147 Journal of Petroleum Geology, Vol.

30(2), April 2007, pp 147-158

2007 The Authors. Journal compilation 2007 Scientific Press Ltd
PERMEABILITY ANISOTROPY DISTRIBUTIONS
IN AN UPPER JURASSIC CARBONATE RESERVOIR,
EASTERN SAUDI ARABIA
A. Sahin
1, 2
, A. Z. Ali
1
, S. Saner
1
and H. Menouar
1
Most classical reservoir engineering concepts are based on homogeneous reservoirs despite
the fact that homogeneous reservoirs are the exception rather than the rule. This is especially true
of carbonate reservoirs in the Middle East which are known to be highly heterogeneous. The
realistic petrophysical characterization of these kinds of reservoirs is not an easy task and must
include the study of directional variations of permeability. Such variation can be incorporated into
engineering calculations as the square root of the ratio of horizontal to vertical permeability, a
parameter known as the anisotropy ratio.
This paper addresses the distribution of anisotropy ratio values in an Upper Jurassic carbonate
reservoir in the Eastern Province of Saudi Arabia. Based on whole core data from a number of
vertical wells, statistical distributions of horizontal and vertical permeability measurements as well
as anisotropy ratios were determined. The distributions of both permeability measurements and
anisotropy ratios have similar patterns characterized by considerable positive skewness. The
coefficients of variation for these distributions are relatively high, indicating their very heterogeneous
nature.
Comparison of plots of anisotropy ratios against depth for the wells and the corresponding
core permeability values indicate that reservoir intervals with lower vertical permeability yield
consistently higher ratios with considerable fluctuations. These intervals are represented by lower
porosity mud-rich and/or mud-rich/granular facies. Granular facies, on the other hand, yielded
considerably lower ratios without significant fluctuations.
1
King Fahd University of Petroleum and Minerals,
Dhahran, Saudi Arabia.
2
KFUPM Box 661, Dhahran 31261, Saudi Arabia.
email: [email protected]
Key words: permeability, permeability anisotropy,
anisotropy ratio, carbonate reservoir, Arab-D, Saudi
Arabia.
INTRODUCTION
Permeability values in most reservoir rocks are
dependent on the direction in which the measurements
are made. Very often, measurements made in the
horizontal plane are not significantly different from
one another, but considerable differences may exist
between measurements made in horizontal and vertical
directions. These differences in permeability
distributions may significantly affect the reservoir
during depletion, and should be taken into account in
reservoir engineering calculations, including well
productivity, formation of water and gas coning,
secondary recovery methods and well test analyses.
Therefore, it is essential to determine permeability
variations in different directions, particularly in
vertical and horizontal directions within the reservoir.
Such variation can be incorporated into engineering
applications as the square root of the ratio of the
horizontal to vertical permeability (Muskat, 1937;
Henley et al., 1961; Forrest, 1971; Wilhite, 1986;
Farouq et al., 1988; Joshi and Ding, 1996; Menouar
and Hakim, 1995). This ratio is commonly referred
to as the anisotropy ratio.
Due to zonation and layering in a reservoir, the
anisotropy ratio may vary from one zone to another
and even from one layer to another. Significant
variations in this ratio within a particular zone or layer
can also be observed. The pattern of variation of this
ratio provides valuable information about flow
148 Permeability anisotropy distributions in an Upper Jurassic carbonate reservoir, E. Saudi Arabia
behavior within the reservoir, and should be
determined prior to any engineering calculations. If
significant variations do not exist within a particular
zone, it is logical to use a single anisotropy ratio to
represent this zone.
Because of the dependence of permeability
anisotropy on lithology, it is also important to
investigate variations in anisotropy ratio within each
lithofacies in the reservoir. This will help to identify
any facies which is prone to major variations in this
ratio. General information related to these topics may
already be available in geological reports in a
descriptive format. However, results based on
numerical data will greatly enhance our understanding
and, at the same time, facilitate the transfer of
information to reservoir engineers.
This study outlines patterns of distribution of the
permeability anisotropy within the Upper Jurassic
Arab-D carbonate reservoir in an active oilfield
located in the Eastern Province of Saudi Arabia. Brief
reviews of reservoir geology and of data sets available
for the study are given in the following two sections.
Then we present and discuss the distributions of
horizontal and vertical permeability as well as
permeability anisotropy. The final part of the paper is
devoted to an analysis of anisotropy ratio - porosity
relationships, and to specific conclusions which can
be drawn from the study.
RESERVOIR GEOLOGY
The stratigraphic section through the Upper Jurassic-
Lower Cretaceous succession in Eastern Saudi Arabia
is illustrated in Fig. 1. The Upper Jurassic in the area
consists mainly of shallow-marine carbonates and
intervening evaporite units. Four carbonate cycles
each having an evaporite seal have been distinguished
in the final stages of the Upper Jurassic. The Arab-D
reservoir, which is the subject of this study, represents
the oldest of these cycles. It comprises an
approximately 300 ft (91.5m) thick carbonate
succession exhibiting an overall decrease in porosity
and an increase in dolomite content with depth.
Grainstones dominate the upper intervals, whereas
wackestones and fine-crystalline dolomitic rocks
dominate the lower units. Lithological and
palaeontological evidence suggests that the reservoir
rocks represents a shallowing-upward depositional
sequence (Hughes, 1996).
Based on porosity-log characteristics, it is possible
to distinguish six zones in the reservoir, referred to as
Zones 1 to 6 (from top to bottom) in our study. Zones
2 and 3 are the most prolific hydrocarbon producing
zones. The data used in this study originated from
Zones 2 to 4, which are briefly described in the
following paragraphs.
Zone 2 is a 54 ft (16.4 m) thick interval consisting
mainly of grainstones which overlies Zone 3 with a
sharp contact. Grains are mainly bioclasts, peloids and
ooids. The most common bioclasts are miliolid and
textulariid foraminifera and echinoid fragments. The
average grain size is 0.5 mm, but some grains reach
up to 1 cm in diameter. The average core-derived
porosity for this zone is approximately 26%.
Zone 3 is an 82 ft (25 m) thick heterogeneous
carbonate unit consisting of alternating bioclastic-
intraclastic grainstones, packstones, wackestones,
dolomite and dolomitic lime-mudstones. This is a
transitional zone between the mud-rich Zone 4 below
and the almost mud-free Zone 2 above. Therefore,
mud-bearing rock types such as packstones,
wackestones and mudstones are common in Zone 3.
Grains are moderately to poorly sorted and range in
diameter from 0.15 to 1 mm. Stromatoporoids, 1 to 7
cm across, were observed in the lower portion of the
zone. The most common bioclasts are miliolid and
textulariid foraminifera, and echinoids. Zone 3 has
an average core-derived porosity of about 21%.
Zone 4 is a 53 ft (16.1 m) thick interval consisting
mainly of fine-crystalline dolomite and dolomitic
mudstones with thin bioclastic packstone/grainstone
Fig. 1. Stratigraphic column for the Upper Jurassic-
Lower Cretaceous succession in Eastern Saudi
Arabia.
149
A. Sahin et al.
)
%
(

Y
T
I
S
O
R
O
P


L
A
T
O
T
CORE PERMEABILITY
1
.
01
0
1
0
0
1
0
0
0
1
0
0
0
0
1
)
'
0
2

=

"
1
(


H
T
P
E
D
S
E
N
O
Z
35 0
ANISOTROPY RATIO
(mD)
0 40 80 120 160
GAMMA RAY
0 GAPI 40
RHOB
2 G/C3 3
CALIPER
5.5 IN 6.5
NPHI
45 V/V -15
kv
kh1
kh2
R1
R2
R3
WELL: A5
920
940
960
980
1000
1020
1040
3


e
n
o
Z
4



e
n
o
Z
Fig. 2. Plot of open-hole log quantities, core permeability measurements, and anisotropy ratios for Well A5.
Horizontal and vertical permeability values almost overlap in higher-porosity intervals, and deviate
significantly from eachother elsewhere.
150 Permeability anisotropy distributions in an Upper Jurassic carbonate reservoir, E. Saudi Arabia
laminations. A 6ft (1.8 m) grainstone horizon marks
the basal contact with Zone 5. The allochems in the
granular horizons are mainly peloids, intraclasts and
bioclasts. Bioclasts include brachiopods, echinoderms,
benthonic foraminifera, stromatoporoids and sponges.
Dolomitization is widespread in fine-grained rocks, and
dolomites appear to have originated from the diagenesis
(Powers, 1962; Cantrell et al., 2001). This zone has
the lowest porosity with an average value of 8.7%.
DATA SETS
The data used for this study include whole-core and
open-hole log data from six vertical wells intersecting
the reservoir. Code letters have been used to represent
the wells, and the actual depth values are not reported
in this paper. However, the relative positions of data
points have been preserved by subtracting a constant
from each depth value. Considering the number of
measurements, results from only two wells (Well A5
and Well A3) are presented to show the general
approach adopted in our study. The data for Well A5
are based on cores from reservoir zones 3 and 4,
whereas the data for Well A3 is from Zone 3 only. Both
of these wells were represented by a significant number
of whole core measurements: a total of 108 core
measurements were available from Well A5, and 81
measurements from Well A3.
The whole-core permeability measurements
provided an ideal data set for this study because they
included one vertical and two horizontal
measurements at each sampling point. The horizontal
measurements were recorded in two perpendicular
directions. Using these permeability measurements,
it was possible to determine three anisotropy ratios
as outlined later.
Complete open-hole log data for each well were
also available. These data consisted of a number of
well-log derived variables, including, gamma-ray,
neutron porosity, total porosity, density, resistivity,
water saturation and caliper measurements. Most of
these variables and the whole-core permeability
measurements were plotted together with calculated
anisotropy ratio values to aid interpretation (Fig. 2).
PERMEABILITY DISTRIBUTIONS
To determine the pattern of permeability distribution,
statistical analyses of two horizontal permeability
measurements (k
h1
and k
h2
), the average horizontal
permeability (k
h
), and the vertical permeability (k
v
)
were conducted. Analyses included the construction
Fig. 3. (a) Histogram of average horizontal permeability measurements (n=108) from Well A5. The highly
skewed nature of the histogram is apparent.
(b) Histogram of log-transformed average horizontal permeability measurements from the same well,
showing a more symmetrical pattern.
(c) Probability plot showing the observed and expected normal values for the histogram in (b).
The symmetrical nature of the distribution is indicated by the approximately straight-line fit.
151
A. Sahin et al.
of histograms, determination of statistical parameters,
and the study of correlations between permeability
measurements.
Histograms of the average horizontal permeability
and the vertical permeability for Well A5 are illustrated
in Figs. 3 and 4, respectively. Both of these histograms
display positively skewed patterns with only a few
outlier values. Most permeability values are below
500 mD. The histograms of the log-transformed
permeability values and the corresponding probability
plots showing the observed values versus expected
normal values are also included in Figs. 3 and 4. These
diagrams indicate the lognormal nature of the
permeability distributions.
Statistical parameters representing each set of
measurements are listed in Table 1, and include values
for the arithmetic, geometric and harmonic mean, the
standard deviation, the coefficient of variation together
with skewness, and kurtosis. The results indicate that
the means of vertical permeability are much smaller
than the corresponding means of horizontal
permeability measurements. Similar observations have
been reported from the Arab-D reservoir in the Abqaiq
oilfield (Sahin and Saner, 2001).
The coefficient of variation is defined as the ratio
of the standard deviation to the mean. This parameter
has commonly been used to define various classes of
heterogeneity (Corbett and Jensen, 1992; Jensen et
Fig. 4. (a) Histogram of vertical permeability measurements (n=108) from Well A5. The pattern is similar to
the corresponding histogram of horizontal permeability in Fig. 3a.
(b) Histogram of log-transformed vertical permeability measurements from the same well, showing
a bimodal distribution.
(c) Probability plot showing the observed and expected normal values for the histogram in (b). The
symmetrical nature of the distribution is indicated by the approximately straight-line fit.
Parameters k
v
k
h1
k
h2
k
h
Samples 108 108 108 108
Arithmetic mean 152.57 405.38 318.95 362.17
Geometric mean 2.69 207.22 150.42 182.41
Harmonic mean 0.07 111.83 76.96 96.95
Standard deviation 586.88 569.98 467.75 499.72
Coefficient of variation 3.85 1.41 1.47 1.38
Skewness 7.36 2.58 2.5 2.3
Kurtosis 61.71 6.96 6.1 4.81
Table 1. Statistical
parameters of the
whole-core permeability
measurements (Well
A5).
152 Permeability anisotropy distributions in an Upper Jurassic carbonate reservoir, E. Saudi Arabia
al., 1997). If the coefficient of variation is greater than
1.0, permeability distributions are considered to be
very heterogeneous and this is the case for our
distributions. The coefficient of variation also provides
a basis for comparisons of the variability of the
different distributions. As shown in Table 1, this
coefficient for vertical permeability is much greater
than corresponding values for horizontal
permeabilities, reflecting greater heterogeneity in the
vertical permeability distribution. It should also be
noted that the variability of two horizontal permeability
measurements are very similar with values of
coefficient of variation ranging between 1.41 and 1.47
(see Table 1).
Skewness is a measure of the deviation of a
distribution from symmetry. The skewness of a normal
distribution is zero, so that a positive value for this
parameter indicates a distribution with a positive skew,
and vice versa. As shown in Table 1, skewness values
for our case are greater than zero, indicating the
positively skewed nature of the distributions. Kurtosis
is used to measure the flatness or peakedness of a
distribution, relative to a normal distribution. A
common measure of kurtosis is defined as the fourth
moment about the mean expressed in dimensionless
form (Spiegel, 1972). Although this measure is equal
to three for a normal distribution, it is commonly taken
as zero, so that a distribution more peaked than the
normal results in a positive value (leptokurtic), and
more flat than the normal results in a negative one
(platykurtic). Most of our permeability distributions
were characterized by positive kurtosis and hence were
leptokurtic. This indicates that there are more than
expected permeability values around the mean
permeability value.
PERMEABILITY CORRELATIONS
To determine the pattern of correlation between various
permeability measurements, Pearsons linear
correlation coefficients between pairs of
measurements were determined. For Well A5, the
correlation diagram for the pair k
h1
-k
h2
is illustrated
in Fig. 5, and the correlation diagram for the pair k
v
-
k
h
in Fig. 6. Similar correlation patterns were recorded
for Well A3. The overall distribution of points and a
very high positive correlation between the two
horizontal measurements in Fig. 5 indicate the close
similarities of corresponding horizontal permeability
values, and hence the isotropic distribution of
permeability in the horizontal plane. The correlation
coefficient between the two horizontal permeability
measurements is equal to 0.84. On the other hand,
the correlation between the vertical and horizontal
measurements is generally poor, as illustrated in Fig.
6. Scrutiny of points in Fig. 6 indicates that there are
two clusters of points. One cluster represents points
having very small k
v
values with a full range of k
h
values. This cluster represents the muddy facies. The
second cluster is characterized by equally variable k
v
and k
h
values and represents granular facies. These
two facies are also distinguished from the anisotropy
ratio-porosity relationship presented later.
The poor correlation between the k
v
-k
h
pair is
possibly a reflection of the fact that geological factors
controlling the horizontal and vertical permeability
values are generally independent in nature. It is well-
known that bedding planes and laminations are
dominant controls enhancing horizontal permeability,
but that these same features act as barriers to flow in
a vertical direction (Lake, 1988). However, this
general pattern may be disturbed by the presence of
vertical and/or inclined fractures or bioturbation,
which may enhance flow in the vertical direction.
PERMEABILITY ANISOTROPY
Permeability anisotropy can be incorporated into
engineering calculations as the square root of the ratio
of the horizontal to the vertical permeability. This
Fig. 5. Correlation plot for
horizontal permeability (k
h1
and
k
h2
) measurements. The
relatively high positive
correlation coefficient (0.84)
indicates the isotropic
distribution of permeability in
the horizontal plane.
153
A. Sahin et al.
parameter is referred to as the anisotropy ratio. Based
on the whole-core permeability measurements described
earlier, it was possible to determine the following
anisotropy ratios:
............................................ (1)
............................................ (2) and
where k
h
= (k
h1
+k
h2
)/2 .... (3)
Because of the similarity of the horizontal
permeability values at each measurement point as
reflected by the correlation patterns, plots against depth
of the above-mentioned three ratios (R
1
, R
2
, and R
3
)
showed almost identical patterns as illustrated in Fig. 2.
These plots are characterized by three thin (each
approximately 20 ft thick) intervals with relatively high
anisotropy ratios between which are quite uniform
distributions of very low values. High ratio intervals are
located at approximately 920 ft, 990 ft and 1,040 ft.
Overall fluctuations in ratios are considerable with a
maximum value reaching approximately 160 and
minimum at 0.
A comparison of the corresponding open-hole log
plots with the anisotropy ratios in Fig. 2 indicates that
high anisotropy ratios are generally correlated with
lithological changes. At 920 ft, the high ratios are
probably due to the presence of thin beds of dolomite
and limestone as identified on the neutron and density
logs. At 990 ft, they are related to alternating low and
high porosity stringers. Finally, at 1,040 ft, a very low
porosity interval with oblique hairline fractures and/or
a very thinly bedded interval with a variable lithology
may have given rise to the fluctuations. The latter
generally provides parallel neutron and density logs
due to the averaging effect, yet the vertical and
horizontal permeability measurements differ
considerably resulting in higher anisotropy ratios.
From the anisotropy ratio equations, it is obvious
that higher ratio values could either be due to very
high horizontal permeability relative to vertical or
to very low vertical permeability relative to
horizontal. A comparison of individual higher
anisotropy ratio values with corresponding
permeability measurements showed that the
majority of these ratios were due to unusually low
vertical permeability values.
The histogram for the anisotropy ratio (R
3
),
illustrated in Fig. 7, shows a strong positive skew
with highly fluctuating behaviour. A significant
proportion of the data are clustered at the lower end
of the distributions with values below 10, and a
much smaller proportion of data with frequent gaps
are displayed in the tails of the distribution. The
log-normal nature of this distribution is clear from
both the probability plot and the histogram of the
log-transformed values displayed in the same figure.
Statistical parameters for the three anisotropy
ratios are listed in Table 2. As in the case of the
permeability distributions, these parameters include
three types of mean, standard deviation, coefficient
of variation, skewness, and kurtosis. Skewness and
kurtosis values are positive in all cases, indicating
that the distributions are positively skewed and
leptokurtic. Due to close similarities in the values
of three ratios, the corresponding statistical
parameters are very similar. A comparison of the
values of three types of means for each ratio
indicates that the arithmetic mean is the largest and
the harmonic mean the smallest, consistent with the
theoretical expectations (Spiegel, 1972).
Fig. 6. Correlation plot for
the average horizontal
permeability (k
h
) and the
vertical permeability (k
v
)
measurements. The
correlation in this case is
poor with the correlation
coefficient being only 0.48.
R k k
h v
1
1
= /
R k k
h v 2 2
= /
R k k
h v 3
= /
154 Permeability anisotropy distributions in an Upper Jurassic carbonate reservoir, E. Saudi Arabia
ANISOTROPY RATIO - POROSITY
RELATIONSHIP
The general pattern of permeability anisotropy in the
reservoir indicates that there is a possible relationship
between the anisotropy ratios and lithology, with
muddy facies generally possessing higher anisotropy
ratios. But it was not possible to quantify such a
relationship because of the lack of complete
lithological information. However, we had sufficient
porosity data to look at the relationship from another
perspective. The statistical parameters of the porosity
data for Well A5 are listed in Table 3, and the same
data are summarized in the form of a histogram in
Fig. 8. This histogram illustrates the bimodal nature
of the porosity distribution and indicates the co-
existence of two populations, one representing muddy
facies and the other granular facies. The cut-off
porosity value separating these two populations
appears to be located at approximately 11%.
A plot of the anisotropy ratio (R
3
) against porosity
for Well A5 is illustrated in Fig. 9. This plot displays
an obvious negative correlation between the
anisotropy ratio and porosity, with a correlation
coefficient of approximately 0.65. However, this
relationship is not a simple one and the scatter diagram
reveals two main clusters of points. One is
characterized by higher porosities and considerably
Fig. 7. (a) Frequency histogram of the anisotropy ratio (R
3
) for Well A5. The pattern of the distribution is
similar to those observed for permeability measurements (see Figs 3 and 4).
(b) Frequency histogram of log-transformed anisotropy ratios from the same well, showing the bi-modal
character of the data.
(c) Probability plot showing the observed and expected normal values for the histogram illustrated in (b).
The bi-modal character of the distribution is reflected by the alignment of points along lines with different
slopes.
Parameters R
1
R
2
R
3
Samples 108 108 108
Arithmetic mean 25.37 20.87 23.38
Geometric mean 8.77 7.47 8.23
Harmonic mean 3.54 3.03 3.35
Standard deviation 37.26 30.46 33.92
Coefficient of variation 1.47 1.46 1.45
Skewness 1.9 2 1.89
Kurtosis 2.81 3.59 2.9
Table 2. Statistical parameters of
anisotropy ratios (Well A5).
155
A. Sahin et al.
lower ratios without any significant fluctuations. The
second cluster of points is characterized by lower
porosities and higher anisotropy ratios with
considerable variation. This relationship is also
obvious from the open-hole log plots illustrated in Fig.
2. A dominant pattern observed in these plots is that
the horizontal and vertical permeability values are
closer to each other in higher-porosity intervals, and
deviate significantly in lower-porosity intervals.
Therefore, higher porosity areas are expected to yield
lower anisotropy ratios, and lower porosity areas
higher ones, as shown in Fig. 9. It should be noted
that the porosity cut-off value separating the two
clusters in Fig. 9 is about 11%. This value matches
closely with the porosity cut-off separating the two
populations in Fig. 8.
To assess whether the anisotropy ratio versus
porosity relationship observed in Well A5 is laterally
persistent, we constructed the same plot for Well A3.
The plot for this well, illustrated in Fig. 10, indicates
that its pattern is very similar to that for Well A5. The
only difference appears to be in the size of the
anisotropy ratio fluctuations representing lower-
porosity facies. The fluctuations for these facies are
relatively small in the case of Well A3. This is due to
the fact that Well A3 intersects Zone 3 which consists
mainly of grain-dominated facies with only a small
proportion of mud-rich intervals. On the other hand,
Well A5 intersects both Zone 3 and Zone 4. The latter
zone includes considerable proportion of mud-rich
facies. Consequently, a reasonable number of
measurements from both granular and muddy facies
is represented in Well A5 data, resulting in a wider
spectrum of anisotropy ratios values.
DISCUSSION
Based on the results from both wells, it can be
concluded that for the reservoir under consideration,
there is a close relationship between the anisotropy
ratio and porosity. Thus, for a good quality reservoir
with relatively high porosities, anisotropy ratios can
be expected to be relatively low without significant
Fig. 8. Frequency histogram of porosity measurements for Well A5, displaying two populations separated by
approximately 11% porosity.
Parameters Porosity (%)
Samples 108
Arithmetic mean 13.25
Geometric mean 11.13
Harmonic mean 8.51
Standard deviation 6.46
Coefficient of variation 0.49
Skewness -0.17
Kurtosis -1.11
Table 3. Statistical parameters of core porosity
measurements (Well A5).
156 Permeability anisotropy distributions in an Upper Jurassic carbonate reservoir, E. Saudi Arabia
Fig. 9. Plot of anisotropy ratio
(R
3
) versus porosity for Well
A5, displaying two distinct
clusters of points: one is
characterized by higher
porosities and considerably
lower ratios without any
significant fluctuations. The
second is characterized by
lower porosities and higher
anisotropy ratios with
considerable fluctuations.
Fig. 10. Plot of anisotropy
ratio (R
3
) versus porosity for
Well A3, displaying a similar
pattern to the corresponding
plot for Well A5 (Fig. 9).
fluctuations. On the other hand, poor quality reservoir
rocks are associated with highly fluctuating ratios. In
other words, the distribution of permeability
anisotropy in the reservoir is dependent on the facies
distribution. Granular facies with higher porosities are
more homogeneous reservoir rocks, and hence yield
considerably lower and more uniform ratio values.
Muddy and mixed granular-muddy facies represent
greater heterogeneity in terms of permeability
distributions. Therefore, the anisotropy ratios in these
facies will show considerable fluctuations.
Considering the rock-fabric, Lucia et al. (2001)
proposed a more complex model for the Arab-D facies
and recognized three classes each having a wide
spectrum of porosity and permeability values.
Considering their model, it is apparent that to produce
the plot illustrated in Fig. 9, we must have different
k
h
and k
v
relationships above and below the 11%
porosity cut-off. For any facies with porosity greater
than 11%, k
h
and k
v
values are very similar, resulting
in anisotropy ratios close to one. On the other hand,
below 11% porosity, k
h
and k
v
values are highly
variable giving rise to greater fluctuations in
anisotropy ratios: the grain-dominated facies show
probably very similar values of k
h
and k
v
and hence
anisotropy values closer to one, in contrast to mud-
dominated facies with greater anisotropy ratio values.
Investigation of the correlation patterns of k
h
and k
v
measurements below and above the 11% porosity cut-
off revealed that there is almost no correlation between
these measurements below the cut-off. However, a
significant correlation was observed between the same
measurements above the cut-off.
As pointed out above, results provided in this study
are based on the whole core measurements. To
determine zonal anisotropy parameters, up-scaling of
anisotropy ratios would be required. Depending on
the direction of flow with respect to the stratification,
different mean values have been used to represent the
permeability of a stratified sequence (Richardson et
al., 1987; Jensen et al., 1987). Thus, if the flow is
parallel to the strata, the arithmetic mean is used. If
157
A. Sahin et al.
the flow is perpendicular to the strata, the harmonic
mean is considered to be more appropriate. However,
in other cases, where the flow is neither strictly parallel
nor perpendicular, or where different facies are not
clearly stratified, the geometric mean is used (Isaaks
and Srivastava, 1989).
Assuming a relatively uniform stratified sequence
with horizontal flow in our case, the horizontal
permeability values can be averaged using the
arithmetic mean and the vertical permeability using
the harmonic mean as discussed above. Combining
the permeability measurements based on samples
within a uniform zone in this manner, it is possible to
derive the horizontal and the vertical permeability
values, and hence to determine the anisotropy ratio
for the entire zone.
CONCLUSIONS
The main conclusions drawn from this study of a
carbonate reservoir in Eastern Saudi Arabia are
summarized as follows:
1. Statistical distributions representing permeability
and anisotropy ratios display similar patterns with a
dominant positive skew. Mean values for the vertical
permeability are considerably lower than the
corresponding parameters for the horizontal
permeability distributions.
2.Values of the coefficient of variation for the
horizontal permeability and anisotropy ratio are
similar with values close to 1.5, indicating the very
heterogeneous nature of these distributions. This
coefficient for vertical permeability is greater than the
corresponding values for the horizontal permeability,
reflecting greater heterogeneity in the vertical
permeability distribution.
3. Higher fluctuations in anisotropy ratios along wells
can be correlated with low-porosity muddy and/or
mud-rich / granular intervals within the reservoir.
4.Granular facies with higher porosities yield
considerably lower anisotropy ratios without any
significant fluctuations. In other words, the best
reservoir facies are expected to present fewer problems
during production.
5.Very high values of the anisotropy ratio are generally
due to unusually low vertical permeability
measurements recorded in compact and undisturbed
mud-rich intervals acting as barriers to vertical flow.
ACKNOWLEDGEMENTS
This study is partly based on a paper (SPE 81472)
originally presented at the 2003 SPE Middle East Oil
& Gas Show and Conference, held in Bahrain. Saudi
Aramco provided the basic data, and the study was
undertaken at the Research Institute, King Fahd
University of Petroleum and Minerals (KFUPM).
Financial support was provided by Saudi Aramco, Al-
Khafji Joint Operations, and Schlumberger Dhahran
Carbonate Research Center . The authors wish to thank
the management of the above-mentioned organizations
for their support and permission to publish this paper.
Review comments on a previous version by Deborah
Bliefnick and Wayne Wright are acknowledged with
thanks.
Nomenclature
k
h1
= Horizontal permeability (maximum)
k
h2
= Horizontal permeability
k
h
= Average horizontal permeability
k
v
= Vertical permeability
R
1
, R
2
, R
3
= Anisotropy ratios
NPHI = Neutron porosity
PHIT = Total porosity
RHOB = Bulk density
GR = Gamma ray
REFERENCES
CANTRELL, D. L., SWART, P. K., HANDFORD, R. C., KENDALL,
C. G., WESTPHAL, H. 2001.Geology and production
significance of dolomite, Arab-D reservoir, Ghawar Field,
Saudi Arabia. GeoArabia, 6 (1), 45-60.
CORBETT, P. W. M. and JENSEN, J. L., 1992. Estimating mean
permeability: how many measurements do we need? First
Break, 10 (3), 89-94.
FAROUQ, A., ROJAS, S.M., ZHU, T., and DYER, S., 1988. Scaled
model studies of carbon dioxide floods. SPE 18083, Annual
Technical Conference and Exhbibition, October 2-5,
Houston, Texas, pp. 273-284.
FORREST, F. C, Jr., 1971. The reservoir engineering aspect of
waterflooding. Monograph Volume 3, Henry L. Doherty
Series, SPE, New York, 141p.
HENLEY, W., OWENS, W. and CRAIG, F. F., 1961. A scale-
model study of bottom water drives. Journal of Petroleum
Technology, 13 (2), 90-98.
HUGHES, G. W., 1996. A new bioevent stratigraphy of Late
Jurassic Arab-D carbonates of Saudi Arabia. GeoArabia, 1
(3), 417-434.
ISAAKS, E. H., and SRIVASTAVA, R. M., 1989. An Introduction
to Applied Geostatistics. Oxford University Press, New
York, 561pp.
JENSEN, J. L., HINKLEY, D.V., and LAKE, L.W., 1987. A
statistical study of reservoir permeability: distributions,
correlations, and averages. SPE Formation Evaluation, 2 (4),
461-468.
JENSEN, J. L., LAKE, L. W., CORBET, P. W. M. and GOGGIN, D.
J. , 1997. Stati sti cs for Petrol eum Engi neers and
Geoscientists. Prentice Hall, Upper Saddle River, New
Jersey, 390pp.
JOSHI, S.D., and DING, W., 1996. Horizontal well application,
reservoir management. Second International Conference
and Exhibition on Horizontal Well Technology, 18-20
November 1996, Calgary.
LAKE, H. D., 1988. The origins of anisotropy. Journal of Petroleum
Technology, 395-396, April 1988.
LUCIA, F. J., JENNINGS, Jr., J. W., RAHNIS, M., and MEYER, F. O.
2001. Permeability and rock fabric from wireline logs, Arab-
D reservoir, Ghawar field, Saudi Arabia. GeoArabia, 6 (4),
158 Permeability anisotropy distributions in an Upper Jurassic carbonate reservoir, E. Saudi Arabia
619-646
MENOUAR, H. K., and HAKIM, A.A., 1995. Water coning and
critical rates in vertical and horizontal wells. SPE 29877,
SPE Middle East Oil Show and Conference, 11-14 March
1995, Bahrain.
MUSKAT, M., 1937. The flow of homogeneous fluids through
porous medi a, Internati onal Human Resources
Development Corporation, McGraw-Hill, New York.
POWERS, R.W., 1962. Arabian Upper Jurassic carbonate
reservoir rocks. In: W. E. Ham, (Ed), Classification of
carbonate rocks. AAPG Memoir 1, 122-192.
RICHARDSON, J. G., SANGREE, J. B., and SNEIDER, R. M.,
1987. Permeability distributions in reservoirs. Journal of
Petroleum Technology, 39 (10), 1197-1199.
SAHIN, A., and SANER, S., 2001. Statistical distributions and
correlations of petrophysical parameters in the Arab-D
reservoir, Abqaiq oilfield, Eastern Saudi Arabia. Journal of
Petroleum Geology, 24 (1), 101-114.
SPIEGEL, M. R., 1972. Theory and Problems of Statistics,
Schaums Outline Series, McGraw-Hill, New York, 359p.
WILHITE, P. G., 1986. Waterflooding. SPE Textbook Series, 3.
SPE, Richardson, Texas, 326pp.