The Electrokinetic Cross-Coupling Coefficient: Two-Scale Homogenization Approach ()
1. Introduction
In numerous studies on the electrolyte flows in rocks, the pore pressure
and the streaming (electric) potential
interplay through the equation
, where
is the current density,
is the saturated rock conductivity, and
is the electrokinetic cross-coupling term. Hydrogeological applications concern the study of water leakage from dams [1], groundwater flows in geothermal fields and volcanoes [2], estimation of water resources [3]. In electrochemistry, the above equation form a basis for managing microchip separations of analytes in nano-channels [4]; there is also an evidence that this equation find applications in hydrocarbon recovery [5,6].
By the Helmholtz-Smoluchowski theory [7], the term
is given by the formula
where
is the porosity,
is the dielectric permittivity of the saturating fluid,
is the viscosity, and
is the so-called
-potential, the electric potential across the diffuse part of the interfacial double layer. In [8,9,10], the above formula is substituted by
(or more sophisticated formulas), with
being a dimensionless formation factor.
The goal of the present paper is to give more mathematical insight into the physico-chemical nature of the cross coupling coefficient
. Restricting ourselves to one-dimensional flows, we derive a representation formula for
by the two-scale homogenization technique [11,12], starting from the equations of the ions transport through a layered membrane with a periodical structure. On this way we arrive at electro-osmotic macro-equations, whereas electrokinetic coupling coefficients can be determined from micro-equations defined on the periodicity cell.
Homogenization is a process in which the composite material with microscopic structure is replaced by an equivalent material with macroscopic, homogeneous properties. There are two methods of up-scaling coupled equations at the microscale to equations valid at macroscale for fluid-saturated porous media. The first is the volume averaging and the second is the two-scale and multiscale homogenization. Volume averaging has been applied successfully to derive the form of Biot's equations of poroelasticity [13], and a wide variety of other up-scaling problems in double-porosity poroelasticity [14]. The averaging theorem used by all these authors is due to J. C. Slattery (1967) [15] and is based on well-known Green's theorem together with the idea that in relatively small regions volume averages of spatial gradient in statistically homogeneous media are presumably closely related to gradient of volume averages.
The two-scale homogenization method requires that the heterogeneous microstructure of a rock sample is described by spatially periodic parameters and the microscale of the heterogeneous porous medium is much smaller than the macroscale of most interest. The approach involves assuming that any quantity can be treated as a function of a macroscale variable and a microscale variable. The two-scale homogenization is a well established method in the theory of partial differential equations with rapidly oscillating periodic coefficients. This method has a lot of important applications in various branches of physics, mechanics and modern technology: porous media, composite and perforated materials, thermal conduction, acoustics, electromagnetism. For general references on the homogenization theory we refer to [12,16,17,18].
The two-scale homogenization method can give formulas for coefficients in the up-scaled equations, whereas volume averaging methods give the form of the up-scaled equations but generally must be supplemented with physical arguments and/or data in order to determine the coefficients. A more detailed comparison of two up-scaling methods can be found in [19].
The present study is applicable to sandstones if surface conductivity can be neglected. When passing to claycontaining rocks one should also take into account bound charges concentrating on the interface surfaces. Such rocks are not considered here.
2. Background
Within the frame of the nonequilibrium thermodynamics, the fluxes (the Darcy's volume fluid velocity
and the electric current density
) are derived as a linear combination of thermodynamical forces (the pressure gradient
and the electric potential gradient
):
(1)
(2)
where
is the permeability. Our goal is to show that these equations, specified for one-dimensional flows through a layered membrane, can be derived by the two-scale homogenization technique starting from the equations valid at microscale. While deriving the up-scaled equations (1) and (2), (which can trace back to Helmholtz and von Smoluchowski) we obtain a formula for the cross-coupling coefficient
.
In this section, we summarize equations that govern the flows of a binary electrolyte solution through the pore space of a solid dielectric. To make clear our hypotheses on physical parameters, we use the Gaussian system of units. Clearly, while comparing final calculations with experiments, we apply the SI units. The electric field E obeys the charge conservation law
(3)
where
is the fluid dielectric permittivity, ![](https://www.scirp.org/html/9-4900031\3308e481-4e8c-4dfc-8730-e25ba38a79c3.jpg)
is the charge of a positive ion, ![](https://www.scirp.org/html/9-4900031\cb8439cd-da3c-4360-9c70-b9e3cedb189d.jpg)
is the charge of the negative ion,
is the ion concentration. Viscous incompressible flows of the electrolyte solution is governed by the Navier-Stokes equations [7]
(4)
with the inertial terms being neglected in the first momentum equation. Here,
is the velocity of the bulk fluid. The motion of both the ionic species satisfies the transport equation
![](https://www.scirp.org/html/9-4900031\ea19ef87-7d2b-4674-804e-aa7c36de3395.jpg)
(5)
with the flux given by the Nernst-Plank relation [7]
![](https://www.scirp.org/html/9-4900031\fdccb4d3-3058-4036-a467-9f1c26b7c6ae.jpg)
where
is the diffusion coefficient,
is the Boltzmann constant,
is the absolute temperature.
Inside the solid dielectric, the electrical field obeys the equation
![](https://www.scirp.org/html/9-4900031\afc11f14-6aa0-4049-ba97-0f31af5e6633.jpg)
where
is the solid dielectric permittivity. In what follows,
stands for the electric potential,![](https://www.scirp.org/html/9-4900031\0144c5bf-9529-4eb1-b656-3c701886e71f.jpg)
.
The solid-fluid boundary conditions will be formulated below for one-dimensional flows.
3. One-Dimensional Flows
To motivate our further study we keep in mind a vertical membrane of thickness
(see Figure 1) when the inflow pressure
(on the left) is grater than the outflow pressure
. It is the pressure gradient
which mainly controls the flow. It is also possible that the flow is due to the external electrical field
. Commonly, an inflow concentration
of the
-th ion is prescribed on the left.
Now, to perform analytical study of the flow equations, we consider a vertical "membrane" of an infinite thickness. We study electrolyte steady flows through the horizontal layer of thickness
consisting of
horizontal thin slits
of the same thickness
separated by layers
of a solid
![](https://www.scirp.org/html/9-4900031\eebb7d1b-242d-4e8b-9d9e-c0ecf0b66147.jpg)
Figure 1. Layered membrane of the thicknes
consisting of
solid/liquid layers.
dielectric of the same thickness
. The central points
of the liquid intervals
are the points of reference where the ion inflow densities
take the prescribed values
.
Let
and
stand for fluid and solid domains
![](https://www.scirp.org/html/9-4900031\c1209b13-d3f1-404b-b10e-2c904d0a6d7f.jpg)
![](https://www.scirp.org/html/9-4900031\bad1d2bb-1ef8-4563-b255-1c316b824c78.jpg)
![](https://www.scirp.org/html/9-4900031\cf072c63-fa46-4ef3-b1d4-66a0f95fbfbe.jpg)
![](https://www.scirp.org/html/9-4900031\5c346575-e888-40c0-8c79-a0ea764347d0.jpg)
![](https://www.scirp.org/html/9-4900031\fbb41d29-eb52-4050-bd77-6024b61a6ab5.jpg)
In the domain
, we look for steady solutions of the fluid Equations (3-5) in the form
![](https://www.scirp.org/html/9-4900031\3a0b1840-89f1-4642-90c6-a68d22871a92.jpg)
![](https://www.scirp.org/html/9-4900031\3ca79e21-2de2-48d1-bc01-2cb060fe474d.jpg)
where
and
are given constants. Under these assumptions, Equations (3-5) in each fluid interval
become
(6)
![](https://www.scirp.org/html/9-4900031\7cc539b2-bb1f-4f93-a9f1-8739750cef00.jpg)
We study horizontal flows along the
-axis, hence
. The latter equality is equivalent to
![](https://www.scirp.org/html/9-4900031\f933ef75-30a2-42f4-9dab-5a81cfca4a9f.jpg)
Integrating between
and
, we exclude the concentration functions from consideration by the formula
![](https://www.scirp.org/html/9-4900031\a63ef5f2-6dd1-4a25-825c-a250910b37e3.jpg)
In the solid intervals
, the potential
satisfies the equatio
(7)
In what follows we assume that the dielectric permittivity function and the fluid indicator function
![](https://www.scirp.org/html/9-4900031\b06853b5-d6aa-45d8-9ddc-3cc98f7bb471.jpg)
are extended periodically on the real line
. Given a function
continuous everywhere except a point
, we introduce the jump as follows
![](https://www.scirp.org/html/9-4900031\db3953e3-c371-4709-b39d-3777078f846e.jpg)
In some sandstones, surface conductivity can be neglected depending on the pore water salinity and the cation exchange capacity of the mineral surface. For such sandstones, the “electric” boundary conditions reduce to the conditions of continuity of the potential
and the normal component the electric induction vector
:
(8)
where k = 1,…, n-1 and l = 0,…, n-1.
The velocity satisfies the no-slip conditions
(9)
We assume that
satisfies the external boundary conditions
(10)
with the prescribed
-potentials
and
. We introduce a function
, which takes the value of the integer part of the number
. Then the functions
![](https://www.scirp.org/html/9-4900031\9959c048-afd2-480b-ad80-d4e8e08fafca.jpg)
![](https://www.scirp.org/html/9-4900031\d0947d83-0a17-47f5-965d-94a645ee5719.jpg)
take constant values
for
. Thus to define
on the whole interval
, one should solve the non-local Poisson-Boltzmann equation
![](https://www.scirp.org/html/9-4900031\59edb16c-f465-4a22-8c5f-a81a24521048.jpg)
(11)
jointly with the conditions (8) and (10). Observe that the function
is periodic, and
on the interval of periodicity
.
With the function
at hand, one can find a velocity
from Equations (6) and the boundary conditions (9).
4. Nondimensionalisation
We look for an asymptotic solution of problem (11), (8), (10), (6), (9) for the functions
and
, assuming that the ratio
is a small parameter for some positive entire number
. We argue by the homogenization approach [24], so the entire interval
is fixed and
varies in
. In that case
and
![](https://www.scirp.org/html/9-4900031\3fde113f-fb69-4212-9b9c-bb15cdee5d7e.jpg)
Here,
is the porosity.
We call
a slow variable and we introduce the fast variable
. With
being small, the periodic functions
and
oscillate strongly and they can be represented as functions of the fast variable:
![](https://www.scirp.org/html/9-4900031\09e35e9f-3251-4a93-a101-17f42c423c7a.jpg)
where
![](https://www.scirp.org/html/9-4900031\15d64a8c-ef74-4c38-8268-19a048f10aeb.jpg)
are periodic functions with the period equal to 1. In what follows the functions
![](https://www.scirp.org/html/9-4900031\6d76304c-f8f6-4870-aa5d-16ea5b563034.jpg)
and
![](https://www.scirp.org/html/9-4900031\c363caef-eac1-402d-925d-10b66bee3f6a.jpg)
are extended periodically for all
. The functions
,
,
can be written as
![](https://www.scirp.org/html/9-4900031\c8cea8e6-7606-4115-8457-0a8753ce1b96.jpg)
and
![](https://www.scirp.org/html/9-4900031\c592ec6d-5535-4ad3-ad51-f9e16f81db6f.jpg)
In the notations used, the function
on the interval
is a solution of the problem
(12)
![](https://www.scirp.org/html/9-4900031\585e692e-dfe8-4525-b79a-aa57c3a57d3e.jpg)
where
is equal to
![](https://www.scirp.org/html/9-4900031\18762790-e8de-4ad2-8daf-fb2745215cfd.jpg)
It follows from Equations (6) that the bulk velocity satisfies the equation
(13)
Let us perform scaling, using the symbol
for a reference value of the dimensional quantity
and the symbol
for a dimensionless quantity of
, i. e.
. We use the following notations:
![](https://www.scirp.org/html/9-4900031\6c95e206-a5ad-48ce-b44c-b88263467a52.jpg)
![](https://www.scirp.org/html/9-4900031\0bdfe1da-a0b8-47d5-ab87-67069e9f32bc.jpg)
![](https://www.scirp.org/html/9-4900031\b80b9038-a243-464a-abc1-350ee8a7ff2f.jpg)
![](https://www.scirp.org/html/9-4900031\f5a9b194-e87f-405d-bb6d-6fff0835c44a.jpg)
The quantity
(14)
is known as the Debye length. In terms of dimensionless variables Equations (13) and (11) in the fluid domain
take the form
![](https://www.scirp.org/html/9-4900031\f49bdf0d-6d27-4cca-b9f0-4d1c6b45be9b.jpg)
![](https://www.scirp.org/html/9-4900031\a16eb858-9fcc-4b2d-b770-da5f3ada7cbc.jpg)
Here,
![](https://www.scirp.org/html/9-4900031\d557d187-a44b-4622-8644-89dedd4780ea.jpg)
In the solid domain Equation (7) becomes ![](https://www.scirp.org/html/9-4900031\963e6732-7585-4bd8-987e-fe766ccff7f3.jpg)
Assuming that the dimensionless quantities
satisfy the equalities
![](https://www.scirp.org/html/9-4900031\e57b57f1-a7cd-4490-956b-1ee248d87cf7.jpg)
we obtain a hierarchy of problems to study. In this paper we restrict ourselves to the case when all the powers
are equal to zero, i. e.
. The meaning of these hypotheses is the following. The relation
implies that electroosmotic force and thermal force are of the same order. Observe that the relation
holds, for example, for the symmetric electrolyte (where
and
) in water at
, with the valency
and with the
-potential equal to
[4]. When
is not small, the Debye-Hückel linearization of the Poisson-Boltzmann equation does not work. Under the condition
the Debye length
can be longer compared to electrical double layer, moreover the double layer overlapping could occur. Indeed, it is a useful rule of thumb that
[4] where
is the valency. For the above mentioned electrolyte with the counterion molar concentration
we have
, whereas the double electric layer is normally only a few nanometers thick [4] and the nanocapillary membrane may have the pore diameter of 15
[20]. For such cases the hypothesis
is natural. Hypothesis
amounts to the effect that the horizontal pressure gradient and the applied horizontal electrical field are of the same order. The relation
means that viscous response is of the same order as the applied horizontal pressure gradient.
There is one more assumption that we impose on the Péclet number
:
(15)
The hypothesis implies that convection and diffusion are of the same order.
We close this section by reminding the Debye-Hückel approach to the Poisson-Boltzmann Equation (11) in the single layer
with the boundary conditions
and
as
and
. In the case of symmetric electrolyte, the linearized equation (11), in the SI system of units where
is substituted by 1, becomes
, since the nonlocal term
vanishes as
. Clearly,
is a solution. This explains the notion (14).
5. Asymptotic Analysis of Electric Field
We proceed by returning to the dimensional variables. Using the method of the two-scale expansions [12], we look for the solution of Equation (12) in the form of an expansion series
(16)
where the functions
are periodic in the variable
,
, with a period equal to 1 for each
. We introduce the flux
(17)
Clearly,
(18)
We present this flux as a series
(19)
where the functions
are 1-periodic in
for all
.
Using the formula
![](https://www.scirp.org/html/9-4900031\8be1a5fa-d990-44e2-87ee-62c6d4fc3763.jpg)
and substituting the series (16) and (19) into equality (17), we obtain an equality which looks like
![](https://www.scirp.org/html/9-4900031\56e52ebd-3dd0-4409-8b29-a10db642cbad.jpg)
Thus
for all
. In particular, the three first equalities can be written as
![](https://www.scirp.org/html/9-4900031\52548e2c-b55f-4e78-a8e4-7c148d953092.jpg)
and
![](https://www.scirp.org/html/9-4900031\8b8a9e61-31d9-4e28-9670-2afe36c7bfbe.jpg)
![](https://www.scirp.org/html/9-4900031\4e0f5b52-e450-44be-a589-d3eac9514f55.jpg)
Substituting the series (16) and (19) into equality (18) and paying attention to the powers
and
, we obtain the equations
(20)
![](https://www.scirp.org/html/9-4900031\5009dfd8-de32-4e95-b8e8-49029f4476fd.jpg)
![](https://www.scirp.org/html/9-4900031\c1bfc860-9ffb-4c41-975f-7324b7a26215.jpg)
(21)
Equations (20) and (21) allow one to determine the functions
,
and
uniquely. Indeed, with a function
independent of the variable
, we look for
by the method of separation of variables assuming that there exists a
-periodic function
such that
![](https://www.scirp.org/html/9-4900031\e49439bf-582d-45b9-aedf-8d46ba7f3513.jpg)
Substituting this presentation into equation (20), we find that the function
solves the following problem on the interval
:
(22)
The latter integral condition serves for uniqueness. We integrate and arrive at the formulas
(23)
![](https://www.scirp.org/html/9-4900031\ed141fbe-f5db-4386-a3ac-f1722df7f9ca.jpg)
Next, we use periodicity and integrate equation (21) with respect to
to obtain the following macro-equation for
:
(24)
As for the function
, we look it in the form
![](https://www.scirp.org/html/9-4900031\06c85832-5093-4b65-b40c-590e090de408.jpg)
Substituting this presentation into equation (21), we find that
is a periodic solution of the problem
![](https://www.scirp.org/html/9-4900031\ff4dcf38-6202-44c4-8be3-8a0133a25ec2.jpg)
(25)
This problem has a unique solution provided
![](https://www.scirp.org/html/9-4900031\cd24fb9b-fb39-43a8-ac18-9281242408e3.jpg)
Thus, we have established the following asymptotic equality for the electric potential:
(26)
6. Asymptotic Analysis of Velocity
Integrating Equation (13), we obtain the following formula for velocity in each fluid domain
:
![](https://www.scirp.org/html/9-4900031\9759713b-cd2c-462d-9b69-26f7e8f6abeb.jpg)
where
![](https://www.scirp.org/html/9-4900031\5b39142f-411e-4062-bc58-dbdf71e672a9.jpg)
We extent the function
by zero to the solid intervals and denote such an extension by
. Now, with
standing for
, we have for all
that
![](https://www.scirp.org/html/9-4900031\d8832f61-b86c-4937-af80-22c12eacb1db.jpg)
(27)
With
given by the expansion series (16), we look for
in the form
(28)
where the functions
are
-periodic in
and
for
. After simple calculations, we find that
![](https://www.scirp.org/html/9-4900031\d531c2c2-0b2a-4da9-b0dc-eecb8b407557.jpg)
![](https://www.scirp.org/html/9-4900031\3ea8bb4b-77f4-4618-ae0d-4a0062fee7de.jpg)
Using the properties of functions
,
,
, we obtain that
is equal to
![](https://www.scirp.org/html/9-4900031\c04d8db3-0711-4b12-a9bf-65dfff34549b.jpg)
for ![](https://www.scirp.org/html/9-4900031\61588df8-e9d9-4e6c-8a15-4427bbbc5d67.jpg)
As for
, we find that it is equal to
![](https://www.scirp.org/html/9-4900031\211a77cc-1875-412b-8a6d-6268604bcd7a.jpg)
for ![](https://www.scirp.org/html/9-4900031\0845b9f1-36d3-4437-b291-86801a812554.jpg)
By virtue of the multiplier
in the right side of equation (11), we can assume that ![](https://www.scirp.org/html/9-4900031\994b8e28-831e-4a38-8b3b-7046e49d0c9d.jpg)
Then, the variables
belong to the interval
![](https://www.scirp.org/html/9-4900031\e481b91d-0bd0-493d-bb21-28d89f0d1348.jpg)
also. As
is between
and
, the inequalities
![](https://www.scirp.org/html/9-4900031\f3f8dc54-3027-49a0-8622-8bbeaf653cda.jpg)
hold and the second derivatives of
in Equation (25) are meaningful. In addition, it follows from Equations (23), (24) and (25) that, for
, the functions
satisfy the equations
![](https://www.scirp.org/html/9-4900031\ff023c4f-86d4-4557-b384-4db8a01bfa33.jpg)
![](https://www.scirp.org/html/9-4900031\904683c2-4374-46b3-b9c4-0c1e4fbede91.jpg)
Thus, we obtain
(29)
Substituting equations (28) and (29) into equation (27) and considering only the power
, one can show that the function
does not depend on the variable
and has the form
(30)
Integrating equation (30) over the periodicity cell, we obtain the macroscopic equation
![](https://www.scirp.org/html/9-4900031\133fb797-2dcc-4af9-831c-771a037ba334.jpg)
where the hydrodynamic and electrochemical mobilities are defined by the formulas
![](https://www.scirp.org/html/9-4900031\765d19c4-9e79-4105-a40b-ed323280454e.jpg)
Thus, we have established the following asymptotic equality for the velocity field:
(31)
We introduce the total electric current
whose horizontal component in the fluid phase is equal to
![](https://www.scirp.org/html/9-4900031\80abe3cd-0c85-46bb-a04a-da9956af7648.jpg)
We extent the function
by zero to the solid intervals and denote such an extension by
. Due to hypothesis (15), we have that
. This is why we look for
in the form of the expansion series
(32)
It follows from (32) and (28) that
![](https://www.scirp.org/html/9-4900031\c7aa2db3-433a-4fbf-9e68-b752b36b7e73.jpg)
By integration, we arrive at the macroscopic equation
![](https://www.scirp.org/html/9-4900031\e0002db4-bf0f-4bd1-8192-dc5924d705af.jpg)
where
and
![](https://www.scirp.org/html/9-4900031\bb82c509-618c-46ec-8bdd-0fe1a5c4856b.jpg)
Thus, we have established the following asymptotic equality for the electric current:
(33)
The asymptotic equalities (26), (31) and (33) are valid in the sense of weak or two-scale convergences; mathematical aspects of these asymptotic expansions are extensively investigated in [21,22,23].
7. Electrokinetic Coupling Coefficients
We introduce the Darcy volumetric flow rate
and the current density
. By the above asymptotic analysis we have derived the macroequations (which are valid up to terms
,
)
(34)
![](https://www.scirp.org/html/9-4900031\81c79e4e-939c-40b1-9050-da842a0a3b08.jpg)
which describe electrolyte flow and distribution of the electric potential
across a layered membrane under the assumption that
![](https://www.scirp.org/html/9-4900031\598a4880-b2e0-4c13-bcf6-18868e3db650.jpg)
are prescribed data. For such a membrane, the effective dielectric permittivity
and the electrokinetic coupling coefficients
are given by the formulas
(35)
(36)
(37)
Formula (35) stating that the effective permittivity
of the layered membrane is the harmonic mean of
and
was first derived by Maxwell (Maxwell 1881) in a different way [24]. Observe, that the Onsager reciprocity relation
[25] is not imposed but derived in the above calculations as a consequence of the homogenization procedure. Moreover, the inequality
(38)
providing nonnegativity of the entropy production rate is also satisfied automatically [26] due to the representation formulas (36) and (37). The inequality (38) becomes equality if both the diffusion coefficients
are negligible. Observe that for some free solutions
[27].
We emphasize that the coupling coefficients
in the macro-equations (34) are given by the representation formulas (36) and (37) as a result of an extensive analysis of the micro-equations (22) and (25) for the functions
and
defined on the periodicity cell.
Clearly, the electroosmosis Equations (1) and (2) should coincide with the system (34) for onedimensional flows. Whereas the formula
![](https://www.scirp.org/html/9-4900031\4d0d9666-b9ba-4aae-9bfc-5dc79790238a.jpg)
for the cross-coupling coefficient have a drawback of measuring the
-potential, the kinetic coefficients
derived by homogenization for the ideal (layered) porous medium do not depend on
. One can exploit this advantage in calculation of the coupling coefficient
for general porous media.
Applying the general Equations (1) and (2) to the ideal (layered) porous medium, we find that
![](https://www.scirp.org/html/9-4900031\3af14122-a1c8-422b-983f-fe3a82231be0.jpg)
![](https://www.scirp.org/html/9-4900031\2e6253f4-4b71-4bbf-8b04-95c4423862b4.jpg)
Now, we have
(39)
and inequality (38) gives the following estimate for the electrokinetic cross-coupling coefficient
:
(40)
As for real rocks, formulas (39) and (40) suggest to take
in the form
(41)
where
is a dimensionless geometrical factor. In applications, the above formula can be of use if no data are available for the diffusion coefficients
and the
potential. We emphasize that formula (41) is not a physical law but rather an engineering formula which can be of help for some sandstones when surface conductivity can be neglected.
Firstly, we evaluate
for a rock sample on the basis of the F.F. Reuss experiment [28]. Such an experiment reveals that a difference in the electric potentials
applied to water in a U-tube results in a change of water levels when the low part of tube is plugged with a sandstone sample (Figure 2).
We calculate the weight
of salt water which fills the cylinder of height
with cross section area
(Figure 2). We have
, where
is the
![](https://www.scirp.org/html/9-4900031\ac17658e-7155-48f6-bdaf-5a0a88245a0f.jpg)
Figure 2. F. F. Reus experiment (1808) with the U-tube plugged with sandstone sample: applied electric field results in water level change of height![](https://www.scirp.org/html/9-4900031\4564109a-5bba-4a55-a4ae-2327d241ca47.jpg)
gravitational acceleration and
is the water density. The pressure drop across the sandstone plug is equal to
. On the other hand it follows from Equations (34) that at equilibrium, when
, we have
.
In [29,30], a mathematical model (jointly with a computer code) is developed for calculation of the electric conductivity
of a saturated rock. The model allows one to find an optimal Archie-like law
![](https://www.scirp.org/html/9-4900031\363e7fed-8a2f-4f0e-8077-83043aeb949a.jpg)
where
is the conductivity of the saturating fluid,
is the percolation threshold porosity,
is the cementation factor. For sandstones, it was calculated in [29] that
0.03,
1.5. Thus, for sandstones, formula (41) becomes
![](https://www.scirp.org/html/9-4900031\ab3ed236-aaaa-4f22-a836-58c579681327.jpg)
Now, the factor
can be evaluated from the formula
![](https://www.scirp.org/html/9-4900031\6e9f5da0-3b7f-4eee-92fd-6176242b2249.jpg)
We perform calculation assuming that, as in [31], the applied potential difference
results in the water level difference
. The rock data are taken from [6]:
,
,
,
,
. With these data at hand, we find that
. It is the cross-coupling coefficient
that can be measured in applications [32]. With the factor
given above, we find that
in agreement with the data in [6].
Next, we calculate the factor
for the rock sample composed of Berea sandstone 500 starting from experimental measurements of streaming potential when a fluid, with a prescribed NaCl concentration (500 ppm), flows through the sample [5]. Given data
,
,
,
, we find from formula (41) that
.
8. Conclusions
We have proposed a two-scale model for one-dimensional horizontal electroosmotic flows in a number of thin horizontal slits, with a horizontal pressure gradient and a horizontal electrical field being driving forces. The model is derived within the framework of homogenization in the up-scaling of the pore-scale description consisting of Stokes equation for bulk fluid flow and the Nernst-Plunk equation for the ion transport. The homogenized model is a generalization both of the Darcy law and the Ohm low. According to this model, both the fluid flux and the electric flux depend linearly on the horizontal pressure gradient and the horizontal electrical field, with the coupling coefficients obeying the Onsager symmetry condition and not depending on the
- potential.
As for three-dimensional general flows in sandstones in the case when surface current is negligible, the cross-coupling coefficient
is obtained in the approximate form
, where
is the fluid saturated rock electric conductivity,
is the rock permeability,
is the fluid viscosity, and
is dimensionless geometrical factor which depends on the sample. We evaluated that
for Berea sandstones.
9. Acknowledgements
The authors were supported by Russian Fund of Fundamental Researches (grant 10-05-00835-a) and State Contract No. 14.740.11.0355 of Federal Special-Purpose Program “Personal”.