SPE 71723

Determination of Fit-For-Purpose Saturation Exponents

Paul F Worthington, SPE, Gaffney, Cline & Associates

Copyright 2001, Society of Petroleum Engineers Inc.

methodology. In many cases of routine petrophysical
This paper was prepared for presentation at the 2001 SPE Annual Technical Conference and interpretation, it is not.
Exhibition held in New Orleans, Louisiana, 30 September–3 October 2001.

This paper was selected for presentation by an SPE Program Committee following review of
information contained in an abstract submitted by the author(s). Contents of the paper, as
The saturation exponent n is an important interpretative
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to parameter in petrophysics, because it defines a quantitative
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at relationship between the electrical properties of a reservoir
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
rock and its water saturation Sw, i.e. the faction of pore space
for commercial purposes without the written consent of the Society of Petroleum Engineers is that is filled with water as opposed to hydrocarbons.1 This
prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300
words; illustrations may not be copied. The abstract must contain conspicuous relationship is often expressed in terms of resistivity index Ir,
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.
the ratio of the resistivity of a partially water-saturated
reservoir rock Rt to the resistivity R0 that the rock would
possess at conditions of full water saturation with the same
electrolyte, as follows:
Abstract
Algorithms that are commonly used for the evaluation of Ir = Rt /R0 = C0 /Ct = Sw–n (1)
water saturation can require different input values of
saturation exponent for meaningful application to the same
where the conductivities Ct and C0 are the reciprocals of
reservoir rock. Further, the differences in the required
resistivities Rt and R0, respectively. Equation (1) can be
saturation exponent can vary according to the reservoir
rewritten:
properties themselves. By taking account of these different
input requirements, it is shown that the discrepancies between
Ct = C0 Swn (2)
predicted water saturations obtained using dissimilar
petrophysical algorithms can be confined to a restricted range.
For a reservoir rock that is fully water-saturated, a formation
The procedure is synthesized for easy operational application
resistivity factor, or simply formation factor, F can be defined
of an ordered, interactive method that allows a fit-for-purpose
as follows:
saturation exponent to be identified. Through this process, the
sensitivity associated with the choice of an interpretative
F = R0 /Rw = Cw /C0 (3)
model for the evaluation of water saturation is demonstrably
contained. The application of the method therefore increases
where Rw is the resistivity of the saturating aqueous electrolyte
confidence in petrophysical interpretation by limiting
and Cw is its conductivity. This definition allows equation (2)
uncertainty.
to be rewritten in the form:
Introduction
Ct = (Cw /F) Swn (4)
A reliable knowledge of saturation exponent is a prerequisite
for the meaningful petrophysical evaluation of water
For the limiting case of fully water-saturated conditions,
saturation. Up to now, saturation exponent has been seen
equation (4) reduces to the following alternative form of
solely as a property of a reservoir rock that is obtained through
equation (3):
special core analysis. However, there are growing indications
that saturation exponent is also dependent upon the nature of
C0 = (Cw /F) (5)
the petrophysical algorithm that is to be used subsequently for
the evaluation of water saturation from well logs. It is
Equations (1) - (5) relate to reservoir rocks that are
therefore imperative that a chosen saturation exponent be fit
electrochemically clean, i.e. conduction takes place solely
for purpose, not only in terms of reservoir properties but also
through the free ions within the formation water. These
from the standpoint of petrophysical interpretative
conditions have been loosely associated with the absence of
2 PAUL F WORTHINGTON
clays and clay minerals, and this is a fair assumption where
the formation water has a high salinity. They can be called
“Archie” conditions after the author of that seminal paper.1
Otherwise a reservoir rock shows conduction phenomena that
do not satisfy equations (1) - (5), because of the electrical
effects of shaliness or of a low-salinity formation water, or
both. In these cases, equations (1) - (5) have been extended to
include at least one additional conductivity term that is
associated with shale and/or low-salinity effects. With one
additional conductivity term X, equation (4) becomes:
Ct = (Cw /F*) Swn* + X Swq* (6)
where q* is a saturation exponent within the additional
conductivity term. For the limiting case of fully water-
saturated conditions, equation (6) reduces to:
C0 = (Cw /F*) + X (7)
The quantity X can be written for mathematical convenience
as:
X = x/F* (8)
Here x is a scaled parameter that quantifies the conduction
associated with shale and/or low-salinity effects. Equations
(6) – (8) relate to reservoir rocks wherein shale and/or a low-
salinity electrolyte give rise to a significant extra conductivity
term X. These conditions can be termed “non-Archie”
conditions. Equations (6) and (7) can be seen as general forms
of equations (4) and (5), respectively. In equations (6) – (8)
the parameters F, n and q have been written F*, n* and q*,
respectively, to indicate that they have to be quantified by
taking account of non-Archie effects. This convention is
adopted throughout.
Equations (2) and (6) are used to evaluate water saturation
from electrical measurements of Archie and non-Archie
reservoir rocks, respectively. Equation (2) is standard for
partially water-saturated Archie reservoir rocks. In contrast,
over the years many authors have presented interpretative
algorithms to describe quantitatively the electrical
characteristics of partially water-saturated non-Archie
reservoir rocks.2 Some of these equations have taken the form
of equation (6): others have taken different forms. All contain
Archie and non-Archie terms. In particular, each equation
takes a position either on the value of exponents such as q* or
on how such an exponent relates to n*. It has long been
appreciated that these multifarious equations lead to
significantly different estimates of water saturation and thence
hydrocarbon saturation.3
One of the reasons why these diverse estimates of water
saturation have not been reconciled is that the saturation
exponent n or n* has generally been regarded exclusively as a
property of the reservoir rock. The same saturation exponent
has been applied regardless of the character of the equation
into which it is to be inserted. The situation has been
exacerbated by the practices of using default values of n or n*
SPE 71723
and basing the selection of algorithms on company culture,
when a consideration of reservoir physics might have
indicated otherwise. Yet, there is evidence to suggest that
each petrophysical model underpinning an available
interpretative algorithm for Sw does require a saturation
exponent that is appropriate to the model itself. In other
words, it is important to identify how the saturation exponent
should be determined for a given interpretative model and to
do so in a way that is compatible with the nature and
requirements of the model. This suggests that a saturation
exponent is not just specific to a particular reservoir rock but
is also specific to the petrophysical model and thence to the
interpretative algorithm that is to be used.
The purpose of this paper is to demonstrate that the use of
model-specific saturation exponents does lead to evaluations
of water saturation that can be reconciled when different
interpretative algorithms are used. To do this, it is necessary
to have some criterion for deciding whether a reservoir rock
should be treated as an Archie reservoir or a non-Archie
reservoir. This criterion is established through the continuum
concept of the electrical character of reservoir rocks.4
Continuum Concept
The continuum concept is founded on the premise that
intergranular reservoir rocks show a continuum of electrical
behavior over a wide range of formation water conductivity
and electrical manifestation of shaliness. This phenomenon is
most readily (but not uniquely) described for the case of fully
water-saturated siliciclastic sediments, on which uncertainties
in the saturation exponent have no bearing. Conduction
through a fully water-saturated reservoir rock is represented
by the parallel resistor model described by equations (7) and
(8), but rewritten in the following form by substituting for C0
using equation (5):
F /F* = Cw / (Cw + x) (9)
For Archie conditions, the ratio F/F* ≈ 1, and this requirement
is attained where x is sufficiently small and Cw is sufficiently
large. For non-Archie conditions, F/F* << 1, and this
situation can prevail where x is sufficiently large or Cw is
sufficiently small. For practical purposes, Archie conditions
are assumed where F/F* ≥ 0.9 and therefore equations (1) –
(5) only apply where this inequality is satisfied. Where F/F*
< 0.5, equations such as (6) – (8) have to be handled with
caution and they can break down in extreme cases.5
Equation (9) is depicted graphically in Fig 1, which also
contains the demarcation lines at F/F* = 0.9 and F/F* = 0.5.
For present purposes, I will refer Fig. 1 to a notional
measurement temperature of 25°C. In this paper, I will
consider only those reservoir rocks that lie within the shaded
area of Fig. 1. This domain is bounded by the line F/F* = 1,
at the limit, and by the line F/F* = 0.5, for reasons already
stated. It is also bounded by the curve Cw = 3.3 S/m (the
electrolyte conductivity at 25°C that broadly corresponds to
20000 mg/liter NaCl, the minimum water salinity used by
SPE 71723 DETERMINATION OF FIT-FOR-PURPOSE SATURATION EXPONENTS
Archie1), and by the curve Cw = 25.0 S/m (close to the
maximum attainable electrolyte conductivity of an NaCl
solution at 25°C). For any reservoir rock within this domain,
the non-Archie formation factor F* becomes numerically
indistinguishable from the Archie formation factor F as x → 0
and thence equation (6) reduces to equation (4). Outside this
domain the situation is less clear, because lithologically-clean
reservoirs that contain low-salinity waters can show pseudo-
shale effects, i.e. F/F* can be significantly less than unity even
where x is very small, as Fig. 1 indicates.
Water-Saturation Algorithms
In the most common practice, where a saturation exponent is
required as input, my earlier type classification of non-Archie
algorithms still accommodates most of the models that have
been proposed for the evaluation of water saturation from
electrical measurements.2 The type classification is as
follows.
Type 1. The algorithm expresses total rock conductivity as
the sum of Archie and non-Archie conductivity terms, but
water saturation appears only in the Archie term and the
exponent q* = 0:
Ct = α Swn* + γ (10)
where α is a generalized coefficient of water saturation in the
Archie conductivity term and γ is a generalized form of the
non-Archie conductivity term. Examples of type-1 models are
those of Hossin6 and Givens.7
Type 2. Again, the algorithm expresses total rock
conductivity as the sum of Archie and non-Archie
conductivity terms, but here water saturation appears in both
terms, so that:
Ct = α Swn* + γ Swq* (11)
where γ is now a generalized coefficient of water saturation in
the non-Archie conductivity term. Examples of type-2 models
are the modified Simandoux,8 Waxman-Smits9 and dual-
water10 methods.
Type 3. The algorithm expresses total rock conductivity as
the sum of three conductivities, an Archie term, a non-Archie
term and a coupling term that brings together both Archie and
non-Archie conductivities. Water saturation does not appear
in the non-Archie term, so that:
Ct = α Swn* + β Swp* + γ (12)
where β is a generalized coefficient of water saturation in the
coupling term for Archie and non-Archie conductivity and p*
is the saturation exponent within the same term. Examples of
type-3 models are those of Husten & Anton11 and Schwartz
and Sen.12
3
Type 4. Again, the algorithm expresses total rock
conductivity as the sum of three conductivity terms, but here
water saturation appears in all three terms, so that:
Ct = α Swn* + β Swp* + γ Swq* (13)
Examples of type-4 models are the Indonesia equation13 and
the method of Raiga-Clemenceau et al.14
Procedures for Determining Saturation Exponent
It is noteworthy that in the particular forms of those
algorithms with more than one exponent of Sw, all exponents
are expressed in terms of n or n*. Therefore the problem of
determining saturation exponent remains one of evaluating n
or n* for equations of all four types. The evaluation procedure
is based on curve-fitting Ir vs. Sw data from a single core plug.
However, the function to be fitted should be governed by the
interpretative model that is to be used subsequently to evaluate
Sw from resistivity logs. This model is described by an
algorithm that will indicate whether the model uses n or n* as
saturation exponent. The adoption of the algorithm leads to a
saturation exponent that is suitable for use with the underlying
interpretative model. Thus, the saturation exponent is
rendered fit for purpose. Values of n* are most reliable if they
are established using control data, in the form of known values
of Cw, F* and x (the last two preferably from multiple-salinity
conductivity measurements). The approach is now described
for each type-model in turn.
Type 1. Here, and in virtually every other non-Archie case, α
= Cw/F*. The non-Archie term can be written in the form γ =
x/F*. For a fully water-saturated reservoir rock, equation (10)
reduces to:
C0 = α + γ (14)
Because Ir = C0/Ct, it follows from equations (10) and (14)
that:
Ir = (α + γ)/(α Swn* + γ) (15)
or:
Ir = (A + 1)/(A Swn* + 1) (16)
where A = Cw/x. Equation (16) therefore allows a
determination of n* as the only unknown parameter by a
regression analysis of Ir vs. Sw data measured on a single core
plug. However, the regression is done in a way that is
consistent with the interpretative model and it is computed in
bilinear as opposed to bilogarithmic space. Type-1 models are
the least flexible of the four types considered.
Type 2. For a fully water-saturated reservoir rock, equation
(11) reduces to equation (14), so that:
4 PAUL F WORTHINGTON SPE 71723

Ir = (α + γ) / (α Swn* + γ Swq*)
or:
Ir = (A + 1) / (A Swn* + Swq*)
(17) Equations (20) and (23) happen to be identical for the
particular models chosen. Once again, equation (23) can be
solved for γ in order to determine a model-compatible value
of the non-Archie coefficient and thence of the coupling
(18) coefficient. Combining equations (22) and (23) we have:

Ir = (α + 2(αγ)0.5 + γ)/(α Swn* + 2(αγ)0.5 Swn* + γ Swn*) (24)

This approach is similar to that described for type-2 models by
Argaud et al., who set q* = n*.15 However, q* = n*-1 in most
other type-2 models.8-10 Again, A = Cw/x and therefore n* can
be evaluated as the only unknown parameter by a regression
analysis of Ir vs. Sw data measured on a single core plug.
Type 3. Here the situation is different. The use of controlling
data from other core measurements is impeded by the coupling
term and the many forms it can take. Therefore a particular
form of type-3 equation has to be assumed in order to
implement the approach. Taking the model of Husten &
Anton,11 we arrive at the following particular form of equation
(12):
Ct = α Swn* + 2 (αγ)0.5 Swn*-1 + γ (19)
or, in the case of fully water-saturated conditions:
C0 = α + 2 (αγ)0.5 + γ (20)
Equation (20) is incompatible with the linear form of C0 vs. Cw
data at higher salinities and therefore those data cannot be
used to control the value of γ during regression. However,
since C0 and α are known, equation (20) can be solved as a
quadratic in γ0.5 in order to determine a model-compatible
value of the non-Archie term and thence of the coupling
coefficient. When this has been done, regression of Ir vs. Sw
data can be effected using the following combination of
equations (19) and (20):
Ir = (α + 2(αγ)0.5 + γ)/(α Swn* + 2(αγ)0.5 Swn*-1 + γ) (21)
where α = Cw/F*. In this case, n* is the only unknown. An
analogous procedure can be followed for the model of
Schwartz and Sen.12
Type 4. Again, a particular model has to be assumed in order
to evaluate the coupling and the non-Archie terms for
controlling the regression analysis. In this case I will begin
with the Indonesia equation,13 which gives rise to the
following particular form of equation (13):
so that:
Ir = 1 / Swn (25)
where n* has now been substituted by n, because the exponent
is obtained by regression as though we were dealing with an
Archie situation. Note, however, that when the Indonesia
equation is applied to field data through equation (22), values
of α and γ are required. An analogous procedure can be
followed for the model of Raiga-Clemenceau et al.,14 which
uses n*.
Synthesis. Table 1 lists the models that are considered here
and indicates the exponents (n or n*) that are required to
characterize their interpretative algorithms. Type-1 models
have not been included, because it became apparent at an early
stage that their mathematical form does not lend sufficient
flexibility to furnish satisfactory matches to Ir vs. Sw data
distributions when used in conjunction with controlling core
measurements. Better fits were obtained when the parameter
A in equation (16) was unconstrained during the regression.
However, the determined values of this parameter then bore
little relation to reality. Further investigation of this problem
has been deferred, because type-1 algorithms are not widely
used.
Application of the Methods
The procedures described above have been applied to
laboratory data in order to investigate the predictive
performance of water-saturation algorithms when used with
fit-for-purpose saturation exponents. Two sample datasets
have been selected. Both of these lie within, and are
representative of, the admissible domain that is identified in
Fig. 1. Sample 1 behaves marginally as a shaly sand in that it
plots just below the line F/F* = 0.9 (Fig. 1). Sample 2 plots
so that F/F* << 1 (Fig. 1). On this basis we can expect fit-for-
purpose saturation exponents to show a greater dispersion for
Sample 2, because differences in non-Archie conductivity
terms between interpretative models are more likely to be
significant where non-Archie effects are greater.

Sample 1. The first dataset to be considered has the

Ct = α Swn* + 2 (αγ)0.5 Swn* + γ Swn*
or, in the case of fully water-saturated conditions:
C0 = α + 2 (αγ)0.5 + γ
(22)
petrophysical characteristics indicated in Table 2, where the
electrical properties are derived from multiple-salinity
conductivity measurements. The distribution of Ir vs. Sw data
for Cw = 9.98 S/m is shown in Fig. 2 with a conventional
(23)
bilogarithmic display. The models of Table 1 have been fitted
to these data using the procedures described above. In
SPE 71723 DETERMINATION OF FIT-FOR-PURPOSE SATURATION EXPONENTS
particular, Fig. 2 also shows a type-3 data fit based on the
model of Husten and Anton.11 Note that although the result of
the data fitting is shown in bilogarithmic space, the regression
itself was effected in bilinear space. The resulting saturation
exponents are listed in Table 3, which also compares the
values of Sw predicted by the various models using the fit-for-
purpose saturation exponents. For the models of Waxman &
Smits,9 Husten & Anton,11 Schwartz & Sen12 and Raiga-
Clemenceau et al.,14 the predicted Sw was calculated by
successive approximation. Although the saturation exponents
do show significant variations, the predicted values of Sw
based on these exponents all lie within a range defined by two
saturation units, except for some departures where Ct = 0.02
S/m, at which point we have moved well beyond the
calibration range for this sample and also have strayed beyond
the lower limit of Sw = 0.15 used by Archie.1 These
predictions, too, draw upon the control provided by core data.
It is unlikely that this precision would be achievable
downhole. However, our purpose here is to contain
uncertainties that can be attributed to the determination of
saturation exponent itself, and this outcome has certainly been
achieved in Table 3.
Sample 2. The second dataset relates to a tighter sample
(Table 2). The distribution of Ir vs. Sw data for Cw = 9.98 S/m
is shown in Fig. 3, once more in bilogarithmic space. Here,
the determined values of saturation exponent show a greater
spread. Yet, the values of Sw predicted by the different models
again lie within a restricted range, with minor departures
where Ct = 0.02 S/m, at which point we have strayed beyond
the calibration range for this sample (Table 4).
Discussion. The procedure that has been exemplified above is
generic. Its application allows model-specific values of
saturation exponent to be evaluated. These values take the
form of n or n* as appropriate.
In the particular case of n*, the method described here
offers a single-salinity desaturation approach to the
determination of a fit-for-purpose saturation exponent, given
that a relationship between n* and p* and/or q* can be
presupposed and that F* is known from supporting
conductivity measurements under conditions of full water
saturation. The method complements the dual-salinity
desaturation approach to the determination of model-specific
saturation exponent,16 which does not presuppose a
relationship between n* and other exponents and does not
require a knowledge of F*.
The examples considered show a linear, or quasi-linear,
data trend of Ir vs. Sw in bilogarithmic space. The approach
would need to be modified in order to accommodate those
non-linear data trends that are encountered in the presence of
microporosity.17 Again, no account has been taken of
formation anisotropy, which is known to increase under
conditions of partial water saturation.18
In a field study, several samples from the same petrofacies
unit would be considered so that a representative saturation
exponent could be identified for application to that unit using
5
the predesignated water-saturation algorithm. In order for the
method to be most effective, core data should also be used to
calibrate the estimates of the non-Archie conductivity term
that are made using well logs, because these estimates can be
tenuous.
Conclusions
By drawing upon the recognized continuum of electrical
properties of fully water-saturated sands, it has been possible
to identify a domain within which the non-Archie formation
factor F* becomes numerically equivalent to the Archie
formation factor F as the electrical manifestation of shaliness
approaches zero. The present study has been located within
that domain.
Building on the type-classification of interpretative
algorithms for the evaluation of water saturation, a structured
approach has been outlined for the determination of a
saturation exponent that is appropriate to an interpretative
algorithm used to evaluate Sw from well logs. In this way, the
derivation of the saturation exponent and the subsequent field
evaluation of Sw draw upon the same model throughout. Of
the models considered, the Indonesia equation13 and the
algorithm of Argaud et al.15 require values of n: all the other
models require values of n*. Yet, notwithstanding these
different requirements, the approach adopted here is generic,
in that it can be applied to water saturation algorithms of all
types. Note, however, that type-1 models have been dropped
from the present approach, where they contain only one term
that can vary and this has been insufficient to fit meaningfully
the observed data when physical controls are applied in the
form of core calibration.
The application of these diverse equations, each used in
conjunction with its model-specific saturation exponent, has
furnished predictions of water saturation that are confined to a
range of about two saturation units. Moreover, the same
outcome has been forthcoming over a range of formation
conductivity Ct. Thus the adoption of a fit-for-purpose
saturation exponent can measurably contain the sensitivity
associated with the choice of an interpretative model for the
determination of water saturation. Thus, provided that n or n*
is determined and applied as described above, the choice of
interpretative algorithm for the evaluation of water saturation
becomes much less critical. The application of the method
therefore contributes to a managed uncertainty in
petrophysical evaluation. In this respect it is beneficial both
technically and commercially.
Note that these comments relate implicitly to the domain
of the continuum chart that we have considered here (Fig. 1).
Further benefits are likely to accrue in very shaly reservoirs
and those containing relatively fresh formation waters. Future
work will include the extension of the method to cases such as
these and an appraisal of how the present methodology is
impacted when type algorithms are used beyond their
conventional range of application.
6 PAUL F WORTHINGTON SPE 71723

References 10. Clavier, C., Coates, G. & Dumanoir, J. : “Theoretical

1. Archie, G.E.: “The Electrical Resistivity Log as an Aid
in Determining Some Reservoir Characteristics,” Trans.
AIME 146 (1942) 54-62.
2. Worthington, P.F.: “The Evolution of Shaly-Sand
Concepts in Reservoir Evaluation,” The Log Analyst
26(1), (1985) 23-40.
3. Fertl, W.H. & Hammack, G.W.: “A Comparative Look
at Water Saturation Computations in Shaly Pay Sands,”
Trans. SPWLA 12th Annual Logging Symposium (1971)
R1-17.
4. Worthington, P.F.: “A Continuum Approach to the
Petrophysical Classification and Evaluation of Reservoir
Rocks,” Petroleum Geoscience 1 (1995) 97-108.
5. Worthington, P.F.: “Recognition and Evaluation of
Low-Resistivity Pay,” Petroleum Geoscience 6 (2000)
77-92.
6. Hossin, A.: “Calcul des Saturations en Eau par la
Méthode du Ciment Argileux (Formule d’Archie
Generalisée),” Bulletin de l’Association Française des
Techniciens du Pétrole 140 (1960).
7. Givens, W.W.: “A Conductive Rock Matrix Model
(CRMM) for the Analysis of Low-Contrast Resistivity
Formations,” The Log Analyst 28(2) (1987) 138-151.
8. Bardon, C. & Pied, B. : “Formation Water Saturation in
Shaly Sands,” Trans. SPWLA 10th Annual Logging
Symposium (1969) Z1-19.
9. Waxman, M.H. & Smits, L.J.M.: “Electrical
Conductivities in Oil-bearing Shaly Sands,” Society of
Petroleum Engineers Journal 8 (1968) 107-122.
and Experimental Bases for the Dual-water Model for
Interpretation of Shaly Sands,” Society of Petroleum
Engineers Journal 24 (1974) 153-167.
11. Husten, P. & Anton, H.: “The Porosity Surface
Concept,” Oil Gas 7(1) (1981) 40-42.
12. Schwartz, L.M. & Sen, P.N.: “Electrolytic Conduction
in Partially Saturated Shaly Formations,” SPE Paper
18131 (1988).
13. Poupon, A. & Leveaux, J.: “Evaluation of Water
Saturation in Shaly Formations,” Trans. SPWLA 12th
Annual Logging Symposium (1971) O1-2. (Full text in:
SPWLA, Houston, Texas, Shaly Sand Reprint Volume
(1982) IV 81-95.)
14. Raiga-Clemenceau, J., Fraisse, C. & Grossjean, Y.: “The
Dual Porosity Model : A Newly Developed
Interpretation Method for Shaly Sands,” Trans. SPWLA
25th Annual Logging Symposium (1976) E1-16.
15. Argaud, M., Giouse, H., Straley, C., Tomanic, J. &
Winkler, K.: “Salinity and Saturation Effects on Shaly
Sandstone Conductivity,” SPE Paper 19577 (1989).
16. Worthington, P.F.: “Characterization of Water
Saturation Algorithms Through Dual-Salinity
Desaturations,” The Log Analyst 37(1) (1996) 31-46.
17. Worthington, P.F. & Pallatt, N.: “Effect of Variable
Saturation Exponent upon the Evaluation of
Hydrocarbon Saturation,” SPE Formation Evaluation 7
(1992) 331-336.
18. Klein, J.D.: “Saturation Effects on Electrical
Anisotropy,” The Log Analyst 37(1) (1996) 47-49.

Table 1 Models used in the analysis Table 2 Petrophysical properties of samples

Model type Name Saturation Sample Porosity Formation Non-Archie

exponent No. Factor Conductivity
(S/m)
2 Waxman & Smits n*
2 Argaud et al. n 1 0.189 27.8 1.34
3 Husten & Anton n* 2 0.107 87.0 2.99
3 Schwartz & Sen n*
4 Indonesia n
4 Raiga-Clemenceau et al. n*

Table 3 Results for Sample 1 Table 4 Results for Sample 2

Model name Fit-For- Predicted Sw for input

Model name Fit-For- Predicted Sw for input
Purpose values of Ct (S/m) =
Purpose values of Ct (S/m) =
Saturation 0.1 0.05 0.02
Saturation 0.1 0.05 0.02
Exponent
Exponent
Waxman & Smits 2.25 0.819 0.574 0.350
Waxman & Smits 1.83 0.430 0.274 0.144
Argaud et al. 1.93 0.813 0.568 0.353
Argaud et al. 1.63 0.422 0.276 0.157
Husten & Anton 2.31 0.822 0.576 0.344
Husten & Anton 1.86 0.425 0.268 0.135
Schwartz & Sen 2.24 0.827 0.583 0.331
Schwartz & Sen 1.77 0.432 0.270 0.111
Indonesia 1.93 0.813 0.567 0.353
Indonesia 1.63 0.418 0.273 0.156
Raiga-Clemenceau 2.08 0.815 0.569 0.352
Raiga-Clemenceau 1.72 0.420 0.273 0.153
et al.
et al.
SPE 71723 DETERMINATION OF FIT-FOR-PURPOSE SATURATION EXPONENTS 7

1
ARCHIE REGION Sample 1
0.9
Sample 2

25.0 S/m
NON - ARCHIE
0.5 REGION
10.0 S/m

3.3 S/m

1.0 S/m

Cw=0.1 S/m

0.1
0.01 0.1 1 10 100
(S/m)

Fig.1 Continuum of shale effects

R esist ivit y R esistivity

In dex In de x

10 10
8 8
6 6

4 4

2 2

1 1
0.1 0.2 0 .4 0.6 0.8 1.0 0 .1 0.2 0.4 0. 6 0.8 1.0
W at er Sa tu ratio n W ater S atur atio n

Fi g.2 R e si sti vi ty ind e x vs . wa ter s a tur ati on Fi g.3 R e s is tiv ity in de x vs . wa ter s a turati on
for S a m ple 1 w ith Ty pe -3 da ta fit for S a m ple 2 with Ty pe -3 da ta fit