SPWLA 53rd Annual Logging Symposium, June 16-20, 2012

CONTINOUS ESTIMATE OF CATION EXCHANGE CAPACITY FROM

LOG DATA: A NEW APPROACH BASED ON DIELECTRIC
DISPERSION ANALYSIS
Mei Han, Schlumberger
Jaime Cuadros, Mansarovar
Cesar Augusto Patiño Suarez, Ecopetrol
Eric Decoster, Olivier Faivre, Laurent Mosse, Nikita Seleznev, Schlumberger

Society of Petrophysicists and Well Log Analysts

Copyright 2012, held jointly by the Society of Petrophysicists and Well Log adjusting the characteristics of this phase based on core
Analysts (SPWLA) and the submitting authors
This paper was prepared for presentation at the SPWLA 53rd Annual Logging data.
Symposium held in Cartagena, Colombia, June 16-20, 2012.
This approach has been validated on core
measurements from several wells and subsequently
ABSTRACT applied on 2 wells drilled in Girasol field, Colombia,
operated by Mansarovar in association with Ecopetrol.
The effect of clay on resistivity has been for many years The well is representative of reservoirs drilled on the
a major difficulty in the interpretation of shaly-sand western flank of the Middle Valley of the Magdalena
reservoirs. Waxman and Smits first formalized the River. A continuous estimate of CEC is derived from
problem in a fundamental paper in 1968, and set the measured dielectric dispersion and compared with
basis for many further works, building upon their initial laboratory measurements performed on rotary cores
approach. The common foundation remains the selected to cover a wide range of CEC values.
essential role of the cation exchange capacity (CEC) in Agreement between the measured values and values
the modelization of the clay effect on formation CC
estimated from the log is very good, opening the long
resistivity. Nonetheless, more than ten years after the sought after possibility of providing a continuous CEC
introduction of the Waxman-Smits model, Istvan estimate directly from logs. We believe that the present
Juhasz (1981) pointed out that the CEC still could not formalism will be applicable to a wide range of shaly
be determined directly from logging data. This situation sand formations, after suitable local calibration on
has largely remained the same up to now. cores.

Until today, CEC measurements have only been Beyond the obvious interest that a continuous CEC
performed in the laboratory. Recently, the analysis of holds for petrophysicists seeking to apply a shaly sand
the dielectric dispersion made possible by a new saturation equation, we also believe that such an
wireline tool holds promise for continuously estimating estimate holds considerable interest for all disciplines
the CEC from its direct effects on conductivity and that need to account for the presence and effects of
permittivity dispersion with frequency. A higher CEC different clay types, including drilling and completion.
leads to higher dispersion.

The new tool measures dielectric dispersion over four INTRODUCTION

frequencies ranging from tens of megahertz up to one Shaly-sand formations with low brine salinity constitute
gigahertz. Many physical phenomena occur close to the a challenging environment for Dielectric log
clay interface, from double layer polarization to bound interpretation. Various phenomena contribute to the
water freezing. They are characterized by the bulk clay observed dielectric dispersion. The clay surface
parameter or CEC. These effects are particularly polarization related to the electrical double layer, called
significant at low formation water salinity. A EDL, (Chelidze and Gueguen, 1999, De Lima and
comprehensive description of these effects would Sharma, 1992, Chew and Sen, 1982) amplifies the
require building a model that involves many dielectric response, leading to strong enhancement of
microscopic parameters and assumptions. Our approach the real part of the complex dielectric permittivity, even
is an alternative to overcome this complexity by up to GHz frequency range. The clay particle platy
including a highly dispersive phase representing the geometry (Wong, 1986, Tyc et al., 1988) also
CEC effects in the formation model and empirically contributes to the dielectric enhancement and the
1
SPWLA 53rd Annual Logging Symposium, June 16-20, 2012

departure from clean sandstones. In addition, water transmitter spacing, of about 1 inch (Hizem et al.,
bound to clay surfaces shows a different dielectric 2008). It is a shallow measurement not exceeding 4
behavior than free water (Or et al., 1999, Friedman, inch in DOI for a dielectric dispersion spectrum.
1998). Consequently, to estimate water saturation and
salinity, the dielectric dispersion models developed for
clean sands are no longer applicable. As a result, the A radial inversion is performed in order to remove the
accuracy of petrophysical parameters estimated will be mudcake influence and other borehole effects. In case
affected if no correction is made to the model. of shallow mud invasion (~ 2 in as invasion depth),
both the invaded zone and the virgin zone are probed by
The clay effects mentioned above are believed to be the tool (Mosse et al., 2009). By contrast, with the deep
characterized by the clay Cation Exchange Capacity invasion seen in the context of this paper, the outputs
(CEC), expressed per weight or per pore volume (Qv). here are only the permittivity ε* of the invaded zone at
The CEC is also an input for the resistivity equations of each frequency.
type ―Waxman & Smits‖ (Waxman and Smits, 1968,
Equation 1
Waxman and Thomas, 1974) and ―Dual water‖ (Clavier .
et al., 1984). Until the advent of dielectric dispersion
tools, obtaining an accurate estimate of CEC from logs It is a complex quantity (identified by the * subscript):
was a challenging problem. (Juhasz 1981, Oilfield the real part is the relative permittivity εr [] and the
Review 1989). imaginary part is the conductivity σ [S/m]; ω is the
circular frequency and ε0 the permittivity in a vacuum.
A recently introduced wireline tool (Hizem et al., 2008; Both permittivity and conductivity depend on the
Mosse et al., 2009) offers the possibility of frequency at which they are measured. They are called
continuously measuring dielectric dispersion spectrum. dispersive. The expected behavior, constrained by the
It provides an opportunity to accomplish a CEC log, underlying physical mechanisms, is to see the
based on the link between CEC and dielectric permittivity εr decreasing with frequency with the
dispersion: Dielectric dispersion, defined as the conductivity σ increasing with frequency. In particular,
permittivity and conductivity variations with frequency, the measured conductivity σ(ω) at frequency differs
is strongly dependant on CEC value. CC
from the DC conductivity σDC : it contains σDC as a
baseline and a frequency dependant contribution
In this paper, we present a new approach for shaly-sand coming from the dielectric losses εdl(ω) due to different
dielectric interpretation. Using dielectric dispersion logs polarization mechanisms:
as inputs, it is aimed at:
 Removing the bias caused by clay effects in Equation 2
classical dispersion models, thus providing ,
accurate water saturation and salinity outputs;
 Providing a reliable continuous CEC from In Figure 1, an example of the dielectric dispersion
dielectric logs. measured by the tool is presented. The data is obtained
from example well-1 of Girasol field. The black dots
The paper is organized as follows: In the first section, are the measurements at four frequencies and the red
we will give an outline of the dielectric log bars correspond to measurement uncertainties. A
interpretation and then introduce the dielectric measure of the dispersion in the frequency range of the
dispersion phenomena in the shaly-sand formation. tool is schematically represented in Figure 1 by Δε and
After that, our new approach will be described in detail, Δσ.
followed by examples applied to the Girasol field in
Colombia. Various petrophysical models (so-called mixing laws,
labeled below) relate the permittivity εr and
conductivity σ to petrophysical parameters such as the
DIELECTRIC LOG INTERPRETATION
water volume fraction , the water salinity and the
OUTLINE
rock textural parameters generically noted below
The Dielectric tool is pad-mounted with two (Archie’s cementation exponent m is an example of
transmitters and eight symmetrically located receivers. such textural parameters). The model reads
At each receiver, the measurements are recorded at four
different frequencies from 20 MHz to 1GHz (labeled F0
up to F3). The vertical resolution corresponds to the with
,
2
 SPWLA 53rd Annual Logging Symposium, June 16-20, 2012

where is the water complex permittivity, w is the origin of this enhancement is of a different nature and
water model that computes water permittivity as a brings into play local, clay- surface related phenomena.
function of water salinity , pressure is P and
temperature T (Hizem et al., 2008). is the dry
DIELECTRIC DISPERSION IN SHALY-SAND
matrix permittivity, constant with frequency and
FORMATION AT HIGH SALINITY LIMIT
varying between 5 and 6 depending on the lithology for
most shaly-sand formations. In clean rocks, dispersion is driven by the accumulation
of electrical charges at the brine - rock matrix interface:
under the action of an electric field, opposite ions travel
in the opposite direction through the brine phase but are
stopped at the isolating matrix surface. This creates an
Δσ interface polarization known as the Maxwell – Wagner
Δε effect. This dispersion enhancement depends on the
brine salinity (the more free ions there are, the larger is
the effect) and on the rock’s pore-network geometry
F0 F1 F2 F3 F0 F1 F2 F3
(the higher the rock’s tortuosity is, the higher is the
observed dispersion). In general, the SMD model
(Stroud et al., 1986) can be used to characterize this
texture-related dispersion.
Figure 1: Example of dielectric dispersion measured on
shaly-sands from Well-1, Girasol field, Colombia.
The SMD model considers that the rock is composed by
an isolating phase including both dry rock matrix and
The analysis of the dielectric dispersion is achieved hydrocarbon, as well as a conductive phase being water.
through an algorithmic inversion of a chosen The petro-physical parameters involved are the water
petrophysical model, as sketched in Figure 2. volume fraction , the water salinity and a rock
textural parameters µ. The complex permittivity of
the rock is expressed as:
CC
𝐶S𝑤
Equation 3
,

where and the function describe the

rock geometry. In clean water bearing sandstones or
Figure 2: Schematical representation of petrophysical carbonates, is equivalent to Archie’s m.
inversion.
It has been shown that when water salinity is higher
than around 30 ppk, the SMD model represents
In formations containing wet clays, the dispersion is reasonably well a shaly-sand formation’s dielectric
enhanced by specific polarization phenomena occurring properties (Pirrone et al., 2011a; Pirrone et al., 2011b).
close to the clay surface. For the same weight percent At this salinity range, the electric double layer
of dry clay, smectite, which has a high CEC value (80 ~ surrounding the clay particles is compressed to the
150 meq/100g), shows higher dispersion than kaolinite minimum thickness and the clay surface effects are also
(4-10 meq/100g) (Ellis et al., 2007). With the same clay minimized. The principal clay contribution is to
type, dispersion will increase with clay content. Hence increase the rock’s tortuosity. Thus, dispersion
in the most general case, the dielectric dispersion is not enhancement can be modeled only with textural
only sensitive to clay content but also to intrinsic parameters.
properties of the clay which is CEC value.
As shown in the Figure 3 (Figure taken from Pirrone et
Clay related dispersion depends on the water salinity. al., 2011) observed that for water salinities higher than
At higher water salinity, above about 30 ppk, dispersion 30 ppk, CEC can effectively be predicted by correlation
enhancement is limited, while at low water salinity with the value of the textural parameter µ.
dispersion enhancement is significant. This is the
opposite of rock texture polarization enhancement that
is proportional to the water salinity. It indicates that the

3
SPWLA 53rd Annual Logging Symposium, June 16-20, 2012

interval. The clay content increases: the dry weight clay

fractions calculated with the elemental capture
spectroscopy tool are 0.20, 0.33 and 0.63 respectively.
In the case of the last two data points, rotary sidewall
cores were taken at nearby depths and CEC
measurements are performed in the laboratory, yielding
CEC values of 22 and 46 meq/100g respectively. The
Figure 3: When salinity is higher than around 30 ppk µ corresponding total porosity is respectively 22, 30
generates dielectric dispersion enhancement. CEC can be
and 34 pu, estimated by triple-combo logs. Note that
predicted by correlating with µ; (Figure 9 in Pirrone et al.,
2011). increasing porosity with increasing clay content would
suggest classifying the clay as structural. Both
formation and mud salinity are low, of a few ppk. From
DIELECTRIC DISPERSION IN SHALY-SAND Figure 4 and 5, we can see that the dispersion increases
FORMATION UNDER FRESH WATER with clay content for both the relative permittivity
ENVIRONMENT and the conductivity σ. In addition, the σ value at the
The situation is totally different in fresh water lower frequency limit (MHz range) also shows
environments below 30 ppk where clay surface effects, dependence on the CEC. It increases with CEC due to
such as the EDL polarization, become dominant over the ―excess conductivity‖ of clay counter ions, resulting
the pure textural contribution to the dielectric response in a enhancement.
of the rock.

This case is illustrated in Figure 4 and 5. Three

dielectric spectra measured by the tool are selected
from well-1 at different depths, all in water zones. The
clay type is assumed to be uniform within the depth

CC

Figure 4: Relative permittivity of 3 data points. The dry clay weights are 0.20, 0.33 and 0.63 respectively from left to right
and the corresponding CEC value measured in laboratory at nearby depths for the last two data points are 22 and 46
meq/100g respectively.

Figure 5: Corresponding conductivity. The data is presented in the same order as in Figure 4.

4
 SPWLA 53rd Annual Logging Symposium, June 16-20, 2012

Petrophysical models that do not take clay contribution closer to the measured dispersion, the inverted
explicitly into account are not only inefficient in petrophysical results diverge considerably from the
honoring the dispersion seen in Figure 4 and Figure 5 known, true values. and salinity are systematically
but also provide biased estimation of water filled over-estimated by this SMD model inversion. Over-
porosity and water salinity. This is because textural estimation of the salinity increases proportionally with
effects are not sufficient to reproduce the measured clay content, being 5, 18 and 26 ppk while both mud
dispersion. and formation water salinity are known to be lower than
15 ppk. The model predicted CEC’s for all three data
In Figure 6, the SMD model with known inputs is points reach a plateau inherent to the model at 30
displayed on top of the three corresponding measured meq/100g whereas the true CEC values should be close
dielectric spectra. The inputs are the total porosity and a to 15, 22 and 46 meq/100g respectively.
water salinity of 5 ppk. The textural parameter is
150 1
fixed at 2.2, 2.3 and 2.5 respectively, based on the CEC Depth-1
- µ correlation shown in the Figure 3. Important Depth-2, CEC ~ 22
0.8
Depth-3, CEC ~ 46
discrepancies are observed between the measurement Depth-1,SMD Inversion

Conductivity [S/m]
100

Permittivity [...]
Depth-2,SMD Inversion
and the predictions of the textural model, especially for Depth-3,SMD Inversion
0.6

the spectra with the highest CEC value: model-

predicted dispersion amplitude and DC conductivity are 50
0.4

much smaller than actual. 0.2

150 1
0 0
1 2 3 1 2 3
Depth-1 10 10 10 10 10 10
Depth-2, CEC ~ 22 Frequency [MHz] Frequency [MHz]
Depth-3, CEC ~ 46 0.8
Depth-1,SMD forward
Conductivity [S/m]

100 Figure 7: SMD model inversion fit. Blue dots are the tool
Permittivity [...]

Depth-2,SMD forward
0.6
Depth-3,SMD forward
measured dielectric spectra with clay dry weight of 0.2;
0.4
Green and red dots refer to the ones with clay dry weights
50 of 0.33 and 0.63 respectively. Their corresponding SMD CC
0.2 model inversion reconstructions are the solid line with the
same color.
0 0
1 2 3 1 2 3
10 10 10 10 10 10
Frequency [MHz] Frequency [MHz] These biases are not related to the use of a particular
textural model. Several different models that only take
Figure 6: SMD model forward fit. Blue dots are the tool into account the rock texture have been tested, and all
measured dielectric spectra with clay dry weight of 0.2; present the same biases. A single frequency CRIM
Green and red dots refer to the ones with clay dry weights model, that does not take any textural parameter into
of 0.33 and 0.63 respectively. Their corresponding SMD account also fails to honor the data, as expected, and
model fit is the solid line of the same color.
also suffers from the same biases in water-filled
As a second step, we can fit the SMD model to the porosity and water salinity estimates. Consequently, a
measured dispersion, following the methodology shown multi-frequency model is necessary to reach a good
in Figure 2. The resulting parameters are presented in petrophysical analysis in shaly-sand formations.
Table 1. The dielectric spectra are then reconstructed
with the inverted petrophysical parameters, as shown in
Figure 7. Although the reconstructed curves are much

Log estimated values SMD model inversion outputs

WCLA
PHIT Salinity (clay dry Core CEC Salinity µ CEC
(pu) (ppk) (meq/100g) (pu) (ppk) (meq/100g)
weight)
Depth1 22 0.2 15* 5.1 24 5 3.2 30
Depth2 30 0.3 22 5.2 32 18 2.6 29
< 15
Depth3 34 0.6 46 5.3 39 26 2.9 30
Table 1: SMD model inversion results: petrophysical parameter outputs. * The CEC value of 15 meq/100g is not directly
measured on the core data, but estimated with WCLA, CEC ≈ 0.2/0.6×46)

5
SPWLA 53rd Annual Logging Symposium, June 16-20, 2012

In this section, we have shown that in fresh water shaly- , Equation 4

sand environments, textural models can no longer be
applied. A new dielectric interpretation approach is
The critical frequency fc locates the point of steepest
therefore needed in order to remove the bias caused by
dispersion. Based on core dielectric spectra and
clay effects, thus provide accurate water saturation and
observations from log data, the steepest dispersion
salinity and provide a reliable continuous CEC
occurs generally around 200 MHz. Figure 8 b shows the
assessment from dielectric logs. In the next section, we
SHSD model inversion misfit on varying fc value from
will present such a model.
50 MHz to 1 GHz on log data at Depth 3. Misfit is
defined as the normalized squared residuals, the
minimum of which is observed at around 200 MHz.
DESCRIPTION OF THE NEW APPROACH
The high frequency limit of the dispersive phase is
set to be 5. In this limit, the permittivity results
Clay effects on dielectric response. Much theoretical exclusively from the electronic and atomic polarization.
work was done during the last 50 years to try and
The low frequency conductivity limit of the dispersive
explain some aspects of the clay effects on electric and
phase, , is computed the same way as introduced in
dielectric properties of a shaly sand formation. They
equation (16) from Clavier’s paper (Clavier et al.,
investigated various physical phenomena:
1984). The low frequency limit of the dispersive phase
1. The Electrical double layer (EDL) polarization leads
should be high enough to characterize the clay surface
to an enhancement of the dielectric dispersion. EDL
effects dispersion. Experimentally, it was reported that
becomes dominant at low frequency range, below 10
EDL polarization is responsible for large (~ 1000) value
MHz (Chelidze and Gueguen, 1999; De Lima and
of at low frequency for rocks containing clay
Sharma, 1992; Chew and Sen, 1982);
particles (Chew and Sen, 1982). Thus, the value of the
2. The clay particle’s platy shape and clay aggregates
low frequency limit has been optimized on the database
fractal surface both increase the dispersion (Wong,
that will be described afterward.
1986; Tyc et al., 1988);
3. Finally, clay bound water has a lower dielectric CC
F0 F1 F2 F3
700 90
12
constant and a lower relaxation frequency (Or et al., 80
600
1999; Friedman, 1998). 10
70
misfit

500
(S/m)

60
8
Permittivity

inversion

The clay effects causing dispersion enhancement are 400 50

Conductivity

6
40
due to the combination of these mechanisms. A 300
Model

4
30
comprehensive model including all of these effects has 200
fc = 50 MHz 20
2
been attempted but it involves many microscopic 100 fc = 200 MHz
10
fc = 350 MHz
parameters and assumptions, rapidly becoming 0
10
7
10
8 9
10
0
0 7
50
10 100 8
200
10 500 9
1000
10
impractical. An efficient model, given the Frequency (Hz) fc [MHz]
Frequency (Hz)

measurements available, should not have more than 5 a b

or 6 free parameters to invert for. Figure 8: a: The higher the fc is, the higher the maximum
dispersion will shift to the high frequency range; b: SHSD
model inversion misfit drawn as a function of fc
Our present approach (called the SHSD model in this
varying from 50 MHz to 1 GHz.
paper) is an attempt to overcome the complexity of the
problem. The fundamental idea is to handle the
dispersion enhancement by including a highly The SMD model is used to mix the three different
dispersive phase with a complex permittivity that phases: water, clay-related and matrix/oil. The clay-
represents the clay-induced dispersion. The textural related dispersive phase is firstly combined with water
effect will still be included in the model in a second in a linear way, thus defining a combined conductive
step to characterize the rock geometry. phase :
The new approach. Many different shapes can be Equation 5
chosen to produce a dispersive phase representing the
clay-induced dispersion . After several trials, we have
selected the Debye relaxation equation (Debye 1929), Where is weighted by . The above combination
based on the goodness of the fit to measured spectra. results in increasing not only the dielectric dispersion,
is described in Equation 4. but also the DC conductivity, thus, captures exactly the
6
 SPWLA 53rd Annual Logging Symposium, June 16-20, 2012

clay dispersion phenomena missing in the classical

textural model.

The last step consists in integrating with non-

conductive dry rock /oil to generate the fluid bearing
rock’s dielectric property :

Equation 6

The corresponding schematic representation of the Figure 9: Correlation between CEC and inverted pd.
model can be found in the Table 2.

The petrophysical parameters obtained by the inversion Figure 9 shows a cross plot of CEC from core versus
of this model include not only , the water salinity dielectric dispersion analysis through the SHSD model
and textural parameter , but also the newly on the same cores. A clear correlation emerges, that can
introduced dispersive phase fraction pd. be parameterized by a hyperbolic tangent function. The
maximum CEC limit is set to be 60 meq/100g because
SMD Model New Approach from application point of view, less precision is
Conductive requested in CEC estimation in shale. At low CEC, we
Water
Conducti Conductive
decided to apply a cutoff : when , is less than 0.01,
ve Water is considered to be 0.
Phase High Dispersive
Phase
The SHSD model inversion is now applied on the data
Isolating presented previously (Figure 4, Figure 5 and Table 1).
Petrophysical parameters estimation from both the CC
Dry Matrix Dry Matrix
Phase
Rock By SMD By SMD SMD model and the SHSD model are compared in
geometry function and µ function and Table 3 to 6. For inverted and water salinity, the
Table 2: Model construction; is the fraction of the SHSD model outputs are closer to the expected value,
dispersive phase. and no clear CEC dependant bias is observed. This
contrasts with the SMD model, meaning that the clay
It is expected that the inverted dispersive phase fraction effects are corrected by the new approach. The CEC
pd will increase with the CEC. The higher the CEC of a from the SHSD model matches reasonably well the
rock, the higher the clay effects are, and this is expected value, shown in Table 5. Besides, the SHSD
characterized by the dispersive phase. To evaluate this model reconstructed dielectric spectra fit better the
assumption, 75 core data were collected worldwide and measured one, especially for the permittivity at low
both CEC and dielectric spectra were measured in the frequency (Figure 10).
laboratory. The core CEC covers a range from 0 to 40
meq/100g.
Water salinity outputs (ppk)

Water filled porosity outputs (pu) Mud &

Formation SMD SHSD
Log Salinity (ppk)
estimated SMD SHSD Depth1 5 3
values (pu) Depth2 < 15 18 8
Depth1 22 24 21 Depth3 26 7
Depth2 30 32 29 Table 4: Salinity predicted by SHSD model.
Depth3 34 39 35
Table 3: predicted by SHSD model.

7
SPWLA 53rd Annual Logging Symposium, June 16-20, 2012

Well-1. Well-1 penetrates an alluvial fan deposit, with

CEC outputs (meq/100g) fresh formation water at about 10-15 ppk generally
found in the reservoir, with however significant
Core CEC variations with depth. The average porosity is about 20
SMD SHSD
(meq/100g) p.u. in the reservoir sands, and increases with shale
Depth1 15* 22 9 content. About 30 preserved cores were taken in this
Depth2 22 29 25 well. Standard core analysis, Dean-Stark, FTIR and
Depth3 46 34 49 CEC measurement were performed on these cores. The
Table 5: CEC predicted by SHSD model. * value not clay type as revealed by cores is mainly illite and
directly measured on the core data, but estimated kaolinite with a small contribution of smectite
approximately by WCLF value, CEC ≈ 0.2/0.6×46). following the illite content. Feldspars are also present in
varying quantities. There is very little carbonate in this
150 1
well .
Depth-1
Depth-2, CEC ~ 22
Depth-3, CEC ~ 46 0.8 A full log suite was acquired in this well, including
Depth-1,SHSD fit
capture spectroscopy, spectral gamma ray, neutron
Conductivity [S/m]

100 Depth-2,SHSD fit

Permittivity [...]

Depth-3,SHSD fit 0.6

density and induction, in addition to the dielectric log.
The available logs allowed to create an ELAN lithology
0.4
50 model following the core data. This point is important
0.2 to ensure that correct total porosity and matrix
permittivity are obtained. Figure 12 shows the
0
10
1
10
2
10
3
0
10
1
10
2 3
10
comparison between minerals’ weight percent as
Frequency [MHz] Frequency [MHz] measured on core and the same weight percent
Figure 10: SHSD reconstructed dielectric spectra on the 3 estimated with the capture spectroscopy and other
log data points presented in Figure 4, Figure 5 and Table available logs. Figure 14 is a log plot with processing
1.
results for well-1.The resulting lithology model can be
found in the track named ―Volumetric Analysis‖. An
obvious dependence between clay content and dielectric CC
Discussions. Our present approach here is a pragmatic
way to overcome the complexity related to clay surface dispersion can be seen: the dielectric dispersion in
phenomena, by including a highly dispersive phase to permittivity and conductivity are shown in the last two
represent the clay induced dispersion. tracks where the color shadings correspond to their
variation between F0 and F1, F1 and F2, then F2 and
We believe that with the continuous advances in F3 respectively. The dispersion enhancement in front of
dielectric dispersion theoretical and experimental works clay-bearing intervals can be clearly seen.
a phenomenological description of the dispersive phase
will replace the current empiric formulation, but justify In the first three tracks on the left, the processing
the bulk phase strategy chosen in this paper. quality flags are displayed, for the petrophysical
analysis, the mudcake/standoff dielectric measurement
Another important field of study is related to research corrections and the dispersion analysis respectively. In
into another clay parameter, Qv, closely related to CEC. the borehole track, the caliper and the SP are displayed.
To clarify whether CEC or Qv is more efficient to A pink color indicates deflection of the SP to the left,
describe the contribution of the clay to the dielectric hence formation water more saline than the mud
dispersion and should be an output of the dielectric filtrate.
analysis, further study is ongoing.
In Figure 14, the curves FSXO_ADT, PWXO_ADT,
SWXO_ADT and MN_XO_ADT stand for the outputs
from the SHSD model inversion. In the ―Salinities‖
APPLICATION EXAMPLES track, the water salinity FSXO_ADT is the invaded
zone water salinity. It is expected to stand between the
mud filtrate and the formation salinities in the sand
The approach described in the previous section is
interval. The salinity limits can be observed in Figure
applied on two wells from the Girasol Field
13, corresponding to FSXO_SHSD . The minimum
representative of the conditions found in the western
value is found in sands with low CEC value and the
flank of the Middle Valley of the Magdalena River.
maximum value in shales ,around 15 ppk.
They are labeled as well-1 and well-2.

8
 SPWLA 53rd Annual Logging Symposium, June 16-20, 2012

Dielectric tool water filled porosity, PWXO_ADT, total

porosity PHIT from ELAN model, and core results are
shown in the ―Porosity Analysis‖ track. PWXO_ADT
matches PHIT in the shale and in the water zones, as
can be seen in the top of the interval.

The trends observed on the database when the

dispersion interpretation model does not take clay effect
into account are confirmed in this well. In the Figure 11
and Figure 13, cross plots show the difference between
the mono-frequency analyses (PWXO_CRIM,
FSXO_CRIM) without textural or clay effect and the
multi-frequency analysis (PWXO_SHSD,
FSXO_SHSD) taking clay effects into account. The
over-estimation of water filled porosity by a mono-
frequency analysis can reach up to 5 to 10 p.u. for high
CEC shales. Salinity overestimation is large also,
leading to values not compatible either with the known
formation salinity or with the filtrate salinity. By
contrast, the SHSD model output salinity is within the
expected range: FSXO_SHSD is close to the mud Figure 12 Comparison between minerals’ weight percent
filtrate salinity in the sands and close to the formation as measured on core and weight percent estimated from
water salinity in the shales. the capture spectroscopy log analysis.

CC
PWXO_CRIM

FSXO_CRIM
CEC_SHSD

CEC_SHSD
PWXO_SHSD FSXO_SHSD
Figure 11: cross plots between mono-frequency analysis Figure 13: Salinity: cross plots between FSXO_CRIM and
(PWXO_CRI) and multi-frequency analysis FSXO_SHSD;The color corresponds to the CEC value
(PWXO_SHSD). The color corresponds to the CEC value from SHSD model estimation.
from SHSD model estimation.

9
SPWLA 53rd Annual Logging Symposium, June 16-20, 2012

X050

X100

CC
X150

X200

X250

X300

Figure 14: Interpretation results of Well-1.

10
 SPWLA 53rd Annual Logging Symposium, June 16-20, 2012

In the ―Textural Properties‖ track, the rock textural formation water salinity, when there is no moveable oil.
parameter MN_XO_ADT and CEC_ADT are The CEC is also slightly increasing from bottom as
displayed, together with the core measurement results. well.
There is a very good match between the CEC measured
on core and CEC_ADT. The middle reservoir has no moveable oil, and shows
higher water salinity. Finally, the upper reservoir, even
The two ―Saturation‖ tracks require some careful if it has a small interval with no moveable oil on its
explanations. In the ―Saturation‖ track, the results of a bottom, shows much better reservoir quality with
conventional resistivity-saturation analysis are significant moveable oil.
presented. In such fresh water condition, the use of
saturation equations is challenging. For the sake of The remaining tracks are left to support the
simplicity, we chose to use Archie’s equation with the interpretation. In the depth track, the gray shading
cementation and saturation exponents set to indicates mudcake/standoff thickness. The borehole is
MN_XO_ADT. SXOT, standing for the invaded zone rough, and in poor shape especially in front of the
water saturation, is obtained using the micro resistivity shaliest sections. The multi spacing of the dielectric
and mud filtrate salinity. SWT, being a deep zone tool allows us to correct for this roughness and obtain
saturation, is obtained from RT, after adjusting the robust results in most of the intervals.
formation salinity based on RW from SP and salinity
limits provided by the dielectric dispersion tool shown There is a good match between the reconstructed RXO
previously in Figure 13. This point is important, as this from the dielectric tool, RXO_ADT and the Micro-
indirect use of the dielectric results may have been resistivity RXO. RXO_ADT is obtained by taking the
overlooked in many cases. These physical constrains low frequency limit of the conductivity dispersion fit.
are: Having a good match between RXO and RXO_ADT is
 SWT should be smaller than or equal to the not only useful for quality control, but also confirms
invaded zone water saturation seen by the dielectric that both tools read the same invaded zone.
tool SWXO_ADT. Indeed, in water-based-muds,
the shallow zone should contain more water than
the deep zone. Well-2. Well-2 benefited from the same log suite as CC
 The formation salinity should be compared to well-1. Only conventional analysis and Dean-Stark
FSXO_ADT as well. As mentioned earlier, the saturation estimation were done on the cores taken in
FSXO_ADT stands in between the formation and this well. CEC values are not available.
filtrate salinity.
Figure 16 present the results of the interpretation based
In the ―Saturation_C‖ track, SWT and SWXO_ADT are on dielectric logging information combined with
displayed, together with Dean-Stark oil saturation conventional logs. Once again, there are significant
―Soil‖ and water saturation ―Sw‖. The comparison is differences between the conventional analysis in the
good, even if the water saturation from Dean-Stark may first saturation track, and the analysis made with
underestimate the true water saturation, due to water dielectric logging in the second saturation track. Based
losses during the core handling process. on resistivity analysis alone, many sands would have
been bypassed. The oil-bearing sandstone reservoir at
There is a clear difference in interpretation results X125-X150ft is an example. This reservoir has no
between a conventional saturation analysis and an moveable oil, but it contains about 20% residual oil, as
analysis done with the dielectric tool results. The confirmed by the Dean-Stark core analysis.
indications of movable oil now are coherent and can be
trusted. A detailed view of a sand sequence will follow Some of the reservoirs with moveable oil have most
in the next section to further examine these results. probably their bottom part in water, for example
intervals at X110-X120ft and X160-X190ft.
In Figure 15 a detailed view of a sand sequence is
presented. In spite of the large variations in resistivity, In this well, the sand bodies are shalier than in Well-1.
the dielectric-tool shows constant remaining oil The volumetric analysis, spectral gamma ray, density-
saturation on the three sands. The bottom sand has neutron and CEC curve from dielectric dispersion
some moveable oil on top, but none on bottom. The analysis are coherent and confirm that evaluation. The
salinity estimated by dielectric tool is also changing, dielectric-based CEC has much more character due to
remaining low, i.e. close to filtrate salinity when there its better vertical resolution.
is moveable oil, and becoming higher, i.e. closer to
11
SPWLA 53rd Annual Logging Symposium, June 16-20, 2012

CC

Figure 15 Zoom on the sand sequence of Well-1.

12
 SPWLA 53rd Annual Logging Symposium, June 16-20, 2012

X050

X100

X150
CC

X200

X250

Figure 16: Interpretation results of Well-2.

13
SPWLA 53rd Annual Logging Symposium, June 16-20, 2012

Debye P, 1929, Polar Molecules, Dover, New York.

CONCLUSION
A new approach to describe fresh-water shaly-sand De Lima O. A. L. and Sharma M. M., 1992, A
formation dielectric behavior has been introduced and generalized Maxwell-Wagner theory for membrane
devised in this paper. In addition to quantitative water- polarization in shaly sands, Geophysics, 57, 431-440.
filled porosity and water salinity, the new dielectric
dispersion model also provides a water phase tortuosity Ellis D. V, Singer J. M, 2007, Well logging for earth
parameter and a continuous estimate of the CEC. This scientists, Springer.
approach has been validated on core measurements
from several wells covering a range of conditions. Haslund E and Nøst B, 1998, Determination of porosity
and formation factor of water-saturated porous
The essence of the new approach is to separate the clay specimens from dielectric dispersion measurements,
dispersion effect into a single conductive and dispersive GEOPHYSICS, 63, 149-153.
phase. The definition of this phase is based on available
core data. With the continuous progress in experimental Hizem M, Budan H, Devillé B, Faivre O, Mossé L and
and theoretical work on dielectric dispersion physics Simon M., 2008, Dielectric dispersion: a new wireline
applied to well logging, this definition may change in petrophysical measurement: SPE paper No 116130.
the future and account for potential clay type or
anisotropy effects. Juhasz I, 1981, Normalised Qv - The key to shaly sand
evaluation using the Waxman-Smits equation in the
The application of the current interpretation absence of core data, Transactions of the SPWLA
methodology on two wells from Colombian fields has Annual Logging Symposium.
demonstrated the significant benefit of using this new
approach. A robust, coherent picture of reservoir Kenyon W. E, 1984, Texture effects on megahertz
properties emerges in intervals where conventional dielectric properties of calcite rock samples: Journal of
interpretation is subjective. Comparison of continous Applied Physics, 55, 3153–3159.
estimates of petrophysics parameters and core analysis CC
results is very good. Mosse L, Carmona R, Decoster E, Faivre O and Hizem
M, 2009, Dielectric Dispersion Logging In Heavy Oil:
We believe that the present formalism will be A Case Study From The Orinoco Belt: Transactions of
applicable to a wide variety of fresh-water shaly-sand the SPWLA Annual Logging Symposium.
formations, after suitable local calibration on cores.
Or D, Wraith J. M, 1999, Temperature effects on soil
bulk dielectric permittivity measured by time domain
reflectometry: A physical model, Water Resources
Research, 35, 371-383.
REFERENCES
Anonymous, 1989, Archie III: electrical conduction in Pirrone M, Han M, Bona N, Borghi M, Galli M. T,
shaly sands, Schlumberger Oilfield Research, 1(3), 43- Pampuri F, Faivre O, Hizem M, Kherroubi J, Mossé L,
53. 2011a, A novel approach based on dielectric dispersion
measurements to evaluate the quality of complex shaly-
Chelidze T. L and Gueguen Y, 1999, Electrical sand reservoirs, SPE 147245-PP.
spectroscopy of porous rocks: a review—I. Theoretical
models, Geophysical Journal International, 137, 1-15. Pirrone M, Bona N, Galli M. T, Pampuri F, Seleznev
N., Han M, Mosse L, Faivre O, Hizem M, 2011b,
Chew W. C and Sen P. N, 1982, Dielectric Experimental Investigation and Modeling of the
enhancement due to electrochemical double layer: Thin Electrical Response of Thinly Layered Shaly Sands (10
double layer approximation, Journal of Chemical MHz – 1 GHz), Trans. Int. Symp. Soc. of Core
Physics, 77, 4683-4693. Analysts, paper SCA2011-10.

Clavier C, Coates G, Dumanoir J, 1984, Theoretical and Stroud D, Milton G. W, De B. R, 1986, Analytical
Experimental Bases for the Dual-Water model for model for the dielectric response of brine-saturated
interpretation of Shaly Sands, SPE Journal, 24, 153- rocks, Physical review B, 34, 5145-5153.
168.
14
 SPWLA 53rd Annual Logging Symposium, June 16-20, 2012

Tyc S, Schwartz L. M, Sen P. N, and Wong P. Z, 1988, Ollivier Faivre graduated from Ecole Polytechnique in
Geometrical models for the high-frequency dielectric 1974. He joined Schlumberger in 1976 and started as a
properties of brine saturated sandstones, Journal of field engineer. In 1982, he entered the Reservoir
Applied Physics, 64, 2575-2582. Description group, in Singapore. After various field
positions in interpretation, he joined Engineering in
Waxman M. H and Smits L. J. M, 1968, Electrical 1991 and has been working since then on the
Conductivities in Oil – Bearing Shaly Sands, SPE development of petrophysical interpretation for new
Journal, 243, 107-122. technology. He is presently Petrophysics Advisor in the
Schlumberger Product Development Center in Clamart,
Wong, P. Z and Howard J, 1986, Surface roughening France.
and the fractal nature of rocks, Physical Review Letters,
Laurent Mossé has an engineering degree from
5, 637-64
Supélec obtained in 1997 and a Ph.D. in fundamental
physics from the Centre of Atomic Energy (CEA) in
“ABOUT THE AUTHOR” SECTION Saclay obtained in 2002. He joined Schlumberger the
same year and first worked on nuclear density tools. He
then turned toward electromagnetic tools and led the
Mei Han received her Ph.D in Physics from Ecole physics and interpretation team for Dielectric-Scanner
Polytechnique (France) in 2008. Since then she has project until end 2011. He currently works as Principal
been working as an interpretation engineer in Petrophysicist in Latin America.
Schlumberger in Paris.
Nikita Seleznev is a Principal Research Scientist at
Jaime Cuadros obtained a geologist degree at the Schlumberger-Doll Research (SDR), Boston,
National University in Bogota Colombia in 1983. He Massachusetts, USA. He conducts research in dielectric
worked in the Petroleum area with Ecopetrol, BP, and resistivity logging tools and techniques,
EXXON, Conoco Phillips and Repsol. For the last three petrophysics of conventional and unconventional
years he has been with Mansarovar Energy Colombia reservoirs. He has been developing interpretation
ltd as the G & G Manager. products that directly measure water volume and rock CC
textural information for the Dielectric Scanner tool. He
Cesar Patiño is a Senior Petrophysicist working for joined Schlumberger in 1998 as a Wireline field
Ecopetrol. He has 15 years working in Formation engineer. Nikita obtained Ph.D. in petrophysics from
Evaluation. He has worked as a Schlumberger Field the Delft University of Technology, The Netherlands.
Service Manager (Latin America- Africa-Europe),
Weatherford Operations manager (Latin America) in
the Wireline segment, and with Oxy and Ecopetrol in
the Reservoir department. Currently working and
giving support to all Colombia basins, and international
reservoir characterization projects. Cesar received a BS
Degree in Petroleum Engineering in 1997 from the
Universidad Industrial de Santander.

Eric Decoster obtained an engineering degree at Ecole

Centrale in Paris in 1976 and an MSc in civil
engineering, flow through porous media, at the
University of Wisconsin, Madison, in 1977. He joined
Schlumberger in 1978. He was recently assigned as
Principal Petrophysicist in the Faja Regional
Technology Center located in Barcelona, Venezuela
and is now Petrophysical Advisor for Schlumberger in
Latin America. He is currently serving as the SPWLA
Latin America Director.

15