Use of Resistivity Logs To Calculate
Water Saturation at Prudhoe Bay
D.D. McCoy, SPE, Exxon Production Research Co., and W.A. Grieves, SPE, Exxon Co. U.S.A.
Summary
At the time of this work, induction- and porosity-log measurements
were available in more than 180,000 ft of hydrocarbon column from
428 wells. Water saturations were determined from this areally and
vertically extensive database after calibration to oil-base-core (OBC)
-derived water saturations. This paper presents the method used to
calibrate the log data to the OBC water saturations and reviews the
methods used to derive resistivity and formation factor values. It also
presents field measurements of as-received core-plug resistivity and
water-saturation values from two oil-base-mud (OBM) -cored wells.
Comparisons are presented with both field water saturations and
“routine” electrical-property measurements on extracted plugs.
Introduction
Calculation of water saturation from logging measurements does
not result in a precise determination of the distribution of the water
saturation in a single well or across a given reservoir. The impreci-
sion in the calculated values of water saturation is a result of both
the application of models, such as the Archie equation, that cannot
account for all factors affecting the relationship between formation
resistivity and water saturation and uncertainty in the parameters
used within the equation.
More than 450 wells at Prudhoe Bay had deep-induction-log data.
The objective of calibrating the log-derived water saturations was
to extend the accuracy of the OBC water saturations to the larger and
more areally extensive database of logged wells. There are two key
requirements for calibration. The first is the need for a representa-
tive sample of the distribution of water saturation in the reservoir.
The other key factor is the accuracy of these water-saturation values.
The OBC water-saturation database was extensive and, in general,
met the criteria for adequately sampling the water-saturation dis-
tribution in the reservoir, despite the variability of those values. In
a given zone within a well, there were, on average, approximately
50 water-saturation measurements used in the calibration. Overall,
there were nearly 7,000 independent measurements and all litholo-
gies were sampled by the OBC data set.
This work will review the derivation of the parameters used to cal-
culate water saturations from the Archie equation. Further, it will
discuss the major factors that led to the need for calibration of the
Archie equation through use of the OBC water-saturation values. Fi-
nally, it will attempt to verify the calibration method by direct com-
parison with a laboratory determination of the calibrating parameter
from an on-site testing program.
Derivation of the Archie Parameters
Model Selection. The Archie equation was selected for the calcula-
tion of water saturation in the Ivishak formation. There are six pa-
rameters in the Archie equation:
S w + ƪ(aR w)ńǒf m R tǓƫǒ1ńnǓ , . . . . . . . . . . . . . . . . . . . . . . . . (1)
where a,m+electrical parameters; n+saturation exponent; f+po-
rosity, fraction; R t +true or formation resistivity, W@m; and
Rw +water resistivity, W@m.
The Archie equation was selected because of the generally low
level of conductive matrix effects within the Ivishak and because it
had been applied and documented in numerous other fields.1
Copyright 1997 Society of Petroleum Engineers
Original SPE manuscript received for review 22 December 1994. Revised manuscript re-
ceived 26 November 1996. Paper peer approved 2 December 1996. Paper (SPE 28578) first
presented at the 1994 SPE Annual Technical Conference and Exhibition, New Orleans,
25–28 September.
SPE Reservoir Engineering, February 1997
Each parameter other than the saturation exponent was estimated
and evaluated from independent sources of information and will be
discussed in this paper. Later in this paper, it is demonstrated that
derivation of saturation exponents from laboratory data yielded un-
reliable parameter values. Because only one parameter in the Archie
equation was unknown, the saturation exponent was chosen as the
calibration parameter.
Resistivity Values. The dual induction log was the standard log for
resistivity logging at Prudhoe Bay. Although resistivities in the up-
structure areas of the field are higher than normally considered ap-
propriate for the induction log, comparisons with laterologs run in
three wells indicate that the induction log is capable of better per-
formance at high resistivities than is normally considered possible.
However, because the induction log measures conductivity, small
errors in the conductivity zero setting can be significant at low con-
ductivities (high resistivities). Consequently, it was necessary to de-
velop a method to detect these calibration errors and to correct for
them in an appropriate manner.
Mapping was chosen as the method to compare conductivity val-
ues on an areal basis (this method works best when there are two
smoothly varying values of key petrophysical properties). The value
chosen for mapping is the log (base 10) of the conductivity corre-
sponding to the 95th percentile of resistivity, after eliminating nonpay
intervals, pyritic intervals, bad data, and water-influx intervals. This
value was intended to represent the maximum resistivity level in a
thick bed and be the value most sensitive to conductivity zero errors.
A map of the conductivity value described previously was devel-
oped. A number of areas with closely spaced and irregular contours
were noted, especially in the east-central and north areas of the field.
The tentative calibration correction was determined as the amount
of conductivity to be added or subtracted to bring the well’s loga-
rithm of conductivity to within one standard deviation of the con-
ductivity of nearby wells as recorded by the deep induction log.
A total of 62 wells was identified as potential candidates for
correction with this procedure. However, the following criteria were
set up to limit corrections to reasonable values.
W
1. No correction less than 0.5 m /m. Correction is too small to
consider.
W
2. No correction greater than 10 m /m. Correction should be
less than anticipated sonde error.
3. No correction if fewer than five wells within 1-mile radius. Too
few wells to apply technique.
4. No correction if 95th percentile resistivity is less than 40 W@m.
Hard to recognize errors in low-resistivity areas.
Wells chosen for calibration corrections are listed in Table 1 along
W
with the corrections. Corrections range from 0.5 to 6.4 m /m. Elev-
W
en negative corrections average *2.4 m /m and 19 positive correc-
W
tions average )1.8 m /m. The average absolute correction is 2.0
W
m /m. The fact that the average correction is close to the published
accuracy of the induction log also lends credence to the corrections
as a whole. The corrected conductivity map is given in Fig. 1.
Derivation of a and m Values. Two sources of data were evaluated
for determination of the relationships between formation factor, F,
(F+a f* m) and porosity in the Ivishak reservoir. Laboratory meas-
urements of porosity and resistivity on representative core samples
were selected for the final calculation of these relationships. Log-
ging measurements of resistivity and porosity were also available in
water-saturated intervals; however, use of this source of data led to
less consistent predictions of the formation factor.
Laboratory data were taken on 975 core samples in an attempt to
represent both the wide areal extent and the vertical zonation of the
45
 TABLE 1—CALIBRATION CORRECTIONS INDUCTION LOG
Well Correction Well Correction Well Correction
W W W
API m /m API m /m API m /m

71 –6.4 213 + 4.6 293 –0.6

73 – 2.6 228 –1.0 294 + 1.0
95 –4.0 229 –1.2 295 –2.0
115 + 0.9 254 + 0.8 338 –1.2
118 + 0.6 255 + 1.7 434 + 1.1
119 + 4.6 257 + 4.3 435 + 1.3
126 –2.8 272 + 0.9 442 + 0.8
136 + 2.3 283 + 1.0 451 + 2.5
144 + 2.5 284 + 0.5 534-2 + 3.8
148 –3.6 291 –0.9 804 + 0.6
Fig. 1—Conductivity map after induction-log calibration.

TABLE 2—ZONAL VALUES OF

ARCHIE PARAMETERS a AND m
Number of
Zone Samples a m
4B 170 0.866 1.971
4A 132 1.320 1.684
3 and 2C 97 0.913 1.933
2B 118 0.828 2.017
2A 81 0.765 2.069
1B and 1A 140 0.976 1.877

field. These data were edited to remove those data that were unrep-
resentative of the reservoir as a result of sampling difficulties, ex-
perimental errors, or difficulties in obtaining measurements repre-
sentative of field conditions. Slightly more than one-half of all edits
were made because the samples contained more than 2% bulk vol-
ume pyrite. (The net result of removing all samples with greater than
2% bulk volume pyrite is that the values of a and m are representa-
tive of only those portions of the field that do not contain any sub- Fig. 2—Formation factor/porosity relationship in Zone 4B based
stantial volume of pyrite.) The electrical measurements were taken on laboratory data.
by five laboratories at varied experimental conditions and converted
to equivalent reservoir conditions.
Owing to sparse sampling, data from certain lithologic zones
were combined and values of a and m were developed. Zones 2C
and 3 had limited sampling, and the data had a limited range of po-
rosity. Zone 1A had only 16 samples, and the data were merged with
data from Zone 1B, which was a similar lithology to Zone 1A. An
example of the formation factor/porosity plot used in the derivation
is shown in Fig. 2. The correlation coefficients ranged from 0.89 to
0.98 and, therefore, the selection of a linear-regression technique
was not critical. The final values of a and m are given in Table 2.

Porosity Values. Porosity values were derived from the compen-

sated sonic log after calibration to an areally extensive core-porosity
database. The sonic log was selected to minimize the influence of
the heavy minerals (pyrite, siderite) present throughout field. Linear
relationships between the core porosity and the sonic transit time
were developed as illustrated in Fig. 3. A y on x (core porosity on
transit time) least-squares fit was used to develop the matrix and
fluid transit-time parameters. Final parameters were selected to ac-
count for areal variations in the transit times; i.e., correlations were
developed for specific well clusters. Fig. 3—Core porosity vs. sonic transit time In Zone 2B for oil-sat-
urated intervals.
Water Resistivity. Ref. 2 discusses values for water resistivity.
Determination of n on a foot-by-foot basis is possible in an oil-
Saturation Exponent. Laboratory saturation-exponent values were base-cored well where water saturations are measured at this fre-
available from measurements on extracted core samples. However, quency. However, it is meaningless to try to relate these depth-spe-
incorporation of these data into the Archie equation yielded signifi- cific values of n to wells that were not cored with an OBM.
cantly and systematically different values of water saturation than Therefore, zone-average values of n were derived in each OBC well.
those measured on OBC’s. This is illustrated in Fig. 4. Structurally Specifically, n was iteratively determined until the porosity-
high wells had an implied water-saturation loss in the cores of greater weighted log-derived water saturation was equal to the porosity-
than 10% PV, while flank wells had an implied water gain in the cores weighted core water saturation. By use of a porosity-weighted
of over 10% PV. This observation led to the need to calibrate the Ar- match of core and log water saturations, this solution calibrates the
chie equation with the OBC water-saturation values.3 logs to the quantity most directly measured on the core plugs, the

46 SPE Reservoir Engineering, February 1997

Fig. 4—Map showing the difference between Sw values derived
from laboratory n values and OBC water saturations, all zones.
water volume. This also ensures that a few feet with unusually low
porosity with higher water saturations would not unduly influence
the zone-average water saturation.
Fig. 5 presents posted values of the saturation exponent for Zone
4B in each oil-base-cored well. Similar maps were developed for the
other zones. Comparisons among the zones demonstrate that the
highest n values are consistently found in Zones 3 and 2C (the con-
glomeratic zones) and that the lowest n values are found in Zones
1A and 1B (the shaliest intervals).
Application of the n values to all logged wells in a given zone was
accomplished by a correlation of the calibrating parameter n with the
zone-average height above the oil/water contact (OWC). This is illus-
trated in Fig. 6. Calibration with height is not surprising because the
effects of microporosity and thin shales are related to the water-satu-
ration level, which in turn is related to the height above the contact.
Reasons for Calibration
The need to calibrate a suite of logs to known core fluid saturations
will vary from field to field because the Archie equation is based on
the assumption that, at a fixed porosity and water salinity, the forma-
tion resistivity is solely a function of water content. Two of the three
primary reasons forcing the need to calibrate the Archie equation in
the Ivishak were violations of this assumption.
The first reason is that the presence of a variety of conductive ma-
trix material altered the relationship between resistivity and fluid
saturation. These conductive matrix effects include the effects of
dispersed pyrite, thin shales, and dispersed clays. Secondly, the
presence of a significant volume of microporous chert (microporos-
ity) caused a significant effect on the Archie equation.4 The final
primary reason forcing calibration of the Archie equation was that,
although care was taken to estimate all the parameters in the equa-
tion, biases remained in the prediction of those parameters. While
these specific reasons for calibration may not be directly applicable
in other fields, variations in the mineralogy and lithology of other
reservoirs will likely cause similar difficulties in log analyses.
Fig. 6—Plot of n vs. height above OWC for Zone 4B.
SPE Reservoir Engineering, February 1997
Fig. 5—Saturation exponents, n, in Zone 4B derived from OBC
water saturations.
Conductive Matrix Effects. The presence of conductive matrix in
the formation will lower the resistivity and, without calibration to an
accurate standard, lead to the prediction of high water-saturation val-
ues. Three types of conductive matrix material were present in the Ivi-
shak: disseminated pyrite (marcasite in lesser amounts), thinly
bedded shales or layers of mudstone ripup clasts, and dispersed clays.
Minor Pyrites. The effect of small (2 to 10 vol%) amounts of dis-
seminated pyrite on formation resistivity has been previously docu-
mented.5 The presence of pyrite was confirmed by both visual de-
scription and thin-section petrography. The resistivity recorded by
the induction log in pyritic regions is depressed, and water satura-
tions calculated from the Archie equation will be overestimated.
Identification of the intervals containing small amounts of pyrite
could not be accomplished from log data owing to the occurrences
of siderite, another heavy but nonconductive iron mineral. In Fig.
7, a map illustrates the existence of pyrite, predominantly in the
northern portion of the field, for the cored wells in Zone 3. Similar
maps were developed for each zone.
Thin Shales. The influence of a thin, conductive layer on the re-
sistivity recorded by the deep induction log can be significant be-
cause of the limited vertical resolution of the deep induction log.
Major shales (those greater than 2 ft in thickness) were identified
and edited from the database, and resistivity data were not used to
calculate water saturation for a distance of 2 to 3 ft from these shales.
This procedure was adequate to account for the influence of thick
shales on the calculated water saturations. Therefore, this discus-
sion will be limited to shales less than 2 ft in thickness.
Even after deconvolution of the deep induction log,6 a 2- to 3-in.-
thick shale can depress the resistivity response over approximately
7 ft. The magnitude of this effect and its influence on calculated wa-
ter saturation is a function of the contrast in resistivity between the
sands and the shales. Hence, the effect is greatest in portions of the
field with low water saturation where sand resistivities exceed 1,000
W@m and shale resistivities are approximately 10 W@m.
To confirm the presence of the influence of thin shales on the in-
duction log, Fig. 8 presents an example from Well API 875. At a
measured depth of 10,026 ft the gamma ray log gives evidence of
Fig. 7—Distribution of minor pyritic intervals in Zone 3.
47

Fig. 8—Example of the effect of thin shales on induction log, Well
API 875; ILD+induction log deep.
a thin shale. This was confirmed by the core description that picked
a 1½-ft-thick shale at this depth. The water saturations in and around
this depth are on the order of 2% PV, and the sand resistivities are
in excess of 2,000 W@m. Opposite this thin shale, the induction log
resistivity is depressed to less than 20 W@m and the influence of the
shale extends approximately 4 ft in either direction. Foot-by-foot
measurements of the water saturation confirm that the water satura-
tion does not vary over this affected interval. The influence of thin
shales is not limited to a few wells, but is important across the field;
in fact, large contiguous areas have similar thin-shale contents.
Dispersed Clays. One of the strongest reasons in support of the
selection of the Archie equation for calculation of water saturation
at Prudhoe Bay is the limited clay content of this reservoir. Kaolinite
is the clay mineral most commonly present and it typically contrib-
utes only 1 to 3% of the bulk volume of the samples. Chlorite and
illite are also found in minor quantities. The effect on the electrical
properties for this level and type of clay is expected to be quite low.
Cation-exchange-capacity measurements were taken on more
than 1,000 samples, with the most common values of cation ex-
change capacity falling between Qv +0.04 to 0.06 eq/PV. No major
trends in cation exchange capacity were apparent across the field;
however, there was a correlation of decreasing Qv with increasing
porosity, as expected. The influence of this level of cation exchange
capacity is small except at the lowest water saturations, which are
present only in the upstructure portion of the field.
Influence of Microporous Chert—Fluid Distribution Effects. As
described by Sneider,7 microporous chert is an important mineral
within the Ivishak reservoir. Microporous chert comprises from 2 to
60% of the total rock volume, with most samples containing 15 to
30%. The porosity of a chert grain is typically 20 to 50%, and the
pores are small, typically having a pore-throat radius on the order
of 0.5 mm. The porosity contained within the microporous chert
grains contributes significantly to the total porosity of the rock, per-
haps accounting for 20 to 40% of the total rock porosity.
The effect of the presence of this large volume of microporosity
on the resistivity/water-saturation relationship is important and not
directly accounted for by the Archie equation. At water saturations
approaching 100%, the water content of both the intergranular and
intragranular pore systems is high and there is little effect of the
chert on the rock resistivity. At intermediate brine saturations (rang-
48
Fig. 9—Effect of microporous chert on Archie relationship.
ing from 40 to 70% PV), the micropores within the chert are highly
water-saturated owing to capillary pressure considerations and the
water saturation of the intergranular pore system is low. Over this
saturation range, the contrast in both brine saturation and resistivity
is greatest between the intragranular and intergranular systems.
Modeling the electrical behavior of this dual pore system predicts
that this nonuniform distribution of brine within the sample will
yield higher-than-expected resistivities for a given total water satu-
ration. Therefore, high values of n must be used in the Archie equa-
tion for this saturation range to account for this effect.
At very low saturations, the contrast in resistivity between the in-
tergranular and intragranular system is reduced. The brine satura-
tion within the chert grains, while still greater than the water satura-
tion of the intergranular pore system, is closer to the saturation of the
intergranular pores. Thus, the value of n needed to predict total wa-
ter saturation correctly is lower than at intermediate saturations. Fig.
9 illustrates this effect. The Archie equation is shown on a log-log
plot of resistivity index vs. water saturation as a straight line. Both
a one-dimensional (1D)—series resistor model—and a three-
dimensional (3D) model—embedded media theory8—were used to
approximate the effect of chert on the Archie equation.
The 1D model assumes that the chert and intergranular porosity are
approximated by a series resistor model. The resistances of the micro-
porous chert and the intergranular pore systems are both calculated
with the Archie relation, assuming similar values of a, m, and n. The
3D model assumes that the chert grains have a resistance and are em-
bedded in a homogeneous media. For the parameters chosen, the
maximum apparent n value was seen at a brine saturation of 30%.
Lower values of n are required at lower Sw. The parameters chosen
include the total porosity, chert fraction, intragranular porosity, and
pore-size distributions of both the intergranular and intragranular
pore systems. The exact values of the parameters are not presented
because the figure is offered only to demonstrate the effect of chert
or a bimodal pore-size distribution on the saturation exponent.
Imperfection in Archie Parameters. Although care was taken to de-
termine each parameter used in the Archie equation accurately, biases
remained in the prediction of those parameters. While existence of
these biases can be demonstrated, quantification or elimination of
these biases could not be done with the limited databases available.
The primary tool used to evaluate the values of a and m derived
from the laboratory data was the backcalculation of water saturation
in the aquifer. With the values of a, m, and the other parameters pres-
ented previously and an assumed value of n+2, a zone-average water
saturation was determined for every well in the aquifer (as shown in
Fig. 10 for Zone 1B). On average, the correct value of water satura-
tion is derived because the mean well-zone water saturation is 102%.
SPE Reservoir Engineering, February 1997

Fig. 10—Water saturations calculated within the aquifer with lab-
oratory a and m values, Zone 1B.
However, the standard deviation of these zone-average water satura-
tions is about 15% PV. Areal effects are even more important because
nearly all wells show an overestimation of the calculated water satu-
ration along the southern flank, while western wells consistently un-
derpredict the water content of the formation. This clearly identifies
the uncertainty in log-derived water saturations to be significant.
Limitations of Calibration. The rationale behind the determina-
tion of OBC n values is that, at wells with known Sw values, zone-av-
erage saturation exponents can be determined accurately. Extrapo-
lation to nearby wells depends on both similar mineralogy and
similar log-response characteristics. If the induction or sonic logs
are incorrectly calibrated at either the well with known Sw or the well
to which the OBC n values are being applied, incorrect porosity and/
or water-saturation values will be calculated. Furthermore, the
zone-average n cannot be applied on a foot-by-foot basis. Prediction
of an accurate n for each foot, even in a well with known values, is
extraordinarily difficult because precise knowledge of core-to-log
depth shifts and log deconvolution procedures must be available.
Zones considered within the Ivishak were typically more than 50 ft
thick. Rigorously, no minimum thickness was estimated for deriva-
tion of reliable n values; however, it was anticipated that at least 20
ft of data would be required to account for core-to-log depth mis-
matches and inaccurate vertical resolution of deconvolved logs.
Validation of OBC n Values
Saturation exponents determined from OBC water saturations, deep
induction log resistivity values, and porosity log data differed signifi-
cantly from routine electrical-property measurements made on ex-
tracted core samples. Therefore, an experimental program was under-
taken to analyze this difference. Additionally, steps were undertaken
to explain the differences in n values determined from routine labora-
tory studies and the OBC calibration. The first program was an on-site
experimental program used to measure the resistivity and water satu-
ration of core plugs and thereby allow calculation of saturation expo-
nents shortly after the core was recovered. This will be referred to as
the on-site or “native-state” saturation-exponent program. The results
of this work confirmed the method of deriving zone-average n values
from OBC water-saturation data. The second program was a review
of the deficiencies in routine laboratory procedures.
Native-State n Values. Native-state saturation exponents were
measured in an attempt to understand the discrepancy in saturation
exponents derived from log data and OBC water-saturation values
and from routine laboratory electrical property measurements on
extracted core plugs. In the upstructure portion of the field, laborato-
ry measurements of saturation exponents on extracted core plugs
had average values over a range from 2.5 to 3.5, while the OBC
method yielded n values in the range of 1.5 to 2.0. Exactly the oppo-
site condition was noticed for the extreme southern and western
flanks of the field, where the laboratory values were lower than
OBC-derived values. Two OBC wells were selected for this pro-
gram, one in each portion of the field. In each well, there was a large
and opposite difference in predicted saturation exponents.
SPE Reservoir Engineering, February 1997
The native-state saturation-exponent program was designed to
obtain resistivity measurements of the core samples at conditions as
near the pressure and wetting characteristics of the reservoir as pos-
sible. The focus of the experimental design was to obtain measure-
ments of the resistivity and water saturation on core samples as
quickly as possible after retrieval of the sample from the core barrel.
This would minimize the effects of redistribution of formation brine
caused by changes in the wettability of the core material that might
have occurred during the coring and core-handling operations. Pre-
vious studies have demonstrated that invasion of the coring fluid
was extremely limited in this core material.3 In this instance, it is be-
lieved that coring with OBM’s did not significantly alter the wet-
tability of the uninvaded portion of the cores. Most importantly,
measurement of the resistivity was taken before extraction of the
core plugs. Previous work that we conducted on Prudhoe Bay core
material had demonstrated the sensitivity of resistivity measure-
ments to the extraction procedure.
On-Site Procedure. On arrival at the surface, the core material was
placed in core boxes lined with diesel-soaked paper. Samples chosen
for the resistivity measurements were plugged with arctic diesel as the
bit lubricant. On 1-in.-diameter plugs, the first resistance measure-
ment was made at atmospheric pressure with a four-electrode system
described by Hoyer and Rumble.9 Resistivity values, Rt , were calcu-
lated knowing the core diameter and electrode spacing. Each core
sample was placed in a pressure cell, and an overburden pressure of
2,800 psi was applied. The measurement of Rt was made at frequen-
cies of 1,000 and 20,000 cycles/sec. The effects of pressure and fre-
quency on the resistivity data were predictable and small, and use of
different measurement systems had a very small effect on the Rt value
recorded. Typically, cores arrived at the trailer within 30 minutes of
arrival at the surface, and atmospheric Rt data were completed within
2 hours. Measurements of Rt at 2,800 psi were completed within 2 to
40 hours. The water content of each plug was determined with Dean-
Stark extraction apparatus with toluene as the solvent. The water con-
tent of the core plugs were determined over 12 to 36 hours, until pro-
duction stopped. The core samples were then dried and wrapped for
shipment to the laboratory in Houston.
Laboratory Procedures. Samples brought to the Houston labora-
tory still contained the salt from the formation brine left by the
Dean-Stark water extraction. This was leached with 100 mL of de-
ionized water in a Soxhlet-type extractor and then with 10 PV of de-
ionized water forced through the sample. The water samples from
these two steps were combined and titrated for chloride ion by
means of the Mohr titration. Porosity, permeability, grain density,
formation factor, and restored-state saturation exponents were then
measured on each plug by routine procedures.
Comparison With Log R t Values. Of greater interest than internal
checks on the R t values are the comparisons between temperature-
corrected core-measured R t values and the resistivity of the forma-
tion as determined by the deconvolved deep induction log. These
comparisons are shown for Well API 855 in Fig. 11 and for Well API
875 in Fig. 12. Comparison between plug resistivity data and induc-
tion log data is difficult because of the difference in vertical resolu-
tion. Even after deconvolution, the vertical resolution of the induc-
tion log is at least 3 to 5 ft.
Considering the mismatch in vertical resolution, the comparison
for Well API 855 is very good. In the thicker beds (e.g., 9,200 ft
where the induction log is less affected by varying rock properties),
the agreement between core and log R t values is very good. Further,
the plug data scatter reasonably on both sides of the induction log.
(On a conductivity basis, the scatter is well matched. The visual im-
pression of the resistivity scale is misleading, and it is shown only
as the more usual method of presenting log data.) Finally, the appar-
ent disagreements between log and plug data at 9,170 and 9,226 to
9,236 ft are most likely a result of thin-bed or bed-boundary effects.
In Well API 875, comparison between core and log data in the up-
per four zones (Zones 4B, 4A, 3, and 2C) is extremely difficult be-
cause of the 2000-W@m limit placed on the induction log. The plug
data show resistivities higher than 2000 W@m, which is likely con-
sistent with the log data. In the lower zones (Zones 2B, 2A, and 1B),
the deconvolved log shows good agreement with the plug-measured
49
Fig. 11—Comparison of native-state resistivity values on plugs
and induction-log readings, Well API 855; ILd+induction log deep.

resistivity values. Again, thin beds account for some of the apparent
discrepancies (i.e., at 10,057 and 10,116 ft). The overall agreement
between plug and log resistivity values is good, thus supporting the
conclusion that the plug samples are representative of the resistivity
of the formation.

Calculation of Native-State Saturation Exponents. The equation

used to calculate the on-site or native-state saturation exponent is
n + ǒlog R ońR tǓ on-siteń log S w , . . . . . . . . . . . . . . . . . . . . . . (2)
where the resistivities are corrected to a constant temperature of
73.4°F and the water saturations are corrected for pressure effects.
The resulting values of n are summarized in Table 3.
In Well API 855, with an average core Sw of 35%, the native-state
saturation exponent averages 3.18 at the experimental conditions
most representative of the reservoir (i.e., 2,800 psi and frequency of
20,000 cycles/sec). Despite the wide variety of experimental condi-
tions, the average saturation exponent in this well varies between
2.90 and 3.41. This is in sharp contrast to the results in Well API 875,
where the average n value lies between 2.03 and 2.27 for all exper-
imental conditions studied.
In summary, the average values of the native-state saturation ex-
ponents within each well are nearly constant despite the variety of
experimental conditions, including confining pressure, electrical
frequency, equipment, and personnel. These average values can be
compared with other methods of determining saturation exponents
in these two wells.
Fig. 12—Comparison of native-state resistivity values on plugs
and induction-log readings, Well API 875; ILd+induction log deep.
Comparison of Native-State n With OBC n Values. On-site na-
tive-state saturation exponents are compared with the saturation ex-
ponents derived from OBC water saturations in these two wells in
Table 3. The comparison is done for each zone. No samples were
available from Zone 3, the conglomeratic zone, because of the in-
ability to cut plugs. With the exception of Zone 2B, the agreement
for each zone in each well is within about 0.2. In Zone 2B the OBC-
derived n value is 1.52. As previously shown in Fig. 8, this value is
depressed as a result of thin shales and inadequate vertical resolu-
tion of the deep induction log. In this zone, four shales, 2 to 3 ft in
thickness, depress the deconvolved induction log. The agreement of
the n values determined from the on-site experimental program with
n values derived from OBC data is outstanding.
Routine LaboratoryĆMeasured n Values
Comparison With Native-State and OBC n Values. Measurements
of saturation exponents by more routine laboratory procedures (i.e.,
brine desaturation of previously extracted water-filled cores) yielded

TABLE 3—COMPARISON OF RESTORED-STATE, NATIVE-STATE,

AND OBC SATURATION EXPONENTS
Well On-Site Number of
API Zone OBC n Native-State n Samples Extracted Core
855 4A 3.14 3.21 13 2.44
High Sw n Low Sw n
875 4B 1.99 1.89 3 3.01 2.06
4A 2.05 2.08 5 2.99 2.07
3 2.25 — — 3.28 2.63
2C 2.28 2.36 4 3.96 2.78
2B 1.52 2.26 6 2.82 2.15
2A 1.94 2.17 7 2.79 2.17
1B 2.06 2.07 4 2.76 2.11
875 Average 1.93* 2.15 2.91 2.16
*Weighted average of zones listed.

50 SPE Reservoir Engineering, February 1997

surprisingly different results from either native-state n values or n val-
ues derived from OBC data. Only in Wells API 855 and API 875 can
a comparison be made among all three means of estimating n values.
The results of this comparison are shown in Table 3.
In Well API 855, the extracted cores had an average n approxi-
mately 0.7 lower than either OBC or native-state n values. These
measurements were all made at essentially equivalent water satura-
tions. Because the extracted cores had lower n values, they predicted
an in-situ water saturation lower than was measured on the samples
from this OBM-cored well. This alone is a strong indication that the
routine electrical-property data on extracted cores do not yield cor-
rect values of the saturation exponent.
In Well API 875, the saturation exponent was measured on ex-
tracted cores over a range of water saturations. At water saturations
higher than approximately 25%, the n value was nearly constant and
averaged 2.91 for these 27 samples. As Sw decreased, the average
n value decreased until, at an average Sw of 6%, the n was reduced
to an average of 2.16. Unfortunately, the water saturation could not
be lowered further in the laboratory accurately without significant
changes in salinity and over a sufficiently short period of time to be
practical. Because both the reservoir saturation as reflected by the
OBC fluid saturations and the native-state n values were determined
at an average Sw of 2 to 3%, it is believed that, if the measurements
on extracted core samples had been made at reservoir saturations,
they would have been lower than the native-state n of 2.15 or even
the OBC Sw derived n of 1.93. In fact, extrapolation suggests that,
at 2% Sw, the extracted n values would average 1.8. From the data
in this well, two observations were made. First, at equivalent water
saturations, the n values recorded on extracted cores appear lower
than those on native-state samples by 0.2 to 0.3, and, second, the val-
ues of n measured on extracted cores are a function of water satura-
tion for this formation (microporosity).
Conclusions
Calculation of water saturation in each well in the Ivishak formation
was best accomplished by calibrating log responses to the measured
water saturations in oil-base-cored wells. The calibration was done
by determining n values from the OBC Sw data in combination with
the sonic- and induction-log responses.
Native-state core resistivities are in reasonable agreement with
those from the deconvolved induction log. This agreement confirms
the fact that the native-state saturation exponents are representative
of the in-situ electrical properties of the Ivishak formation.
The native-state n measurements are in excellent agreement with
the saturation exponents from OBC Sw data, thereby confirming the
calibration technique used in this formation.
Measurements of saturation exponent in extracted cores do not
match either native-state or OBC n values. This in part results from
recording values at water saturations very different from those in the
formation but, more fundamentally, is because the n values recorded
on extracted cores are sensitive to the extraction procedure and typi-
cally decrease with more thorough removal of hydrocarbon.
Acknowledgments
The data-gathering methods and analytical techniques are the result
of a number of dedicated people. We especially thank Mike Gay for
his attention to detail in gathering the data in the field and Marita Tee
for her precision in the Houston laboratory. We also wish to ac-
knowledge the support of Exxon Co. U.S.A. and Exxon Production
Research Co. Most importantly, we thank J.G. Richardson and R.M.
Sneider for their guidance. The opinions stated here are not neces-
sarily shared by the Prudhoe Bay Unit working-interest owners.
Nomenclature
a,m+ parameters in the Archie equation
SPE Reservoir Engineering, February 1997
F+ formation factor, Ro /Rw
n+ saturation exponent as defined by the Archie
equation
Qv + cation exchange capacity, eq/unit PV
Ro + resistivity of a rock fully saturated with brine, W@m
Rt + resistivity of a partially brine-saturated sample, W@m
Rw + resistivity of brine, W@m
Sw + brine saturation, fraction
f+ porosity, fraction
References
1. Archie, G.E.: “The Electrical Resistivity Log as an Aid in Determining
Some Reservoir Characteristics,” Trans., AIME (1942) 146, 54.
2. McCoy, D.D., Warner, H.R. Jr., and Fisher, T.E.: “Water Salinity Varia-
tions in the Sag River and Ivishak Reservoirs,” SPERE (February 1997).
3. Richardson, J.G. et al.: “Validation of As-Received Oil-Base-Core Water
Saturations From Prudhoe Bay Field,” SPERE (February 1997).
4. Swanson, B.F.: “Microporosity in Reservoir Rocks—Its Measurement
and Influence on Electrical Resistivity,” paper F presented at the 1995
SPWLA Annual Logging Symposium, 17–20 June.
5. Clavier, C., Heim, A., and Scala, S.C.: “Effect of Pyrite on Resistivity and
Other Logging Measurements,” paper presented at the 1976 SPWLA
Annual Logging Symposium, 9–12 June.
6. Woodhouse, R., Greet, D.N., and Mohundro, C.R.: “Induction Log Verti-
cal Resolution Improvement in Vertical and Deviated Wells Using a Prac-
tical Deconvolution Filter,” JPT (June 1984) 993.
7. Sneider, R.M. and Erickson J.W.: “Rock Types, Depositional History, and
Diagenetic Effects, Ivishak Reservoir, Prudhoe Bay Field,” SPERE (Feb-
ruary 1997).
8. Landauer, R.: “Electrical Conductivity in Inhomogeneous Media,” Elec-
trical Transport and Optical Properties in Inhomogeneous Media, J.C.
Garland and D.B. Tanner (eds.), American Inst. of Physics, New York City
(1978) Chap. 2.
9. Hoyer, W.A. and Rumble, R.C.: “Dielectric Constant of Rocks as a Petro-
physical Parameter,” paper O presented at the 1976 SPWLA Annual Log-
ging Symposium, 9–12 June.
SI Metric Conversion Factors
cycles/sec 1.0* E)00 +Hz
ft 3.048* E*01 +m
°F (°F*32)/1.8 +°C
in. 2.54* E)00 +cm
mile 1.609 344* E)00 +km
psi 6.894 757 E)00 +kPa
*Conversion factor is exact. SPERE
David D. McCoy is currently the supervisor of the Reservoir StudĆ
ies Section at Exxon Production Research Co. in Houston. He has
been with Exxon since 1978 and has held positions in reservoir
engineering and formation evaluation. McCoy holds a BS deĆ
gree from Cornell U. and MS and PhD degrees from the U. of IlliĆ
nois, all in chemical engineering. W.A. (Bill) Grieves retired from
Exxon Co. U.S.A. in 1986. He spent approximately 20 years with
a predecessor of Exxon Production Research Co., then joined
Exxon Co. U.S.A. where he specialized in well evaluation for the
Rocky Mountain states, California, and Alaska. Grieves holds a
BS degree in chemical engineering from the U. of Oklahoma.
McCoy Grieves
51