Waxman Smiths Equation

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

PETROPHYSICS, VOL. 48, NO. 2 (APRIL 2007); P.

104–120; 17 FIGURES, 3 TABLES

Pore-Scale Analysis of the Waxman-Smits Shaly-Sand Conductivity Model1

Guodong Jin2, Carlos Torres-Verdín2, Sarath Devarajan2, Emmanuel Toumelin2,3, and E. C. Thomas4

ABSTRACT
Waxman-Smits and dual-water models of electrical resent homogeneous shaly sands that include the struc-
conductivity of shaly sands account for the dual conduc- tural effects of compaction, cementation, and distribution
tive pathways formed by pore brine and clay mineral of grain-coating clay minerals. Our pore-scale model is
exchange cations, while Archie’s equations describe the implemented on the digitized representative of rock sam-
electrical conductivity behavior of clay-free rocks. These ples, which allows us to consider arbitrary media bound-
empirical models are widely used in the interpretation of aries. Two-phase immiscible fluids are geometrically dis-
resistivity logs acquired in homogeneous reservoir rocks. tributed in the pore space using the ordinary percolation
However, the models are not explicit in their predictions algorithm. We introduce grain-coating hydrated clay min-
of electrical conductivity with respect to rock structure, erals and their corresponding electric double layer in the
spatial fluid distribution in the pore space, wettability, or synthetic rock models with a grain “shell” of variable
clay mineral distribution. The objective of this paper is to dimensions and diffusivity with respect to those of
quantify the influence of exchange cations associated pore-filling brine. Waxman-Smits formation factors and
with hydrated clay minerals, salinity of pore bulk water, resistivity indices are calculated directly with random
and water saturation on the electrical conductivity of walk simulations of late-time diffusion within the conduc-
shaly siliciclastic rocks. We accomplish this objective by tive space formed by the pore water and clay mineral
cal cu lat ing excess con duc tiv i ties asso ci ated with exchange cations in the digitized shaly-sand samples.
hydrated clay minerals for explicit pore geometries of Simulations with this simple model correctly reproduce
brine- and hydrocarbon-saturated shaly granular rocks. In the nonlinear behavior of rock conductivity at low and
so doing we introduce synthetic pore-scale representa- high salinity, and are consistent with reported laboratory
tions intended to reproduce experimental observations of measurements.
electrical conductivity of shaly sands. Keywords: shaly sand, pore-scale model, conductiv-
The synthetic pore-scale models are constructed to rep- ity, clays

Manuscript received by the Editor November 10, 2006; revised manuscript received January 26, 2007.
1
Originally presented at the SPWLA 47th Annual Logging Symposium, Veracruz, Mexico, June 4–7, 2006, Paper M.
2
The University of Texas at Austin, Department of Petroleum and Geosystems Engineering, One University Station, Mail Stop C0300,
Austin, Texas, 78712; E-mail: gjin@mail.utexas.edu, cverdin@mail.utexas.edu, sarath@mail.utexas.edu.
3
Currently with Chevron; 11111 S. Wilcrest Dr., Houston, TX, 77099; E-mail: toumelin@chevron.com
4
Bayou Petrophysics; P. O. Box 1027, Tomball, TX, 77377; E-mail: ecthomas@charter.net
©2007 Society of Petrophysicists and Well Log Analysts. All rights reserved.

104 PETROPHYSICS April 2007


Pore-Scale Analysis of the Waxman-Smits Shaly-Sand Conductivity Model

INTRODUCTION electrically equivalent effective volume conductivity. Kim


and Torquato (1990) and McCarthy (1990) calculated the
The most fundamental empirical relationship for inter-
effective conductivity of random mixtures by reducing
preting electrical conductivity measurements in reservoir
them to a multi-component conductivity problem. A com-
rocks was experimentally advanced by Archie (1942) as
mon feature of these models is that they assume a conduc-
sw a tive rock matrix. Using a diffusion current model, Sen
FR = = , (1) (1987) demonstrated that the nonlinearity of so versus sw
so f m
follows as a consequence of the distribution of the electric
where the formation resistivity factor is denoted by FR, and field between the pore bulk water and clay-bound water
the conductivities of the saturating water (or brine) and the regions. Revil and Glover (1998) advanced an effective
water-saturated rock are denoted by sw and so, respectively. medium theory that accounts for the different behavior of
The formation factor FR relates to the total interconnected anions and cations in these two conductive regions.
rock porosity f and the cementation (lithology) exponent m, The most widely used log-interpretation methods for
with the latter depending mainly on the degree of consolida- shaly sands in the oil indus try are based on the
tion of the rock. The coefficient a is a constant that depends Waxman-Smits and the dual-water models (Waxman and
on tortuosity, shape of the pores, and their connectivity. Smits, 1968; Clavier et al., 1984). Both models assume that
Archie’s law is applicable only to water-wet clean or clay exchange cations associated with clay minerals are the
mineral-free sands, i.e. non-conductive rock matrices agents behind clay conductance. The pore space in shaly
(Klein and Sill, 1982). However, laboratory measurements sands is composed of two zones (Clavier et al., 1977, 1984;
(Waxman and Smits, 1968; Waxman and Thomas, 1974) for Klein and Sill, 1982). The first is occupied by water that is
fully water-saturated shaly sands have shown that the rela- not affected by surface charges of clay minerals. The sec-
tionship between so and sw is strongly nonlinear as reflected ond is the electrical double-layer zone in near proximity to
the HCM surface, where the ionic distribution is controlled
by a convex-upward increase in so as sw ®0. This nonlinear
by the unbalanced charge defects of clay minerals. The ini-
behavior observed at low salinity is attributed to the addi-
tial curvature of the so – sw plot is attributed to an increase
tional conductivity resulting from the exchange cations
of the HCM counterion mobility and to the compression of
associated with hydrated-clay minerals (HCM) (Waxman
the double-layer thickness with increasing salinity of the
and Smits, 1968; Clavier et al., 1977, 1984; Bussian, 1983;
electrolyte.
Johnson and Sen, 1988; Revil and Glover, 1997, 1998;
An alternative interpretation of these electrochemical
Tabbagh et al., 2002). When clay minerals are present, ionic
concepts leads to the proposition that different values of
substitutions within the lattices result in an excess of nega- electrical conductivity exist close to the grain surface com-
tive charge near the HCM surface. The counterions (or cat- pared to the electrical conductivity of the bulk pore fluid.
ions) required to balance these charges form a thin “electri- Almon (1979) suggests that the conductivity in the dou-
cal double layer” around the HCM surface, where the ble-layer zone be an order of magnitude larger than the bulk
hydrated cations are in rapid exchange with those in the pore water conductivity even for moderately saline waters.
pore bulk water (Clavier et al., 1977, 1984). Electrical con- A decrease in salinity is expected to emphasize the relative
duction within this layer can contribute substantially to the contribution from the HCM counterions in the double-layer
electrical conductivity of the whole porous system at low zone, regardless of the model interpretation (Winsauer and
salinity or/and high temperature. McCardell, 1953). Revil et al. (1998) and Leroy and Revil
Many empirical and theoretical models were proposed to (2004) assign a specific surface conductivity in lieu of a
describe the role of water and clay content on the electrical clay mineral continuum by integrating the “anomalous”
conductivity of porous media (Waxman and Smits, 1968; surface conductivity over the double-layer thickness. An
Bussian, 1983; Clavier et al., 1984; Sen and Kan, 1987; analogous approach has been used for modeling the con-
Lima and Sharma, 1990; Revil and Glover, 1998). To be ductivity of the interfacial zone in cement mortar (Garboczi
useful for petrophysical interpretation, these models should et al., 1995).
be able to capture the nonlinear (convex-upward) behavior However, the nature of the excess conductivity associ-
of the plot of so versus sw at low values of salinity as well as ated with the presence of clay minerals requires further
the non-zero value of so when sw = 0. Bussian (1983) pro- investigation. There are properties specific to the dou-
posed a general model that related the electrical properties ble-layer that warrant a careful appraisal of parameters cho-
of any heterogeneous two-component mixture to the prop- sen and results obtained. One such property concerns the
erties of the individual components. Lima and Sharma choice of the true thickness d of an individual HCM cation
(1990) conceived of the surface conductivity of clays as an double-layer. This is unlike the mortar conductivity prob-

April 2007 PETROPHYSICS 105


Jin et al.

lem (Garboczi et al., 1995), where the width of the interfa- rent attributed to ions in the bulk water. This behavior is
cial zone is comparable to the grain radius. Clay minerals due to rapid exchange of HCM cations between the clay
do stack up on top of one other to form layers as thick as 12 mineral surface and bulk water. In other words, the
mm. However, individual double-layer thicknesses are still tortuosity is the same for all ions,
in the sub-micron range. Thus, d is the thickness of a repre- 3. The conductivity of HCM exchange cations, scl, is
sentative volume continuum that would describe the cumu- dependent upon the conductivity of bulk water, sw.
lative effect from surface phenomena at each double-layer. For fully water-saturated shaly sands, the W-S model is
Likewise, there is no simplistic relation that associates sur- defined as
face conductance with volumetric excess HCM exchange
cation conductivity. Parameter calibration is crucial owing 1
so = ( sw + scl ) , (2)
to the freely varying nature of d. FR*
The electrical response of shaly sands is also determined
by the distribution, amount, shape and type of clay miner- where the shaly-sand formation resistivity factor, FR* ,
als, the salinity of pore water, and water saturation. It is relates to total interconnected rock porosity and is inde-
essential to develop a clear understanding of the effect these pendent of formation-water salinity. The value of FR* can be
factors have on the electrical conductivity of shaly sands. In determined at high salinity where surface electrical conduc-
this paper we introduce a new pore-scale method based tion can be neglected. The conductivity of HCM exchange
upon the three-dimensional (3D) digital representation of cations scl is expressed as
the porous medium aimed at investigating the effect of the s cl = BQv , (3)
electrical conductivity contrast between bulk water and
clay-bound water, and their relative amount on the shaly where Qv is the volume concentration of HCM exchange
sand con duc tiv ity. First, we briefly describe the cations (equivalent/liter or meq/ml) and B represents the
Waxman-Smits model, which forms the theoretical basis of average mobility of the HCM counterions near their sur-
our study. In order to illustrate our simulation approach, we faces (mho cm2/meq), which at 25°C is given by
consider a clean, well-sorted, quartz-cemented, water-wet
shaly-sand model. The geometry of the model consists of B = 0.046[1- 0.6 exp(-s w / 0.013)]. (4)
electrically insulating spherical grains coated with a HCM
The W-S model captures the nonlinear behavior of so versus
counterion double-layer of thickness d, conductivity scl,
sw at low values of salinity by allowing the counterion
and the bulk pore electrolyte of conductivity sw. Electrical
mobility B to increase exponentially at low values of sw
conductivity of shaly sands is calculated directly from ran-
until it attains a constant and maximum value at high values
dom walk simulations of late-time diffusion. Subsequently,
of salinity. When sw = 0, the W-S model gives a finite value
we describe the algorithm of random walk simulation and
of so, i.e., the rock is still conductive even if it is saturated
our implementation on the digitized rock samples. Simula-
with fresh water. While B accounts for the well understood
tion results are presented and analyzed for two modeled
electrochemical phenomenon of ionic mobility, it is the one
shaly sands with different values of porosity.
free parameter that is used in the empirical W-S regression.
Regardless of the phenomenological accuracy of equation
THE WAXMAN AND SMITS MODEL (4), it is worthwhile recognizing that the product BQv repre-
sents the experimental excess conductivity due to the HCM
Waxman and Smits (W-S, 1968) proposed a physical
exchange cations at various values of salinity and can be
model to describe the dependence of shaly-sand conductiv-
used to calibrate a proposed conductivity model in response
ity on clay content, expressed as cation exchange capacity
to type and amount of clay, and type of clay distribution in
per unit pore volume. The W-S model has become the most
the pore space.
fully developed and widely accepted approach to the under-
Another parameter of interest is the “excess conductiv-
standing of the electrical conductivity of shaly-sand forma-
ity” associated with the HCM exchange cations. For better
tions. The main assump tions of the W-S model are
understanding, equation (2) is rewritten in its general form
(Bussian, 1983):
as
1. The electrical conductivities associated with both bulk
water in the pore space and HCM exchange cations con- sw
tribute to the electrical conductivity of shaly-sand for- so = (1 + X ) , (5)
FR*
mations in a manner analogous to a parallel circuit,
2. The electric current transported by HCM exchange ions where the dimensionless excess conductivity X, defined as
travels along the same tortuous path as the electrical cur-

106 PETROPHYSICS April 2007


Pore-Scale Analysis of the Waxman-Smits Shaly-Sand Conductivity Model

BQv ous cements inside the pore space of consolidated rocks


X= , (6)
sw (Bloch and Helmold, 1995). In this study, we assume that
cement, independent of its composition, is homogeneously
is a characteristic variable used directly or otherwise deposited on the grain surfaces within the sample according
included in most shaly-sand con duc tiv ity mod els to a proposed cement overgrowth algorithm (Roberts and
(Worthington, 1985). Grain conductivity models (Bussian, Schwartz, 1985; Bryant et al., 1993; Jin et al., 2003, 2006a).
1983; Lima and Sharma, 1990) use X in power laws involv- For simplicity, all grains in Figure 2 are allowed to grow
ing the cementation exponent. Realistic values of X at low uniformly, i.e. without moving their centers, to render low
values of salinity vary with bulk conductivity sw depending values of porosity. The increased sphere’s radius simulates
on Qv. At very high values of salinity this ratio approaches the growth of quartz cement (Schwartz and Kimminau,
zero. Figure 1 shows typical values of X calculated as a 1987).
function of sw and Qv. Dispersed clay minerals in sandstones are of three mor-
phological types: (1) pore lining, (2) pore bridging, and (3)
discrete particles (Wilson and Pittman, 1977; Neasham,
MODEL OF SHALY SAND
1977). Pore-lining clays are attached to pore walls to form a
We construct a model of shaly sand with a dense random relatively continuous and thin (£ 12 mm) clay mineral coat-
packing of mono-sized spheres. For simplicity, we use the ing (Neasham, 1977). In the model used here, we have cho-
real packing constructed by Finney (1968), who measured sen to describe the electrical properties of shaly sands with
the spatial coordinates of 8000 balls comprising the central a “simple” grain coating model (Johnson et al., 1986; Lima
part of the packing, in which 25000 mono-sized spherical and Sharma, 1990), in which a clay mineral shell with thick-
ball bearings were packed densely. The size of ball bearings ness d and corresponding HCM cations is added to the elec-
is 200 mm. Our simulation domain consists of 1000 spheres, trically insulating grains. Figure 3 shows the idealized grain
whose spatial coordinates were obtained from Finney’s data coating model, where d is the thickness of a representative
set (Figure 2). The initial porosity is approximately 36.5%. volume continuum intended to synthesize the cumulative
In addition, we assume that grains are electrically insulat- electrical effect from surface phenomena due to the corre-
ing.
Cementation is the process of mineral nucleation and
precipitation that binds the rock grains. Depending on the
chemical and crystallographic properties, cements precipi-
tate on the grain surfaces and affect the specific surface area
and tortuosity of the rock. However, it is not an easy task to
predict the location and chemical composition of the vari-

FIG. 2 Graphical illustration of a sand pack based on 1000


mono-size spherical grains. The dimensions of the packing are
FIG. 1 Plots of dimensionless excess conductivity X as a func- 2100 mm x 2100 mm x 2100 mm, with grain diameter equal to 220
tion of sw and Qv. mm and the porosity approximately equal to 36.5%.

April 2007 PETROPHYSICS 107


Jin et al.

sponding double layer. This thickness is small compared to tion of the discretization. However, such dependence is
the dimensions of both pores and grains. negligible if the spatial resolution is sufficiently high. For
Another interpretation of the clay shell thickness d is that our study, we determined a small voxel size of 6 mm from
the grain surface is electrically charged where electrical sensitivity studies of calculated permeability (Jin et al.,
neutrality requires an equal concentration of counterions 2006b).
that are confined to the surface within a distance d. These
additional HCM counterions contribute to the electrical
RANDOM WALK ALGORITHM AND
conductivity of the shaly sand beyond the contribution due IMPLEMENTATION
to bulk water. The water in the d zone is called the
clay-bound water, similar to the concept of the double layer Steady-state flow of electric current through a porous
developed by Clavier et al. (1977, 1984). As the salinity of medium saturated with an electrolyte is governed by the dif-
the pore water increases, d decreases and the surface cat- ferential equation (Revil and Glover, 1997)
ions are packed closer together near the surface (Johnson
and Sen, 1988; Sposito, 2004). Ñ× [ sÑF ]= 0 , (7)
To construct a 3D model of the pore space, we discretize where F is the local electrical potential. Additionally, the
the macroscopic cube with grains in the form of a 3D array no-flux boundary condition is enforced at the solid-void
of identical microscopic cubes (voxels). Our convention is interface when the solid phase is assumed to be insulating,
that a voxel is an element of the pore space if its center is i.e.
within the pore space; otherwise, it becomes an element of
the solid skeleton. Figure 4 displays the corresponding n× ÑF = 0 , (8)
discretized pore space of the grain packing shown in Figure
2, with voxel dimensions of 250 x 250 x 250, and with the where n is the unit normal vector outward from the solid to
side length of each voxel equal to 6 mm. Porosity can be the pore space. The solution of this problem in terms of the
directly evaluated from the fraction of void-space voxels in conductivity of the porous medium so is given by
the rock image, here approximately equal to 36.5%. The
j =-s o ÑF , (9)
porosity calculated this way depends on the spatial resolu-

FIG. 3 Idealized grain-coating clay mineral morphology shows


the spatial separation between the HCM exchange cations (or
clay-bound water) and bulk water regions. The size of the shell FIG. 4 Pore space volume of the grain pack shown in Figure 2.
of HCM exchange cations relative to the grain radius is exagger- The number of voxels is 250 x 250 x 250. Voxel resolution is 6
ated for better illustration. mm and the porosity is approximately 36.5%.

108 PETROPHYSICS April 2007


Pore-Scale Analysis of the Waxman-Smits Shaly-Sand Conductivity Model

where j and ÑF are the macroscopic electrical current sw 1 Dw


FR = = . (13)
density and potential gradient vectors, respectively. so f D¥
The differential equation (7) can be numerically solved
via the finite-difference or finite-element methods (Adler et For simplicity of calculation, the bulk water conductivity sw
al., 1992; Martys and Garboczi, 1992). However, such cal- and diffusion coefficient Dw are assigned the value of unity
culations become computationally expensive when the size in our simulations.
Our implementation of random walk simulation is based
of the simulated sample increases or when different con-
on the voxel-based rock sample. There are two kinds of
ducting phases exist in the heterogeneous sample with dif- voxels used in this study: coarse and fine (Figure 5). Each
ferent length scales. Moreover, they are difficult to imple- coarse voxel can be occupied by either the solid phase (A),
ment on synthetic 3D grain packs because of both variable oil (B), bulk water (C and D), or clay-bound water with
geometrical surfaces of grain boundaries and complex some oil or bulk water (E). We define the type of voxels E as
boundaries between immiscible fluids. those having a common face, edge, or vertex with voxels
Alternatively, the electrical conductivity of heteroge- occupied by the solid phase (A). The type of voxels D is
neous porous media can be calculated by simulating the defined as those occupied by bulk water and in contact with
Brownian motion of diffusive particles in the porous media voxels E. To explore the properties of the HCM exchange
cation region, voxels close to this region (type of voxels D
where the transport process is governed by a diffusion equa-
and E) are re-discretized into finer voxels. Fine voxels in E
tion together with the appropriate boundary conditions at can be occupied by clay-bound water, bulk water, or oil. In
the multiphase interface (Schwartz and Banavar, 1989; Kim the simulation, we assume that the rock is water wet.
and Torquato, 1990; Ioannidis et al., 1997). The diffusion In order to calculate a reliable ensemble average of the
equation is given by diffusion coefficient, many walker trajectories are gener-
¶Y
Ñ× [ DÑY ]= , (10)
¶t

where Y is the local potential and D is the diffusion coeffi-


cient. For late-time, i.e. steady-state diffusion, the character
of the above equation is identical to that describing the
behavior of the static electric potential. An important
parameter included in the simulation of the Brownian
motion of diffusive particles is the time-dependent diffusion
coefficient (or diffusivity) D(t) in the porous media,

r 2 (t )
D (t ) = , (11)
6t

where r 2 ( t ) is the mean-square displacement of walkers at


time t (Tobochnik et al., 1990; Latour et al., 1993; Regier
and Schuchmann, 2005). In the long-time limit, i.e., t ® ¥,
the diffusion coefficient reaches a constant value D¥ for clay
mineral-free rocks (Latour et al., 1995; Nakashima and
Watanabe, 2002). This is expressed by the relationship
D¥ 1
= 2, (12)
Dw t
FIG. 5 Description of voxel types in the coarse and fine grids.
where Dw is the diffusion coefficient of the bulk water, and t There are five types of coarse voxels, each occupied by either
is the tortuosity of the pore space. Through the tortuosity, solid phase (A), oil (B), bulk water (C) and D), or clay-bound
water with either oil or bulk water (E). Each of coarse voxels D
the diffusion coefficient is intimately connected to the for-
and E is rediscretized into finer voxels. Each fine voxel in the
mation factor FR (or rock conductivity so) (Latour et al., coarse voxels E may be occupied by clay-bound water (dark
1995; Toumelin, 2006), as cells), bulk water (medium cells), or oil (light cells). The rock is
assumed to be water wet.

April 2007 PETROPHYSICS 109


Jin et al.

ated and evaluated. In our model, the starting points of walk- We applied the random walk algorithm to calculate the
ers are randomly specified using a Monte Carlo algorithm. formation factor of a sandstone computed-tomography
Only those walkers within the pore space occupied by the (CT) image to validate this approach. There is no clay pres-
electrically conductive components, i.e. bulk water and ent in the rock core sample from which the CT image was
clay-bound water, are retained. The walker distribution is acquired. The sample has a measured porosity of 20.3% and
proportional to the ratio of the volume of the clay-bound formation factor of 17.5. The size of the CT image is
water zone to that of the bulk water zone. Each walker is dis- 300´300´300 voxels with the side length of each voxel
placed for a sufficiently long time. Length steps of a random equal to 4.5 mm. The calculated CT image porosity is
walker are dependent on the type of voxels or fluid in it, approximately equal to 21.1%. We carried out 10 independ-
whereas the direction is distributed equally and randomly. In ent simulations, each consisting of 1000 walkers, on the
the coarse voxels C, a large step size is used, which is propor- image and calculated the effective diffusivity for each sim-
tional to the size of coarse voxels, while the step size in ulation. Figure 6 shows an example of the variation of the
voxels D and E is proportional to the size of fine voxels. A effective diffusivity with simulation time for one simula-
random step is possible and the new location of the walker is tion, which describes an ensemble average of 1000 random
stored if the new location is not occupied by the non-conduc- walkers. From equation (13), we calculated the formation
tive phase (solid or oil). However, if the new position is factor for each simulation. The mean value of the formation
within solid or oil the step is not carried out, i.e. the walker is factor from 10 independent simulations is 18.5 with a stan-
returned to the previous position, and the clock is allowed to dard deviation of 0.21. Considering the difference in poros-
advance by one time step (Schwartz and Banavar, 1989). ity and spatial scales between the CT image and the core
A random walker can jump from the bulk water zone to
sample, our calculation is in good agreement with the
the clay-bound water zone with a given jump probability
experimental measurement.
pb,cl, or stay at the original position with probability (1 –
pb,cl). In either case, the simulation time is incremented by
one unit. The reverse condition occurs for the case of walk- RESULTS AND DISCUSSION
ers jumping from the clay-bound water zone to the bulk We performed the random walk simulations on two
water zone. We enforce a periodic boundary condition in models of shaly sands with different porosities, and investi-
the simulations, that is, a walker that moves out of the pore gated the effect of the double layer thickness d, diffusivity
space image at one side reappears at the opposite side. contrast Dcl /Dw between clay-bound water and bulk water,
and water saturation Sw on the electrical conductivity of the
modeled shaly sands.

Single-phase simulation
All grains in the Finney packing are uniformly grown
without moving their centers to generate two different poros-
ities of 25.4% and 18.9%. A homogeneous HCM exchange
cation shell of thickness d and diffusivity Dcl is added to each
electrically insulating grain. The pore space is saturated with
the electrolyte, which is assigned a diffusivity equal to Dw.
Note that different values of electrical conductivity exist
close to the clay mineral surfaces, i.e. in the double-layer
zone, compared to the electrical conductivity of the bulk pore
fluid, which could be an order of magnitude larger than the
bulk pore water conductivity even for moderately saline
waters (Almon, 1979). We assign a specific conductivity to
the water in the double-layer zone, and a different value to
the bulk pore water, in which the electrical conductivity is
FIG. 6 Change of effective diffusivity with increased simulation assumed to be proportional to the diffusivity of the corre-
time. The solid line describes an ensemble average of 1000 ran-
dom walkers. Fluid bulk diffusivity is assumed equal to 1
sponding water (Schwartz and Banavar, 1989).
mm2/ms, and the calculated effective diffusivity (asymptotic Input parameters in the simulation are the bulk water
value) is equal to 0.256 mm2/ms (dash line), with the standard diffusivity Dw, the clay-bound water diffusivity Dcl, the
deviation equal to 1.86 x 10–4 mm2/ms. ratio of the coarse voxel size to the random step size rc for

110 PETROPHYSICS April 2007


Pore-Scale Analysis of the Waxman-Smits Shaly-Sand Conductivity Model

TABLE 1 Parameters used in the random walk simulations. TABLE 2 The corresponding values of clay-bound water
saturation Sw,cl, and bulk water saturation Sw,b for each con-
Notation Values figuration of d in the two modeled shaly sands with different
values of porosity f and shaly-sand formation factor, FR*.
Dw (mm2/ms) 1
Dcl (mm2/ms) 1, 5 10, 25, 50, 100, 250, 500, 1000 f(%) FR* Notation Values
rc 2 (coarse voxel size 6 mm)
rf 2 (finer voxel size 0.6 mm) d (mm) 1.8 3.0 3.6 4.8
N 1000 18.9 16.6 Sw,cl (%) 18.1 29.0 33.7 42.0
tT (ms) 1.0´107 Sw,b(%) 81.9 71.0 66.3 58.0
d (mm) 1.8, 3.0, 3.6, 4.8
pb,cl (%) 50 d (mm) 1.8 3.0 3.6 4.8
25.4 9.0 Sw,cl (%) 15.6 25.3 29.5 37.3
Sw,b (%) 84.4 74.7 70.5 63.7

the type of voxels C, the ratio of the finer voxel size to the
random step size rf for the type of voxels D and E, the num-
tion (5) is used to compare our simulation results to the W-S
ber of walkers N, the travel time of each walker tT, the thick-
model. The simulated values of dimensionless excess con-
ness of clay-bound water d, and the probability pb,cl (Table
ductivity X vary with the diffusivity contrast Dcl /Dw and the
1). The calculated values of the shaly-sand formation factor
thickness d for the modeled shaly sands with porosity of
FR* and saturation of clay-bound water Sw,cl and bulk water
18.9% and 25.4%, respectively (Figures 7 and 8). The HCM
Sw,b are shown for each configuration of d (Table 2). The
exchange cation layer cannot be strictly treated as a contin-
value of FR* is calculated by assigning Dcl = Dw during the
uum of finite conductivity. The conductivity of HCM
simulation. It is discussed later that values of the diffusivity
exchange cations decreases from a “surface value” to the
contrast Dcl /Dw are close to 1 when the pore bulk water is
bulk value for both shaly-sand samples and is a function of
highly conductive. The value of FR* from the W-S equation
the normal distance from the grain surface (Schwartz et al.,
corresponds to the slope of the linear portion of the so ver-
1989). As d increases, the clay-bound water saturation in
sus sw plot that occurs at high values of sw.
the rock increases (Table 2), and its relative contribution to
If the bulk water conductivity sw and diffusivity Dw are
the overall rock conductivity increases, resulting in a higher
assigned the value of unity in equation (13), then the rock
excess conductivity X. With the porosity increasing, the rel-
conductivity can be calculated directly as so = fD¥. Equa-

FIG. 7 Simulated values of dimensionless excess conductivity FIG. 8 Simulated values of dimensionless excess conductivity
X as a function of diffusivity contrast Dcl /Dw and thickness d for X as a function of diffusivity contrast Dcl /Dw and thickness d for
the shaly sand with porosity equal to 18.9%. the shaly sand with porosity equal to 25.4%.

April 2007 PETROPHYSICS 111


Jin et al.

ative fraction of the HCM exchange cation region in the model based on the relative conductivity of the two phases,
rock decreases, and hence one obtains lower values of X. i.e., Dcl /Dw (Kim and Torquato, 1990) or the harmonic
In the simulations we use an unbiased jump probability mean of Dcl and Dw (McCarthy, 1990). However, a rigorous
(pb,cl = 50%) for walkers stepping across the phase bound- justification for these two approaches is yet to be provided.
aries, since HCM exchange cations on the mineral surface At low water conductivity the simulated rock conductiv-
are in rapid exchange with those in the bulk water. Other ity so is described as a function of the pore water conductiv-
biased probabilities could also be implemented in our ity sw for the shaly sand with porosity equal to 18.9% (Fig-
ure 9). Values of Qv = 0.20 meq/ml and d = 3.6 mm are used
in this plot. The values of so are derived by rescaling the
simulation results with the corresponding values of sw that
are obtained by choosing the appropriate value of Dcl /Dw in
Figure 7 and by resolving the dimensional excess conduc-
tivity from Figure 1. Note that the simulation curve of
shaly-sand conductivity is not linear at low values of sw. For
clarity, we calculated the differences between the simula-
tion results so, and the values obtained from the linear curve
connecting the two end-points of the simulation curve of
shaly sand, so,linear, and W-S calculation, so,W-S, from equa-
tion (2) (Figure 10). As expected, so is an increasing func-
tion of sw, and exhibits convex-up behavior at low values of
sw (Waxman and Smits, 1968; Waxman-Thomas, 1974).
The simulation rock conductivity so is consistent with the
calculations from Waxman-Smits model. It is also interest-
ing to note that the chosen diffusivity contrasts are propor-
tional to 1 / sw at high values of salinity but proportional
to 1 / sw at low values of salinity (Figure 11).
FIG. 9 Plot of so versus sw for the shaly sand with porosity equal
Reasonable parametric values can be surmised from a
to 18.9% at low values of salt concentration. Values of Qv = 0.20
meq/ml and d = 3.6 mm were used to construct the plot. The better understanding of the relationship between diffusivity
dashed line (parallel to that of clean sand) is used to emphasize contrasts and pore fluid conductivity sw. Diffusivity con-
the nonlinearity of the relationship between so and sw. trasts are strongly affected by different values of d (3.6 mm

FIG. 10 Differences between the simulation results so, values FIG. 11 Chosen values of Dcl /Dw as a function of sw used to
so,linear calculated from the linear relation connecting the two calculate the results shown in Figure 9. Values are selected to
end-points of the simulation curve of shaly sand in Figure 9, and yield an excess conductivity consistent with W-S model predic-
Waxman-Smits calculations so,W-S from equation (2). tions.

112 PETROPHYSICS April 2007


Pore-Scale Analysis of the Waxman-Smits Shaly-Sand Conductivity Model

and 4.8 mm) in a synthetic shaly sand with Qv = 0.20 meq/ml is also of interest. Excess conductivity generated by the
and f = 18.9% (Figure 12). It is clear that the desired value HCM counterions results in decreased values of ma at low
of Dcl /Dw at different values of salinity is influenced by the values of salinity. As salinity increases, ma reaches a con-
choice of d. In the case of sw < 0.005 mho/cm, the value of d stant value as the relative contribution of HCM conductivity
exerts a strong influence on the diffusivity contrast that decreases. Figure 13 describes the variation of ma with pore
needs to be chosen. The ratio Dcl /Dw can be maintained rel- fluid conductivity calculated for the shaly sand with f =
atively constant if d is allowed to increase with decreasing 18.9% and d = 3.6 mm. The variation of ma closely resem-
values of salinity. This is in keeping with increments in dou- bles the trends in bilogarithmic plots of Fa /FR versus sw
ble-layer thickness that have been found to occur below a reported by Worthington (1985).
critical value of salinity (Clavier et al., 1984; Sen, 1987; In addition to the rapid exchange with ions in the bulk
Bassiouni, 1994; Hill et al., 1979). Silva and Bassiouni water, the HCM exchange cations on the clay mineral sur-
(1988) calculated the fractional volume occupied by the face can move along the surface when the shaly sand is sub-
double layer as a function of far-water conductivity. A jected to an external electrical field (Revil and Glover,
decrease of far-water conductivity below a value of 0.03 1998). This movement results in finite rock conductivity
mho/cm is accompanied by a drastic increase in volume even if the shaly sand is saturated with fresh water (Dalla et
occupied by the double layer. This further justifies the al., 2004; Lima et al., 2005). Our simulations confirm the
selection of higher values of d at low values of salinity existence of this finite conductivity. Figure 14 shows the
while maintaining invariant diffusivity contrasts. In the variation of the simulated rock conductivity so with the
case of very shaly sands (Qv > 1 meq/ml) and under low thickness of clay-bound water d when the shaly sand is sat-
equilibrating brine conductivity (sw < 0.03 mho/cm) the urated with fresh water (sw = 0 or negligible). One can
desired excess conductivity is realizable only when d is observe that so increases linearly as d increases, and that the
comparable to the bulk/pore dimensions. The ratio Dcl /Dw conductivity becomes null when d = 0. This behavior is
can then be as low as 1, which eventually reduces to consistent with the property that the conductivity of dry
Waxman and Smits’ assumptions of a uniformly enhanced shaly sand is negligible (Bassiouni, 1994). The simulated
pore electrolyte (Waxman and Smits, 1968). rock conductivity can be rescaled back to the “true” value if
The variation of the apparent formation-porosity expo- we knew the conductivity of clay-bound water and its
nent ma, defined as diffusivity. In the simulation, we assume values of the con-
ductivity and diffusivity of clay-bound water equal to 1
1 sw mho/cm and 1 mm2/ms, respectively.
Fa = = , (14)
f ma
so

FIG. 13 Values of ma calculated from simulation results as a


FIG. 12 Diffusivity contrasts chosen to generate W-S model function of sw and Qv for the shaly sand with f = 18.9% and
results for the shaly sand with f = 18.9% and Qv = 0.20 meq/ml. d = 3.6 mm.

April 2007 PETROPHYSICS 113


Jin et al.

Two-phase simulation
We investigated the effect of water saturation on the
electrical conductivity of the modeled shaly sands at low
and high values of water conductivity, sw. For the simula-
tions, we maintain values of the clay shell thickness con-
stant at d = 3 mm, while the volume concentration of HCM
exchange cations is set to Qv = 0.20 meq/ml for both sam-
ples. At low sw, we choose the diffusivity contrast Dcl /Dw =
5. From Figure 1, Figure 7 and Figure 8, we determine that
sw = 0.031 mho/cm for the shaly sand with porosity of
18.9% and 0.025 mho/cm for the sample with porosity of
25.4%. At high sw, we set Dcl /Dw = 1.
We assume that these two samples are water-wet sand
and that a film of water is always maintained on the grain
surfaces. Under such a condition, the wetting phase (water)
occupies the corners of large pores and small pores, while
the non-wetting phase (oil) occupies the central parts of the
invaded pores. To assign the spatial fluid distributions in
the pore space, we use the technique of maximal-inscribed
spheres to obtain several distributions of water and oil for
each sample corresponding to a different value of capillary
pressure (Silin et al., 2003, 2004). The ordinary percolation
algorithm is used for that purpose. Figures 15(a) and (b) dis-
play an example of part of the pore space occupied by water
and oil in the sample with porosity of 18.9%, where water
saturation is Sw = 0.668 including clay-bound water and bulk
water. In the following simulations, we only consider the
cases in which water is connected through the pore space.
The resistivity index IR is used to quantify the effect of
partial saturation on the rock conductivity and is defined as

FIG. 15 Pore space occupied by the wetting phase, water (a),


and non-wetting fluid, oil (b), respectively, in the modeled shaly
sand with porosity equal to 18.9%. Water saturation, including
FIG. 14 Plot of so versus d for the shaly sand with f = 18.9% clay-bound water and bulk water, is Sw = 0.668. The dimension
under fresh water conditions (i.e., conductivity of bulk water of each image is 250 x 250 x 250 voxels, and the voxel resolu-
sw = 0). The fitting dashed line is so,fit = 5.6 x 10–3d. tion is 6 mm.

114 PETROPHYSICS April 2007


Pore-Scale Analysis of the Waxman-Smits Shaly-Sand Conductivity Model

low. We can represent this curved relationship as two linear


TABLE 3 Computed values of coefficients b* and saturation segments, with slopes n*L and n*H, respectively, and a
exponents n* in equation (16) for two modeled shaly sand cross-over saturation Sw,cr. However, there is no imperative
samples at low and high values of sw. to impose linearity if some other relationship is suggested
by the data. Similar anomalous relationships between I R*
Low sw High sw and Sw were reported by Diederix (1982) as due to surface
f (%) b*L n*L b*H n*H b* n* roughness based on laboratory measurements of Rotliegend
sandstone, which has a rough clay coating with Qv = 0.20
18.9 1.09 0.73 1.01 1.07 0.93 1.72 meq/ml on the grain surface. Other studies confirm the exis-
25.4 1.05 1.07 1.01 1.33 0.95 1.93 tence of this relationship (Taylor and Barker, 2002; Dalla et
al., 2004).
The values of coefficients b* and saturation exponents
n* from equation (16) are calculated for the two modeled
so
IR = » bS w-n , (15) shaly-sand samples (Table 3). At low values of water satu-
st ra tion, the con duc tiv ity enhance ment aris ing from
densely-packed HCM exchange cations offsets the resistiv-
where so is the conductivity of the rock fully saturated with
ity associated with lower values of Sw, thereby resulting in a
water and st is the conductivity of the rock partially satu-
lower value of n* = n*L. At high values of water saturation,
rated with water (Archie, 1942). The value of the coefficient
however, the relative contribution from HCM exchange
b is approximately equal to 1, and the saturation exponent n
cations to the rock conductivity becomes smaller and a
is typically about 2 for clean sands. However, the value of n
higher value of n* = n*H is obtained. Compared to Archie’s
varies with the volume of clay minerals present in the rock
formula I R* = S w-2 , the coefficient b* is very close to 1
and represents an apparent saturation exponent when calcu-
(±0.09), values of saturation exponent n*L and n*H are much
lated using equation (15) for shaly sands. When the forma-
smaller than the normally assumed value of n* = 2. This
tion is partially saturated, the HCM exchange cations are
result is in marked agreement with Diederix’s experimental
densely packed within the clay mineral zone and the relative
observations on Rotliegend sandstone (Diederix, 1982).
contribution to overall rock conductivity from the HCM
We also note that the values of n*L and n*H depend on the
layer is expected to increase. Based on experimental evi-
sample porosity. In our simulations at low sw, we assumed
dence, Waxman and Smits (1968) predicted this excess con-
that both samples had the same characteristics in the HCM
tribution to be inversely proportional to the value of water
saturation. They formulated an empirical relationship by
including the effect of the HCM exchange cations through
the function BQv /Sw as
é sw + BQv / Sw ù * -n*
I R* = I Rê ú» b S w , (16)
ë sw + BQv û

where I R* and n* are the “clay-corrected” resistivity index


and saturation exponent, respectively and the value of b* is
close to 1.0 (Diederix, 1982).
For each case of water saturation modeled we computed
the bulk conductivity of the corresponding part of pore
space occupied by water. Next, we calculated the clay-cor-
rected saturation exponent n* from equation (16) for each
shaly-sand sample. In the random walk simulations, we
assume that the electrical conductivity of oil is zero. Fig-
ure 16 shows the variation of the computed resistivity
index I R* with water saturation Sw for two samples at low sw.
These two samples do not exhibit the straight-line relation- FIG. 16 Computed resistivity index versus water saturation Sw
for two shaly sand samples for the case of low sw. Archie’s equa-
ship between I R* and Sw predicted by Archie’s law on a
tion used in the calculations is given by IR* = Sw-2. The fitting
bilogarithmic plot. Instead, the relationship is curved such curves have the form IR* = b *Sw-n *, where the values of b* and n*
that at low values of Sw the resistivity index I R* is relatively are listed in Table 3.

April 2007 PETROPHYSICS 115


Jin et al.

exchange cation zone: d = 3 mm, Qv = 0.20 meq/ml, and simulations were also observed in experimental measure-
Dcl/Dw = 5. This implies that the relative contribution on rock ments (Diederix, 1982; Taylor and Barker, 2002), we make
conductivity from HCM exchange cations is smaller in the the following remarks stemming from our simulation results:
sample with high porosity than in the sample with low poros- 1. The curved relationship of log(I R* ) versus log(Sw) in Fig-
ity. The sample with high porosity will exhibit relatively ure 16 was obtained at low sw, where the conductivity
larger values of n*L and n*H, as observed in our simulations. enhancement arising from the HCM exchange cations
The calculated values of cross-over critical water satura- becomes more prominent. We could not anticipate this
tion Sw,cr are 0.80 for both samples. This cross-over could be behav ior from Diederix’s exper i men tal results
larger and reach the value of 1.0 if the rock porosity were (Diederix, 1982).
low enough or else if d were large enough. Such behavior 2. The dependence of the saturation exponent n* on poros-
indicates that only the HCM exchange cations contribute to ity could be more likely related to the relative contribu-
rock conductivity or that their contribution is relatively tion of HCM exchange cations on rock conductivity. For
large compared to that due to pore bulk water. In the latter the rock with low porosity, the relative volume fraction
case, the two piecewise linear segments would become a of the HCM exchange cation zone to the bulk (or free)
single line with a slope equal to n*. water zone is large, which results in a substantially
In contrast to the curved relationship between log(I R* ) and higher electrical conductivity and a lower value of n*.
log(Sw) at low values of sw, our simulations exhibit the nor- 3. Water saturation may affect the electrical behavior of
mal linearity at high sw (Figure 17). One can observe that the shaly sands at low sw much more than at high sw. The
effect of water saturation on the exponent n* is negligible. dependence of I R* on water conductivity or salinity was
Although an increase in sw decreases the relative contribu- also observed in experimental measurements by Taylor
tion from the HCM counterions to the rock conductivity, and Barker (2002). However, such a statement cannot be
regardless of the model interpretation, this contribution still generalized without further investigation and verifica-
makes the value of n* smaller than 2. The value of n* is tion by experiments.
larger for the sample with high porosity of 25.4% (n*=1.93)
than for the sample with low porosity of 18.9% (n*=1.72).
CONCLUSIONS
Again, this result is due to the assumption of equal properties
in the HCM exchange cation zone for both samples. We developed a pore-scale random-walk model based on
Even though the anomalous resistivity index/water satu- the 3D digital representation of a porous medium and used
ration relationships of very low n*-values obtained in our it to simulate pore-level phenomena that govern electrical
conduction in shaly sands. With this model, we reproduced
realistic values of excess conductivity associated with the
HCM exchange cations at various electrolyte conductivi-
ties. Our model can be extended to various clay mineral
morphologies since we have made no assumptions regard-
ing continuous electrical pathways for the HCM cations.
Several previous works on this subject did not consider the
effective volumetric contribution to excess conductivity
resulting from surface conductivities associated with the
HCM cations.
We have shown that the relationship between interfacial
clay mineral conductivity and effective excess conductivity
of the HCM exchange cations needs to be calibrated using
the thickness of the volume continuum, d. The requirement
of increased values of d with a decrease in electrolyte salinity
is consistent with observed swelling of the anion-free layer at
low values of salinity. By maintaining constant diffusivity
contrasts while allowing only d to increase, we have simu-
lated the entire range of measured excess conductivities
FIG. 17 Computed resistivity index versus water saturation Sw associated with moderately shaly sands. The fact that the
for two shaly-sand samples for the case of high sw. Archie’s
equation used in the calculations is given by IR* = Sw-2. The fitting
thickness of the clay mineral zone d needs to be comparable
curves have the form IR* = b *Sw-n *, where the values of b* and n* in size to the pore dimensions at low values of salinity con-
are listed in Table 3. firms the W-S assumptions of a uniform pore electrolyte.

116 PETROPHYSICS April 2007


Pore-Scale Analysis of the Waxman-Smits Shaly-Sand Conductivity Model

Clay in the rock matrix plays an essential role in electri- n*H High value of n*
cal conduction, but only in the presence of an electrolyte. N Number of walkers
The conductivity of dry shaly sand is negligible. Our results pb,cl Jump probability from the bulk water zone to
confirmed the existence of a finite electrical conductivity of the clay-bound water zone
shaly sand saturated with fresh water. It was found that Qv Volume concentration of HCM exchange
there is no electrical conduction when the thickness of the cations, meq/ml
clay mineral zone d is zero. r Displacement of a walker at time t, mm
Simulations considered in this paper reproduced the rc Ratio of the coarse voxel size to the random
anomalous curved resistivity index/water saturation rela- step size
tionships of very low clay-corrected saturation exponent rf Ratio of the fine voxel size to the random step
n*, which are consistent with reported experimental obser- size
vations (Diederix, 1982; Taylor and Barker, 2002). The Sw Water saturation
simulated values of n* are lower than the typical value of 2. Sw,b Saturation of bulk water
It was found that the relative amount of the HCM exchange Sw,cl Saturation of clay-bound water
cation zone to the total conductive volume could dramati- Sw,cr Cross-over critical water saturation
cally affect the electrical behavior of shaly sands. t Time, ms
Our pore-scale model allows one to include arbitrary tT Travel time of each walker, ms
media boundaries and takes into account the effect of clay X Dimensionless excess conductivity
minerals on the electrical properties of the rock. The ability
of our approach to correctly predict the so versus sw curves Greek symbols
and the subsequent sensitivity to model parameters lends cre- scl Conductivity of HCM exchange cations or
dence to the physical consistency of the method. Future work clay-bound water, mho/cm
will involve studying the effects of wettability, saturation so Conductivity of the water-saturated rock,
cycles, and varying rock geometry on the model parameters. mho/cm
so,linear Conductivity interpolated from a linear
relationship, mho/cm
NOMENCLATURE so,W-S Conductivity calculated from Waxman-Smits’
a Coefficient in Archie’s equation model, mho/cm
b Coefficient in Archie’s equation st Conductivity of the rock partially saturated with
b* Modified coefficient in Archie’s equation water, mho/cm
bL* Low value of b* coefficient in Archie’s equation sw Conductivity of the saturating water, mho/cm
bH* High value of b* coefficient in Archie’s f Effective rock porosity
equation d Thickness of the shell of HCM exchange
B Mobility of HCM counterions, mho cm2/meq cations, mm
D Diffusivity, mm2/ms F Local electrical potential
Dw Diffusivity of bulk water, mm2/ms Y Local diffusion potential
Dcl Diffusivity of clay-bound water, mm2/ms t Tortuosity
D¥ Asymptotic value of time-dependent diffusivity,
mm2/ms ACKNOWLEDGMENTS
Fa Apparent formation resistivity factor
FR Rock formation resistivity factor The work reported in this paper was funded by the Uni-
FR* Shaly-sand formation resistivity factor versity of Texas at Austin’s Research Consortium on For-
HCM Hydrated-clay minerals mation Evaluation, jointly sponsored by Aramco, Baker
IR Resistivity index Atlas, BP, British Gas, ConocoPhillips, Chevron, ENI E&P,
I R* “Clay-corrected” resistivity index ExxonMobil, Halliburton Energy Services, Hydro, Mara-
j Electrical current density thon Oil Corporation, Mexican Institute for Petroleum,
m Cementation exponent in Archie’s equation O c c i d e n ta l P e t ro l e u m C o r p o ra t i o n , P e t ro b r a s ,
ma Apparent cementation exponent Schlumberger, Shell International E&P, Statoil, TOTAL,
n Unit normal vector and Weatherford.
n Saturation exponent in Archie’s equation We are also thankful to Professor Tad W. Patzek of the
n* “Clay-corrected” saturation exponent University of California at Berkeley and Dr. Dmitry B. Silin
n*L Low value of n* of Lawrence Berkeley National Laboratory for their help in

April 2007 PETROPHYSICS 117


Jin et al.

partitioning the images occupied by wetting and non-wet- conductivity and percolation aspects of statistically homoge-
ting fluids. A note of gratitude goes to Dr. James J. Howard neous porous media: Transport in Porous Media, vol. 29, no. 1,
p. 61–83.
and three anonymous reviewers of Petrophysics whose con- Jin, G., Patzek, T. W., and Silin, D. B., 2003, Physics-based recon-
structive editorial and technical suggestions improved the struction of sedimentary rocks, SPE-83587: Society of Petro-
original conference paper. leum Engineers, presented at Western Regional/ AAPG Pacific
Section Joint Meeting.
Jin, G., Patzek, T. W., and Silin, D. B., 2006a, Dynamic reconstruc-
REFERENCES tion of sedimentary rock using the distinct element method:
SPE Journal, in press.
Adler, P. M., Jacquin, C. G., and Thovert, J.-F., 1992, The forma- Jin, G., Patzek, T. W., and Silin, D. B., 2006b, Microstructure and
tion factor of reconstructed porous media: Water Resources permeability analysis of reconstructed reservoir rocks: Journal
Research, vol. 28, no. 6, p. 1571–1576. of Petroleum Science and Engineering, submitted.
Almon, W. R., 1979, A geologic appreciation of shaly sands, paper Johnson, D. L., Koplik, J., and Schwartz, L. M., 1986, New
WW, in 20th Annual Logging Symposium Transactions: Soci- pore-size parameter characterizing transport in porous media:
ety of Professional Well Log Analysts. Physical Review Letters, vol. 57, no. 20, p. 2564–2567.
Archie, G. E., 1942, The electrical resistivity log as an aid in deter- Johnson, D. L., and Sen, P. N., 1988, Dependence of the conduc-
mining some reservoir characteristics: Transactions of the tivity of a porous medium on electrolyte conductivity: Physical
American Institute of Mining and Metallurgical Engineers, vol. Review B, vol. 37 no. 7, p. 3502–3510.
146, p. 54–62. Kim, I. C., and Torquato, S., 1990, Determination of the effective
Bassiouni, Z., 1994, Electrical resistivity of rocks, in Theory, Mea- conductivity of heterogeneous media by Brownian motion sim-
surement, and Interpretation of Well Logs: Society of Petro- u la tion: Journal of Applied Physics, vol. 68, no. 8, p.
leum Engineers, Richardson, TX, p.1–24. 3892–3903.
Bloch, S., and Helmold, K. P., 1995, Approaches to predicting res- Klein, J. D., and Sill, W. R., 1982, Electrical properties of artificial
ervoir quantity in sandstones: AAPG Bulletin, vol. 79, no. 1, p. clay-bearing sandstone: Geophysics, vol. 47, no. 11, p.
97–115. 1593–1605.
Bryant, S. L., Cade, C., and Mellor, D., 1993, Permeability predic- Latour, L. L., Kleinberg, R. L., Mitra, P. P., and Sotak, C. H., 1995,
tion from geologic models: AAPG Bulletin, vol. 77, no. 8, p. Pore-size distribution and tortuosity in heterogeneous porous
1338–1350. media: Journal of Magnetic Resonance, Series A, vol. 112, no.
Bussian, A. E., 1983, Electrical conductance in a porous medium: 1, p. 83–91.
Geophysics, vol. 48, no. 9, p. 1258–1268. Latour, L. L., Mitra, P. P., Kleinberg, R. L., and Sotak, C. H., 1993,
Clavier, C., Coates, G., and Dumanoir, J., 1977, The theoretical Time-dependent diffusion coefficient of fluids in porous media
and experimental bases for the “dual water” model for the inter- as a probe of surface-to-volume ratio: Journal of Magnetic Res-
pretation of shaly sands, SPE-6859: Society of Petroleum Engi- onance, Series A, vol. 101, no. 3, p. 342–346.
neers, presented at 52nd Annual Technical Conference and Leroy, P., and Revil, A., 2004, A triple-layer model of the surface
Exhibition, 18p. electrochemical properties of clay minerals: Journal of Colloid
Clavier, C., Coates, G., and Dumanoir, J., 1984, Theoretical and and Interface Science, vol. 270, no. 2, p. 371–380.
experimental bases for the dual-water model for interpretation Lima, O. A. L. d., Clennell, M. B., Nery, G. G., and Niwas, S., 2005,
of shaly sands: Society of Petroleum Engineers Journal, vol. A volumetric approach for the resistivity response of freshwater
24, no. 2, p. 153–168. shaly sandstones: Geophysics, vol. 70, no. 1, p. F1–F10.
Dalla, E., Cassiani, G., Brovelli, A., and Pitea, D., 2004, Electrical Lima, O. A. L. d., and Sharma, M. M., 1990, A grain conductivity
conductivity of unsaturated porous media: pore-scale model approach to shaly sandstones: Geophysics, vol. 55, no. 10, p.
and comparison with laboratory data: Geophysical Research 1347–1356.
Letters, vol. 31, no. 5, paper L05609. Martys, N., and Garboczi, E. J., 1992, Length scales relating the
Diederix, K. M., 1982, Anomalous relationships between resistiv- fluid permeability and electrical conductivity in random
ity index and water saturations in the Rotliegend sandstone (the two-dimensional model porous media: Physical Review B, vol.
Netherlands), paper X, in 23rd Annual Logging Symposium 46, no. 10, p. 6080–6090.
Transactions: Society of Professional Well Log Analysts. McCarthy, J. F., 1990, Effective conductivity of many-component
Finney, J., 1968, Random packings and the structure of the liquid composites by a random walk method: J. Phys. A: Math. Gen.,
state: Ph.D. thesis, University of London. vol. 23, no. 15, p. L749–L753.
Garboczi, E., Schwartz, L., and Bentz, D., 1995, Modeling the Nakashima, Y., and Watanabe, Y., 2002, Estimate of transport
influence of the interfacial zone on the dc electrical conductiv- properties of porous media by microfocus x-ray computed
ity of mortar: Advanced Cement Based Materials, vol. 2, no. 5, tomography and random walk simulation: Water Resources
p. 169–181. Research, vol. 38, no. 12, p. 1272.
Hill, H. J., Shirley, O. J., and Klein, G. E., 1979, Bound water in Neasham, J. W., 1977, The morphology of dispersed clay in sand-
shaly sands and its relation to Qv and other formation proper- stone reservoirs and its effect on sandstone shaliness, pore
ties: The Log Analyst, vol. 21, no. 3, p. 3–19. space and fluid flow properties, SPE-6858: Society of Petro-
Ioannidis, M. A., Kwiecien, M. J., and Chatzis, I., 1997, Electrical

118 PETROPHYSICS April 2007


Pore-Scale Analysis of the Waxman-Smits Shaly-Sand Conductivity Model

leum Engineers, presented at 52nd Annual Technical Confer- of the pore-space morphology in sedimentary rocks: Journal of
ence and Exhibition. Petroleum Technology, vol. 56, no. 5, p. 69–70.
Regier, M., and Schuchmann, H. P., 2005, Monte Carlo simula- Silva, P. L., and Bassiouni, Z., 1988, Hydrocarbon saturation equa-
tions of observation time-dependent self-diffusion in porous tion in shaly sands according to the s-b conductivity model:
media models: Transport in Porous Media, vol. 59, no. 1, p. SPE Formation Evaluation, vol. 3, no. 3, p. 503–509.
115–126. Sposito, G., 2004, The Surface Chemistry of Natural Particles:
Revil, A., and Glover, P. W. J., 1997, Theory of ionic-surface elec- Oxford University Press, Oxford UK.
trical conduction in porous media: Physical Review B, vol. 55, Tabbagh, A., Panissod, C., Guerin, R., and Cosenza, P., 2002,
no. 3, p. 1757–1773. Numerical modeling of the role of water and clay content in
Revil, A., and Glover, P. W. J., 1998, Nature of surface electrical soils’ and rocks’ bulk electrical conductivity: Journal of Geo-
conductivity in natural sands, sandstones, and clays: Geophysi- physical Research, vol. 107, no. B11, p. 20–1.
cal Research Letters, vol. 25, no. 5, p. 691–694. Taylor, S., and Barker, R., 2002, Resistivity of partially saturated
Revil, A., III, Losh, S., and Nunn, J. A., 1998, Electrical conduc- Triassic sandstone: Geophysical Prospecting, vol. 50, no. 6, p.
tivity in shaly sands with geophysical applications: Journal of 603–613.
Geophysical Research, vol. 103, no. B10, p. 23925–23936. Tobochnik, J., Laing, D., and Wilson, G., 1990, Random walk cal-
Roberts, J. N., and Schwartz, L. M., 1985, Grain consolidation and culation of conductivity in continuum percolation: Physical
electrical conductivity in porous media: Physical Review B, Review A, vol. 4, no. 6, p. 3052–3058.
vol. 31, no. 9, p. 5990–5998. Toumelin, E., 2006, Pore-scale petrophysical models for the simula-
Schwartz, L. M., and Banavar, J. R., 1989, Transport properties of tion and combined interpretation of nuclear magnetic resonance
disordered continuum systems: Physical Review B, vol. 39, no. and wide-band electromagnetic measurements of saturated
16, p. 11965–11971. rocks: Ph.D. dissertation, The University of Texas at Austin.
Schwartz, L. M., and Kimminau, S., 1987, Analysis of electrical Waxman, M. H., and Smits, L. J. M., 1968, Electrical conductivi-
conduction in the grain consolidation model: Geophysics, vol. ties in oil-bearing shaly sands: SPE Journal, vol. 8, p. 107–122.
52, no. 10, p. 1402–1411. Waxman, M. H., and Thomas, E. C., 1974, Electrical conductivi-
Schwartz, L. M., Sen, P. N., and Johnson, D. L., 1989, Influence of ties in oil-bearing shaly sands: I. the relation between hydrocar-
rough surfaces on electrolytic conduction in porous media: bon saturation and resistivity index. ii. The temperature coeffi-
Physical Review B, vol. 40, no. 4, p. 2450–2459. cient of electrical conductivity: Society of Petroleum Engineers
Sen, P. N.,1987, Electrochemical origin of conduction in shaly for- Journal, vol. 14, p. 213–225.
mations, SPE-16787, Society of Petroleum Engineers: pre- Wilson, M. D., and Pittman, E. D., 1977, Authigenic clays in sand-
sented at 62nd Annual Technical Conference and Exhibition. stones: recognition and influence on reservoir properties and
Sen, P. N., and Kan, R., 1987, Electrolytic conduction in porous paleo-environmental analysis: Journal of Sedimentary Petrol-
media with charges: Physical Review Letters, vol. 58, no. 8, p. ogy, vol. 47 no. 1, p. 3–31.
778–780. Winsauer, W. O., and McCardell, W. M., 1953, Ionic double-layer
Silin, D. B., Jin, G., and Patzek, T. W., 2003, Robust determination conductivity in reservoir rock: Transactions of the American
of the pore space mor phol ogy in sed i men tary rocks, Institute of Mining and Metallurgical Engineers, vol. 198, p.
SPE-84296, Society of Petroleum Engineers: presented at 78th 129–134.
Annual Technical Conference and Exhibition. Worthington, P. F., 1985, The evolution of shaly-sand concepts in
Silin, D. B., Jin, G., and Patzek, T. W., 2004, Robust determination reservoir evaluation: The Log Analyst, vol. 26, no. 1, p. 23–40.

April 2007 PETROPHYSICS 119


Jin et al.

ABOUT THE AUTHORS


Guodong Jin received a PhD in Reservoir Engineering and Emmanuel Toumelin received a PhD degree in Petroleum
Petrophysics from the University of California at Berkeley in Engineering at the University of Texas at Austin and an engineer-
2006, and is currently a post-doctoral fellow in the Department of ing degree from the Ecole Centrale, Lille. From 2001 to 2005, his
Petroleum Engineering at the University of Texas at Austin. Dur- research at UT focused on developing pore-scale models of elec-
ing 1997-2000 he held the position of Research Associate with tromagnetic and nuclear magnetic resonance measurements in sat-
Institute of Mechanics, Chinese Academy of Sciences. His urated rocks. He also held research internship positions with Baker
research interests include pore-scale modeling of petrophysical Atlas, Schlumberger-Doll Research and ChevronTexaco between
response in reservoir rocks and shaly sands, rock fracture and dam- 2001 and 2003. Emmanuel joined Chevron North America E&P in
age mechanism, fluid flow in porous media. 2006 as a petrophysicist for Midcontinent. He serves as a technical
Carlos Torres-Verdín received a PhD degree in Engineering editor for SPE Reservoir Evaluation and Engineering and SPE
Geoscience from the University of California, Berkeley, in 1991. Journal.
During 1991–1997 he held the position of Research Scientist with E. C. Thomas received a PhD in Physical Chemistry from
Schlumberger-Doll Research. From 1997–1999, he was Reservoir Stanford University in 1966, completed a post-doctoral year at
Specialist and Technology Champion with YPF (Buenos Aires, Princeton University, then joined Shell Development Co. and per-
Argentina). Since 1999, he has been with the Department of Petro- formed research in the electrical behavior of shaly sands and many
leum and Geosystems Engineering of The University of Texas at other areas of Petrophysical Engineering. E. C. spent 32 years in
Austin, where he currently holds the position of Associate Profes- the Shell organization in positions encompassing field work, oper-
sor. He conducts research on borehole geophysics, formation eval- ating division engineer, research supervisor, technical training and
uation, and integrated reservoir characterization. Torres-Verdín technical oversight of Petrophysics. Upon retiring, E.C. formed
has served as Guest Editor for Radio Science, and is currently a Bayou Petrophysics to continue his work in Petrophysics. E.C. has
member of the Editorial Board of the Journal of Electromagnetic served on numerous SPE and SPWLA technical committees and
Waves and Applications, and an associate editor for Petrophysics chaired several SPE technical forums. He also served as an SPE
(SPWLA) and the SPE Journal. He is co-recipient of the 2003 and Distinguished Lec turer and presently serves as a technical
2004 Best Paper Award by Petrophysics, and is recipient of reviewer for SPE and an associate editor for SPWLA. In 2004 he
SPWLA’s 2006 Distinguished Technical Achievement Award. was awarded the SPWLA Gold Medal for technical achievement.
Sarath Devarajan received his BS in Chemical Engineering
from the Indian Institute of Technology, Madras and is currently a
Master’s student in the Department of Petroleum Engineering at
the University of Texas at Austin. His research interests include
reservoir characterization, electrical conduction in shaly sands and
simulation of fluid flow through disordered porous media.

120 PETROPHYSICS April 2007

You might also like