs 2(a) and (b), our modified permeability model is tested
against the data reported by Spangenberg et al. (1998) for salt rocks
and Chilindar (1964) for natural sandstones. In the first case, we
check the relationship between k and F using α = 8. We obtain an
excellent agreement between eq. (21) and the experimental data
reported by Spangenberg et al. (1998) over six orders of magnitude.
In Fig. 2(b), permeability versus porosity data show the necessity
to incorporate a percolation threshold in the equations. For these
types of sandstone, this yields a percolation porosity ∼0.025 like in
Section 3.

6 STREAMING POTENTIAL COUPLING

COEFFICIENT
Introduction of a percolation threshold in the electrokinetic coupling
coefficient C is performed taking eq. (17) and using (18) to (20),
CHS  χd 
C= 1 − 4m(F − 1) , (30)
H(ξ ) d

εf ζ
CHS ≡ , (31)
ηf σf
Figure 2. (a) Permeability versus electrical formation factor (data from
where CHS is the streaming potential coupling coefficient corre- Spangenberg et al. 1998, their Fig. 6, and corresponding to salt rocks).
sponding to the Helmholtz-Smoluchowski (HS) formula. The HS The initial mean grain diameter of the samples is in the range 0.5–2 mm
formula is widely used in an important number of publications to in- for the coarse-grained samples and 0.1–0.5 mm for the fine-grained sam-
terpret electrokinetic laboratory experiments or field data. It predicts ples. (b) Comparison between the model and the data from Chilindar (1964,
that the streaming potential coupling coefficient does not depend on Fig. 2, p. 73) (fine and silty-grained natural sandstone naturally compacted).
Parameter used: m = 1.5, φ p = 0.025 and the best fit of the model with the
the microstructure of the porous medium. Actually, experimental
experimental data yields d = 116 µm for the fine-grained sandstone (filled
data reported in the literature showed that the streaming potential circles) and 46 µm for the silt (filled squares).
coupling coefficient depends quite strongly on the microgeometry
of the porous medium at porosity below 0.30 (e.g. Bull & Gortner
1932). Actually the HS formula is valid only at very high salin- low porosity. For the Fontainebleau sandstone, this yields a perco-
ity and/or high porosity where the two conditions H(ξ  1) = 1+ lation porosity ∼0.035 ± 0.10, quite similar to the value reported in
O(ξ ) + · · · and χd  d/[4(F − 1)] are simultaneously satisfied. Sections 3 and 4 (0.025).
According to eqs (30) and (31), the streaming potential coefficient
depends generally on the porosity. The normalized streaming po-
7 CONCLUDING STATEMENTS
tential coefficient C/CHS is shown as a function of the porosity
in Fig. 3(a). The model predicts that the ratio C/CHS increases In this paper, I show (1) how the hydroelectric equations result
strongly with the porosity, especially at low salinity. This prediction from the Gibbs-Duhem equation and (2) that the three materials
is very well confirmed by the experimental data reported by Jouniaux properties (electrical conductivity, permeability, and electrokinetic
& Pozzi (1995) (Fig. 3b). The surface conductivity determined from coupling coefficient) entering the hydroelectric equations exhibit a
eqs (30) and (31) and the data by Jouniaux & Pozzi (1995) is similar percolation porosity. For sandstones and granular materials,
σ S = 0.84×10−4 S m−1 , which is in good agreement with that deter- the percolation porosity is in the range 0.02–0.04, which is quite
mined independently from electrical conductivity data in Fig. 1(a) small. For other types of porous rocks like vesicular basalts, it is
(σ S = 1.03 × 10−4 S m−1 ). In addition the data of Fig. 3(b) show a expected that this percolation porosity could be much higher than
good agreement with a model including a percolation threshold at the previous value owing to their different texture.


C 2002 RAS, GJI, 151, 944–949
 Birch, F.S., 1998. Imaging the water table by filtering self-potential profiles,
Ground Water, 36, 779–782.
Bull, H.B. & Gortner R.A., 1932. Electrokinetic potentials, X. The effect of
particle size on the potential, J. Phys. Chem., 36, 111–119.
Carman, P.C., 1937. Fluid flow through a granular bed, Trans. Inst. Chem.
Eng. London, 15, 150–156.
Chilindar, G.V., 1964. Relationship between porosity, permeability and grain
size distribution of sands and sandstones, in Deltaic and Shallow Marine
Deposits, Vol. I, pp. 71–75, ed. Van Straaten, L.M.J.U., Elsevier, New
York.
Guéguen, Y. & Palciauskas, V., 1992. Introduction à la Physique des Roches,
Hermann, Paris, p. 299.
Ishido, T. & Pritchett, J.W., 1999. Numerical simulation of electrokinetic
potentials associated with subsurface fluid flow, J. geophys. Res., 104,
15 247–15 259.

Downloaded from https://academic.oup.com/gji/article-abstract/151/3/944/623281 by guest on 08 November 2018

Johnson, D.L., Plona, T.J., Scala, C., Pasierb, F. & Kojima, H., 1982. Tortu-
osity and acoustic slow waves, Phys. Rev. Lett., 49, 1840–1844.
Jouniaux, L. & Pozzi, J.-P., 1995. Streaming potential and permeability
of saturated sandstones under triaxial stress: consequences for elec-
trotelluric anomalies prior to earthquake, J. geophys. Res., 100, 10 197–
10 209.
Kostek, S., Schwartz, L.M. & Johnson, D.L., 1992. Fluid permeability in
porous media: comparison of electrical estimates with hydrodynamical
calculations, Phys. Rev. B, 45, 186–195.
Lorne, B., Perrier, F. & Avouac, J.-P., 1999. Streaming potential measure-
ments. 1. Properties of the electrical double layer from crushed rock sam-
ples, J. geophys. Res., 104, 17 857–17 877.
Pengra, D.B., Li, S.X. & Wong, P., 1999. Determination of rock proper-
ties by low-frequency AC electrokinetics, J. geophys. Res., 104, 29 485–
29 508.
Pride, S.R., 1994. Governing equations for the coupled electromagnetics and
acoustics of porous media, Phys. Rev. B, 50, 15 678–15 696.
Prigogine, I., 1949. Le domaine de validité de la thermodynamique des
phénomènes irréversibles, Physica, 15, 272–284.
Revil, A. & Cathles, L.M., 1999. Permeability of shaly sands, Water Resour.
Figure 3. (a) Normalized streaming potential coupling coefficient versus Res., 35, 651–662.
porosity. Parameters used: m = 1.5, φ p = 0.02, and σ S = 1 × 10−4 S m−1 . Revil, A. & Leroy, P., 2001. Hydroelectric coupling in a clayey material,
(b) Influence of the porosity upon C. Comparison between the data reported Geophys. Res. Lett., 28, 1643–1646.
by Jouniaux & Pozzi (1995, Table 1 samples #F82, F17, F01 and F44) and Revil, A., Cathles, L.M., Losh, S. & Nunn, J.A., 1998. Electrical conduc-
the model described in the main text. Parameters used: σ f = 1×10−3 S m−1 tivity in shaly sands with geophysical applications, J. geophys. Res., 103,
(NaCl, measured), ζ = −40 mV (from Lorne et al. 1999, measured their 23 925–23 936.
Fig. 20), ε f = 7.1 × 10−10 F m−1 , and η f = 1.0 × 10−3 Pa s. The porosity Revil, A., Naudet, V., Nouzaret, J. & Pessel, M., Principles of electrogra-
of samples F17 and F01 is determined from the permeability/porosity trend phy applied to self-potential electrokinetic sources and hydrogeological
typical of the Fontainebleau sandstone and the permeability data reported for applications, Water Resources Research.
these samples in Jouniaux & Pozzi (1995, Table 1) (uncertainty better than Revil, A., Hermitte, D., Voltz, M., Moussa, R., Lacas, J.-G., Bourrié, G. &
10 per cent in relative value). A non-linear fit of the data is performed using Trolard, F., 2002. Self-potential signals associated with variations of the
the model developed in the main text. The values of surface conductivity hydraulic head during an infiltration experiment, Geophys. Res. Lett., 10,
and percolation porosity are reported on the graph. 1029/2001GL014294.
Roberts, J.N. & Schwartz, L.M., 1985. Grain consolidation and electrical
conductivity in porous media, Phys. Rev., 31, 5990–5997.
ACKNOWLEDGMENTS Schwartz, L.M. & Kimminau, S., 1987. Analysis of electrical conduction in
I am grateful to Frédéric Perrier for fruitful discussions and the the grain consolidation model, Geophysics, 52, 1402–1411.
two referees for their useful comments. I thank the Ministère Sen, P.N., Scala, C. & Cohen, M.H., 1981. Self-similar model for sedimentary
rocks with application to the dielectric constant of fused glass beads,
de l’Education Nationale de la Recherche et de la Technologie
Geophysics, 46, 781–795.
(MENRT) in France (ACI Jeunes and ACI ‘Eau et Environnement’)
Spangenberg, E., Spangenberg, U. & Heindorf, C., 1998. An experimental
and the French National Research Council (CNRS). B. Dupré, M. study of transport properties of porous rock salt, Phys. Chem. Earth, 23,
Cara, P. Choukroune, and B. Hamelin are especially thanked for 367–371.
encouraging supports during the last three years. Trique, M., Richon, P., Perrier, F., Avouac, J.P. & Sabroux, J.C., 1999. Radon
emanation and electric potential variations associated with transient de-
REFERENCES formation near reservoir lakes, Nature, 399, 137–141.
Wildenschild, D., Roberts, J.J. & Carlberg, E.D., 2000. On the relationship
Archie, G.E., 1942. The electrical resistivity as an aid in determining some between microstructure and electrical and hydraulic properties of sand-
reservoir characteristics, J. Pet. Tech., 5, 1. clay mixtures, Geophys. Res. Lett., 27, 3085–3088.


C 2002 RAS, GJI, 151, 944–949