You are on page 1of 16

Relative Permeability Coupled

Saturation-Height Models on the Basis of


Hydraulic (Flow) Units in a Gas Field
Maclean O. Amabeoku, SPE, David G. Kersey, SPE, Rami H. BinNasser, SPE, and Ali R. Belowi, SPE, Saudi Aramco

Summary
Saturation/height functions on the basis of unique flow units have
been developed as part of an integrated petrophysical analysis of a
gas field. Furthermore, coupling the saturation/height functions
with appropriate relative permeability models has effectively
quantified hydrocarbon saturation, classified producibility of intervals, and defined critical water saturation. The results show that
linking depositional and diagenetic rock fabric with flow units and
then linking the flow units with zones that have similar core capillary pressure and relative permeability relationships have enhanced the utility of the saturation models. The saturation/height
functions provided more-accurate water saturation in the study
field, and potentially they can overcome uncertainties associated
with log interpretation by use of Archie or shaly-sand models.
The saturation/height models were developed from core capillary pressure (Pc) data to calculate water saturation vs. depth,
which is independent of logs. The relative permeability models
were obtained from special-core analysis (SCA). Consequently,
the core-based saturation/height functions can be useful in the
calibration of log-based petrophysical models and with relative
permeability can also be used to estimate water/gas ratios (WGRs)
and critical water saturation.
Capillary pressure and relative permeability curves from SCA
studies were distributed into corresponding flow units, on the basis
of the calculated flow-zone indicators (FZIs). Saturation/height
functions were then developed for each unit and were used to
calculate water saturation in the study field. The most accurate
flow-unit-based saturation model that evolved is a function only of
porosity and of height above the free-water level; it does not require permeability in its application; and it performed better than
the Leverett J-function in this field.
Coupled with hydraulic unit (HU)-based relative permeability
curves, the saturation models may provide more comprehensive
petrophysical interpretation in gas-bearing formations and may
highlight potential differences in reservoir producibility.
Introduction
Models used to calculate water saturation from logs in this gasfield
case study include the deterministic Archie equation (Archie
1942), Waxman and Smits (1968), and an optimizing dual-water
(DW) model presented by Eyvazzadeh et al. (2003).
Extensive laboratory measurements conducted by Amabeoku
et al. (2005a) and Efnik et al. (2006) show variability of the saturation (n) and cementation exponents (m) vertically within the well
and from well to well. This makes the use of single-valued (average) parameters untenable.
The presence of illite, even though in small quantities, has
necessitated the use of the DW model routinely to calculate porosity and water saturation in this field. Illite is filament-like, nonswelling clay that coats grain surfaces. It is thought that the DW
model, which was optimized for this formation, provides moreaccurate water saturation. The model uses as input all available

Copyright 2008 Society of Petroleum Engineers


This paper (SPE 102249) was accepted for presentation at the SPE Annual Technical
Conference and Exhibition, San Antonio, Texas, 2427 September, and revised for publication. Original manuscript received for review 25 June 2006. Revised manuscript received
for review 16 May 2008. Paper peer approved 1 September 2008.

December 2008 SPE Reservoir Evaluation & Engineering

logs, mineral analyses, and electrical parameters, and it solves for


clay-bound water and free fluids in the flushed and unflushed
zones of the wellbore.
The relationship between capillary pressure (Pc) and water
saturation offers a means to estimate water saturation vs. depth,
which is independent of wireline logs and provides the ability to
calibrate log-derived saturations. Saturation/height models, if
implemented successfully, would also minimize, or eliminate, the
uncertainties associated with electrical-parameter measurements.
Some uncertainties that can impact the accuracy of electrical parameters include electrode configuration, saturation and resistivity
equilibration, and incomplete core cleaning. The experimental protocols should also be designed to determine intrinsic saturation and
cementation exponents (n* and m*, respectively) and not apparent
properties in shaly formations such as those discussed in this paper.
All that is required to develop these models are routine core
properties and capillary pressure data.
Several saturation/height models have been proposed and used
in the petroleum industry with different degrees of difficulty and
amount of data required. The Leverett J-function (Leverett 1944)
is one of the most commonly used and is written as
JSw =

0.2166Pc
cos

k
, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (1)

where Pc is the capillary pressure, is the contact angle, and is


interfacial tension between the fluid pair. The ratio (k/) implicitly
recognizes the reservoir quality. A log-log plot of J(Sw) vs. Sw is
nearly linear (Ibrahim et al. 1992; Cuddy 1993), as illustrated in
Fig. 1, and will yield a relationship of the form
Sw =

a
JSwb

, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2)

where a and b are regression coefficients. Substituting the functional form of J(Sw) in the height domain and solving explicitly for
Sw, Eq. 2 transforms to
Sw =

a
0.2166Pcres
cosres

. . . . . . . . . . . . . . . . . . . . . . . . . . . (3)

Cuddy (1993) developed the FOIL function by use of log


data from the Southern North Sea in net-reservoir intervals and
away from bed boundaries. Cuddy did not define the acronym. He
also demonstrated that FOIL can be applied equally to core capillary pressure data and does not require permeability. It provides
correlation between the bulk volume of water (BVW), which is the
product of water saturation and porosity, and the height above
free-water level. Cuddy (1993) presents the FOIL function as
BVW = Sw =

A
HB

, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (4)

where H is height above free-water level and A and B are fitting


parameters. The formulation was derived from the J-function. A
porosity/permeability relation was used to substitute for k in the
J-function. The detailed derivation is provided in Cuddy (1993).
Cuddy showed examples in his paper to demonstrate that, for the
North Sea gas fields he investigated, the BVW at a certain height
1013

under study when a porosity exponent was introduced. The resulting empirical equation was presented by Amabeoku et al. (2005a)
as the Modified FOIL function:
Sw =

A 1
HB C

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (6)

A, B, and C are parameters that are obtained by robust nonlinear


regression and other optimization schemes.

Fig. 1Illustration of development of equation to simplify relationship between Sw and J-function.

above the free-water level was virtually independent of the rock


properties and almost entirely dependent on the height.
The development is based on the observation that the product
of porosity and saturation (Sw) will remain virtually constant in
zones of irreducible water saturation and will increase toward the
free-water level. Although explicitly independent of permeability,
Cuddy suggests separate functions may be used to refine the saturation/height relationships within a particular geological zone,
lithofacies, or HU with fundamentally different pore geometries.
After an appropriate function has been derived, water saturation
can then be calculated by use of
Sw =

A 1
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (5)
HB

Although the capillary pressure behavior is controlled by both


porosity and permeability, as recognized by the J-function, the
implementation of saturation-vs.-height functions, which directly
include permeability as a rock-quality discriminator, requires a
robust permeability predictor. Permeability and porosity data required to compute saturation with the J-function may be available
in cored wells. In such wells, a J-function can be developed and
validated against the core data. Only a limited number of wells in
a field are cored. Porosity can be obtained from logs, but permeability would have to be predicted and mapped in a 3D geocellular
model in the field. This would be an arduous task. The choice of
the FOIL function over the J-function would then be a matter of
expediency. Cuddy suggested constructing separate functions for
different geologic units or lithofacies with distinct and coherent
/k relationships.
Modified FOIL Function. Amabeoku et al. (2005a) first incorporated HU zonation to overcome the initial limitations of the
FOIL function (Cuddy 1993) and observed an improvement in
water-saturation calculation. Mitchell1 had suggested the FOIL
water-saturation model could be modified to fit the core capillary
pressure data by adjusting the porosity term. Further improvement
in water saturation was obtained with the data from the gas field

From personal communication with P. Mitchell, 2004, PGLs capabilities with reference to
tight gas field study.

1014

Geology of Study Field


Prudden2 described the study field as fault-bounded in the west
and north and dipping gently away to the south and east. The
reservoir section consists of terrestrial clastics of PermianCarboniferous age. The field is divided into two reservoir units: an
approximately 200-ft-thick upper Reservoir A and a lower Reservoir B. The two reservoir units are separated by a siltstone unit of
variable thickness (10 ft up to 115 ft). Stratigraphically, the rocks
of Reservoir B continue below the known gas/water contact.
Reservoir A consists of an eolian sequence of fine- to coarsegrained, highly laminated, crossbedded dune sands and sheet
sands. The dune and sheet sands are interspersed with interdune
playas of very fine silty sand with some paleosols near the top of
the unit. The dunes prograde east (wind blowing west to east) as a
series of transverse dunes, which have a strong permeability anisotropy. The original fabric is overprinted with a significant
amount of quartz-overgrowth cements that have reduced porosity
substantially. Despite a composition of more than 95% quartz, the
internal architecture of Reservoir A is complex. The siltstone unit
is also geologically complex and comprises a series of very fine
silty sandstones and pure siltstones. This unit is a barrier between
Reservoirs A and B.
Reservoir B is dominantly sandstone of a fluvial origin, probably braided-stream deposits related to glacial outwash. These
rocks are characteristically quartz-rich sandstones with
extensive quartz overgrowths (>98% quartz). The reservoir section
includes very frequent, horizontal to low-angle stylolites with
associated tension gashes. Although the porosity is often slightly
lower than that of Reservoir A, the lower clay content tends to
improve the permeability, and Reservoir B is regarded as the principal reservoir in this field. Although local reservoir architecture is
also quite complex, Reservoir B is a more continuous sequence
than the overlying Reservoir A. In general, the highest-porosity/permeability units have the lowest amounts of quartz-overgrowth
cements.
Hydraulic (Flow) Unit Zonation
Ebanks (1987) defined an HU as a volume of the total reservoir
rock within which geologic and petrophysical properties that affect
fluid flow are internally consistent and predictably different from
properties of other rock volumes. It is a zone that is continuous
over a defined volume of the reservoir and has similar average
rock properties that affect fluid flow and has similar bedding characteristics. Distribution of flow units is related to facies distribution, but flow-unit boundaries do not necessarily coincide with
facies boundaries.
The HU-zonation scheme devised by Amaefule et al. (1993)
was used to compute and distinguish the different flow units. The
equation is given as
logRQI = logz + logFZI, . . . . . . . . . . . . . . . . . . . . . . . . . (7)
where

z =

, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (8)
1

RQI = 0.0314

k
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (9)

From personal communication with M. Prudden, 2005.

December 2008 SPE Reservoir Evaluation & Engineering

The FZI is calculated by


FZI =

RQI
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (10)
z

The derivation of these simplified equations by Amaefule et al.


(1993) originated from the generalized form of the KozenyCarmen (Kozeny 1927; Carmen 1937) relationship among porosity, permeability, surface area, tortuosity, and pore shape factor.
Amaefules derivation shows that a log-log plot of RQI vs. z will
yield a straight line with a unit slope. The intercept of this straight
line at z1 is the FZI. Core samples that lie on the same straight
line have similar pore-throat characteristics and therefore constitute a flow unit. Samples with different FZI will lie on other
parallel lines.
Fig. 2 shows the plot of reservoir-quality index vs. z calculated from porosity and permeability data from some wells to
determine the HUs. Five units were identified, but the fifth HU
was not considered net-reservoir rock. Therefore, the samples in
HU-5 were excluded from further analysis. Samples that were used
for the determinations of capillary pressure and electrical parameters are denoted by the red squares. Basic properties and computed hydraulic units of these samples are shown in Table 1.
Geologic Characteristics of Samples
Constituting the HUs
The clay content is generally low and ranges from 1% to a maximum of 13% on a bulk basis. Initially, the clay fraction was separated and analyzed. The dominant clays from the analysis of the
clay fraction are kaolinite, dickite, illite/mica, and chlorite.
HU-1. Fig. 3 shows representative thin-section (TS) photomicrograph, scanning electron micrographs (SEMs), and pore-throatsize distribution of the rock that constitutes HU-1. This unit is
characterized by medium-grained, moderately to poorly sorted
quartz arenites. All the detrital grains are subrounded quartz with
authigenic clays comprising approximately 3% of the grain vol-

ume. Most of the clays are pore-filling and grain-replacing kaolinite. Lesser amounts of authigenic illite/mica and illite/smectite are
also present. Quartz overgrowths are the only cements. TS porosity
is characterized by well-developed intergranular pores. Minor
amounts of intraparticle and grain-moldic dissolution porosity are
also present. The pore throats can be described as macropores,
with a large proportion of the pore throats in the 5- to 10-m range.
HU-2. This unit is a lower medium-grained, well-sorted quartz
arenite. All the detrital grains are rounded to subrounded quartz,
with authigenic clays comprising approximately 1% of the grain
volume. Most of the clays are pore-filling illite with lesser amounts
of pore-filling kaolinite. Most of the grains are cemented by quartz
overgrowths. Authigenic illite is also a common cement. TS porosity is characterized by isolated intergranular pores. The pore
system is reduced by quartz overgrowths and authigenic clays. The
TS and SEM are shown in Fig. 4. The dominant pore-throat size is
in the 2- to 5-m range, followed by mesopores less than 2.5 m.
HU-3. HU-3 is characterized by laminated, lower medium- and
fine-grained quartz arenites. As seen in Fig. 5, both medium and
fine-grained laminae are poorly sorted. All the detrital grains are
subangular to subrounded quartz, with authigenic and detrital clays
comprising approximately 6% of the grain volume. TS porosity is
characterized by isolated intergranular pores partially filled with
quartz overgrowths and authigenic clay.
The detrital clays are commonly mica concentrated along bedding laminae. Most of the authigenic clays are pore-filling kaolinite. Smaller amounts of grain-coating illite are also present. Most
of the grains are cemented by quartz overgrowths. Authigenic illite
is also a common cement. The micropores are in the range of 0.025
to 0.25 m.
HU-4. The rock in this unit is a fine-grained, well-sorted sublitharenite. The general rock composition is detrital quartz (89%),

Fig. 2HU zonation for study-field core samples. Intercept of unit-slope lines at z=1 is FZI value.
December 2008 SPE Reservoir Evaluation & Engineering

1015

rock fragments (6%), and authigenic and detrital clay (5%). The
detrital grains are subangular to subrounded. Many of the rock
fragments are deformed as a result of compaction. Clay minerals
include detrital chlorite associated with rock fragments and authigenic pore-filling illite. Cements include quartz overgrowths,
grain-rimming clay, and minor amounts of calcite and dolomite.
The TS and SEM in Fig. 6 show most of the visible porosity is
isolated intergranular pores and secondary porosity in rock fragments. The pores are dominantly micro with a cluster around
0.025 micron.
Mineralogy. The presence of clay minerals can have implications
in the measurements of electrical properties of the rock. Illite grain

coatings can trap water and provide continuous conductive phase


even at lower water saturations.
SCA Data for Sw Calculation
Electrical Parameters. An extensive SCA program was undertaken to acquire data that would be used in all renditions of the
water-saturation equations. Experiments were conducted to determine apparent cementation and saturation exponents. Multiple salinity tests (Co, Cw) were also conducted to ascertain the effects of
clays on log analysis. These tests involved completely saturating a
core sample with varying brine salinities. The conductivity (Co) of
the fully brine-saturated sample is plotted against the conductivity
of the saturating brine (Cw). From these experiments, the intrinsic

Fig. 3TS (left), SEM (middle), and pore-throat-size distribution (right) of HU-1 rock.
1016

December 2008 SPE Reservoir Evaluation & Engineering

Fig. 4TS (left), SEM (middle), and pore-throat-size distribution (right) of HU-2 rock.

formation resistivity factor (F*), intrinsic cementation exponent


(m*), intrinsic saturation exponent (n*), and volume concentration
of exchangeable cations (Qv) used for the Waxman-Smits shalysand water-saturation model can be estimated. Experimental details and examples are given in Amabeoku et al. (2005a).
Reservoirs A and B have an average composite cementation
exponent (m) of 1.89. The saturation exponent (n) for Reservoir A
is 1.79. For Reservoir B, the saturation exponent is 1.91.
Centrifuge Capillary Pressure. Ambient-condition high-speed
centrifuge air/brine capillary pressure data were obtained by use of
core samples from three wells in the study field. The data are given
in Table 2, and the capillary pressure curves are shown in Fig. 7.
Strictly on the basis of the FZI values of the samples, the
capillary pressure data were assigned to their respective HUs.
Samples X718.2, X713.8, and X742.8 belong in HU-1. Sample
X727.3 was assigned to HU-4. The other samples were assigned to
HU-2 and HU-3, as appropriate.
Saturation/Height Models by Use of Core
Capillary Pressure Data
Centrifuge capillary pressure data were used in the developments
of the saturation-height models that follow. The ambientconditions air/brine centrifuge capillary data were converted to
reservoir conditions by use of the following equation:

cosres
Pc-res = Pclab
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (11)
coslab

The relationship between capillary pressure and the height


above free-water level (H) is given by
Pc = 0.433*Hw g, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (12)
where, in oilfield units, Pc is in psi; H is in feet; and w and g
are both in g/cm3. 0.433 is a conversion factor from laboratory to
field units.
These definitions were then used to develop practical forms of
the J-function, FOIL function, and the modified FOIL function
cited previously in Eqs. 1, 5, and 6, respectively, for each of the
four HUs.
Practical HU-Based Saturation Functions
J-Function. Eq. 12 was substituted for Pc in Eq. 3 for further
analysis. The J-function/saturation relationship was established by
robust nonlinear regression. The resulting model was transformed
so that water saturation can be solved for in terms of the reservoir
properties and height. Eqs. 13 through 16 are the practical forms
for HUs 1 to 4, respectively
Sw1 =

Sw2 =

0.2249
0.0.094w gH
cosres

0.5150

0.3739
0.094w gH
cosres

0.3784

. . . . . . . . . . . . . . . . . (13)

. . . . . . . . . . . . . . . . . . (14)

Fig. 5TS (left), SEM (middle), and pore-throat-size distribution (right) of HU-3 rock.
December 2008 SPE Reservoir Evaluation & Engineering

1017

Fig. 6TS (left), SEM (middle), and pore-throat-size distribution (right) of HU-4 rock.

Sw3 =

Sw4 =

0.2380
0.094w gH
cosres

0.4301

0.2038

0.094w gH
cosres

0.4082

. . . . . . . . . . . . . . . . . . (15)

. . . . . . . . . . . . . . . . . . (16)

Sw4 =

Sw1 =

FOIL Function. The FOIL function is written as BVWSw.


Having established the relationship with the height above the freewater level and the bulk volume of water, Sw can now be calculated
explicitly for the different HUs by use of Eqs. 17 through 20:

Sw2 =

Sw2 =
Sw3 =

1018

0.2032
H 0.5542
0.1727
0.3857

0.2132
H0.4651

0.4082

1
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (20)

Modified FOIL Function. The multiple nonlinear regression algorithm was used to obtain the parameters for Modified FOIL. The
final forms of the function for the HUs are given in Eqs. 21
through 24:

Sw1, Sw2, Sw3, and Sw4 are the water-saturation calculation


equations for HUs 1 through 4, respectively.

Sw1 =

0.1705

1
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (17)

Sw3 =

1
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (18)

Sw4 =

1
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (19)

2.897
H0.5485

0.5239
0.3765

2.9214
H0.4604
0.2476
0.4082

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (21)

0.2223

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (22)

0.5353

0.1528
1

0.8486

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (23)

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (24)

The water saturation can now be calculated with Eqs. 13 through


24, after substituting the fluid properties provided in Table 3 or

December 2008 SPE Reservoir Evaluation & Engineering

relation coefficient of 0.968; and FOIL function had a correlation


coefficient of 0.952.

Fig. 7Air/brine centrifuge capillary pressure curves.

those deemed to be appropriate for the field under study. The


regression parameters (A, B, C) shown in the equations above and
the fluid densities in Table 3 are specific to this field under study
and would vary for different data sets. The interfacial tensions and
contact angles are generally assumed values.
Comparison Between Saturation-Model
Performance and Core Data
Water saturations were calculated for each of the following models: J-function, FOIL function, and Modified FOIL function, and
these were compared with the measured water saturations from the
core capillary pressure experiments. For illustrative purposes,
Figs. 8 through 10 show plots of model-derived water saturation
vs. core water saturation for HU-1. Rock properties of the samples
were used in the calculations.
Among the three models, Modified FOIL performed the best,
with a correlation coefficient of 0.985 between core water saturation and predicted water saturation; J-function followed, with cor-

Applicability of the Saturation/Height Models for


Reservoir Water-Saturation Calculations
From HU classification, Reservoir B rock is of better quality than
that of Reservoir A. Whereas the B reservoir is mostly HU-1, the
A reservoir is predominantly HU-4.
A permeability-modeling approach developed in Amabeoku
et al. (2005b) was used with sparse core data to develop a continuous permeability trace in each well. This was done so that the
performance of the J-function could be evaluated. The laboratoryconditions air/brine centrifuge capillary pressure data were converted to reservoir conditions by use of the data in Table 2.
Routinely, water saturation in this field is calculated by use of
deterministic Archie expressions and by an optimizing DW model.
Of these two methods, DW is believed to be superior. Each of the
saturation models was run across Reservoirs A and B. Fig. 11
compares the performance of each of the models to DW.
The lithology track (far left) shows that Reservoirs A and B are
very clean. Tracks 4 and 5 (from left) show the J-functions for
HU-1 and HU-4, respectively. The J-functions are featureless in
Reservoir A and calculate water saturations that are higher than the
DW water saturations. In Reservoir B, both models compare reasonably well with DW. In Track 6, the HU-1 FOIL model matches
DW quite well in Reservoir A and most of Reservoir B. It deviates
close to the base of the reservoir. The HU-4 FOIL model shows a

Fig. 8Comparison between J-function predicted water saturation and core water saturation for HU-1.
December 2008 SPE Reservoir Evaluation & Engineering

1019

Fig. 9Comparison between FOIL-function predicted water saturation and core water saturation for HU-1.

good match in Reservoir A but fails in Reservoir B. Tracks 8 and


9 show the performance of the Modified-FOIL functions for HU-1
and HU-4, respectively. The Modified-FOIL HU-1 model exhibits
the best match of all in Reservoir B. The water saturation in
Reservoir A is too low. That is not surprising: HU-1 was developed for Reservoir B and was not expected or intended to be
applied to Reservoir A. Modified FOIL for HU-4 calculates good
water saturation in Reservoir A. The water saturation in Reservoir
B is higher and is considered to be wrong. Again, HU-4 models
were not intended to be used in Reservoir B.
A catalog of all the wells evaluated shows the following common observations:
1. The J-function models calculate water saturations reasonably
well in Reservoir B but unreliably in Reservoir A. Good permeability models are required, and this can limit their application.

2. FOIL for HU-1 performs very well in reservoir A but can be


questionable in Reservoir B. The FOIL HU-4 model works well in
Reservoir A but not in B.
3. The Modified-FOIL HU-1 model works the best in B. The
HU-4 model is also good in Reservoir A but not as consistently
good as FOIL HU-1 in Reservoir A.
With the aforementioned observations, it became practical and
necessary to take advantage of the reservoir tops and develop a
hybrid model that is fit for purpose in this field. The computer code
would use a simple logic, such as
If TopA, compute Sw with FOIL HU-1. If TopB, compute
Sw with Modified-FOIL HU-1.
Relative Permeability. Four sets of steady-state relative permeability data at reservoir conditions were obtained on single-core

Fig. 10Comparison between Modified-FOIL-function predicted water saturation and core water saturation for HU-1.
1020

December 2008 SPE Reservoir Evaluation & Engineering

Fig. 11Comparison of the HU saturation models in study field.

plugs that are representative HU-1 and HU-4. Table 4 lists the
properties of the core plugs constituting the HUs.
The two sets of steady-state relative permeability data from
each HU were normalized, averaged, and denormalized to obtain
the relative permeability curves used in these analyses. Techniques
for normalizing relative permeability are available in Ahmed
(2001). Figs. 12 and 13 illustrate averaged relative permeability
curves for HU-1 and HU-4, respectively.
Water/gas relative permeability ratios were also calculated and
plotted. Fig. 14 is an overlay of HU-1 and HU-4 relative permeability ratios. The plots show that for any given water saturation,
krw/krg is higher for HU-4. This indicates that HU-4 is more likely
to produce water in comparison with HU-1, if the water was above
the irreducible saturation.
The relationship between the relative permeability ratio and
water saturation can be written as given by Ahmed (2001):
krw
= aebSw, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (25)
krg
where a and b are coefficients that are obtained from regression
by use of the linear portion of the semilog plot of krw/krg vs.
water saturation.
December 2008 SPE Reservoir Evaluation & Engineering

Relative permeabilities were also calculated from mercuryinjection capillary pressure (MICP) and centrifuge capillary pressure data, by use of the Brooks-Corey (1966) method. This was
done for comparison and to establish the feasibility of this approach in areas where experimental relative permeability data
might not be readily available. The procedure is first to obtain
from the capillary pressure data as follows:
1

Pc = PeS*
w , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (26)

where Pe is the entry pressure and is the pore-distribution index.


Sw
* is the normalized wetting-phase saturation, and it is defined, for
the drainage case, as
S*w =

Sw Swir
, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (27)
1 Swir

where Swir is the irreducible water saturation.


Eq. 27 was used when the residual nonwetting-phase (gas)
saturation was low and could not be measured accurately. Otherwise, adaptation of the Corey model for the imbibition case would
take the form
1021

S*
w=

Sw Swi
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (28)
1 Swi Snwr

where Swi and Snwr are the initial water saturation and the nonwetting phase saturation, respectively.
The wetting-phase and nonwetting-phase relative permeabilities (krw, krnw), respectively, are given by
krw = S*
w

2+3

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (29)
2+
.

2
krnw = 1 S*
w 1 S*
w

. . . . . . . . . . . . . . . . . . . . . . . . . . (30)

Fig. 15 is an overlay of relative permeability curves obtained


from capillary pressure data on the averaged steady-state data for
HU-1. The abscissa uses the normalized water saturation, calculated with Eq. 27, to reconcile the effect of the different irreducible
water saturations obtained in the cores. In the steady-state calculation of Sw
*, the residual-gas saturation is subtracted from the
denominator, and the expression is that given in Eq. 28.
The relative permeability curves derived from MICP data and
from centrifuge capillary pressure agree very well in nearly all the
analyses discussed in this field. This indicates that, for water-wet

and less-complex rocks, the less expensive and faster-to-measure


MICP data can be used in place of the more expensive air/brine
capillary pressure data but should not be substituted for experimental relative permeability data. Moreover, pore-throat-size distribution and structural information are also extracted from MICP data.
The difference between the steady-state gas relative permeability and those calculated with capillary pressure data is not significant. The steady-state water relative permeability is generally less,
as the normalized water saturation increases beyond 50%.
Fig. 16 confirms the similarity between the capillary pressure
relative permeability ratios. In the middle to upper ranges of water
saturation, all three ratios agree closely. The divergence between
the steady-state ratio and the capillary pressure ratios occurs at the
lower water saturation. In spite of this, relative permeability from
capillary pressure data provides very good approximation for
semiquantitative analysis. This is particularly relevant because it is
the linear part of the curve (middle section) that generally is used
to develop the relationship between the relative permeability ratio
and water saturation given in Eq. 25.
Coupling Relative Permeability With
Saturation Functions
Steady-state relative permeability data were used in these analyses.
The data derived from capillary pressure could equally be used.
The functional forms of the relative permeability ratio relationship
are given by the following:
for HU-1,
krw
= 3.2 106e19.48Sw1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (31)
krg
for HU-4,
krw
= 6.98 106e22.06Sw4, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (32)
krg
where Sw1 and Sw4 are the water-saturation values calculated with
the saturation/height models for HU-1 and HU-4, respectively.
WGR
Water producibility in this analysis is presented as the volume of
water that would be produced with 1 scf of gas. This is the WGR,
with units of STB/scf, and it is given as

Fig. 12Averaged steady-state relative permeability of HU-1 samples.


1022

December 2008 SPE Reservoir Evaluation & Engineering

Fig. 13Averaged steady-state relative permeability of HU-4 samples.

WGR =

Bg g krw
, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (33)
Bw w krg

where Bg and Bw are gas and water formation factors, respectively,


and g and w are gas and water viscosities, respectively. Substi-

tuting the relative permeability ratio for HU-1, for example,


WGR becomes:
WGR1 =

Bg g
3.2 106e19.48Sw1. . . . . . . . . . . . . . . . . . . . . (34)
Bw w

Similarly, for HU-4, the ratio would be calculated as:


WGR4 =

Bg g
6.98 106e22.06Sw4. . . . . . . . . . . . . . . . . . . . (35)
Bw w

With the applicable values of Bg, Bw, g, and w, as shown in


Table 5, the WGR can be calculated for saturation at each depth.

Fig. 14Relative permeability ratios of HU-1 and HU-4 samples.


December 2008 SPE Reservoir Evaluation & Engineering

Assessment of WGR and Verification With


Test Data
Having established the utility of the HU-based saturation functions, we shall now incorporate relative permeabilities to calculate
WGR. These calculations are based on the single hybrid saturation
model from the preceding section. Remarks on water producibility
are based on the saturation/height model. The DW water saturation
is plotted alongside for comparison purposes.
In Fig. 17, the mirror image of the gamma ray (GR) log in
Track 2 (from left) shows the thick porous sands in Reservoirs A
and B in Well 5X. The density-neutron overlay in Track 3 and
the water saturation in Track 6 show these sands to be gas-filled.
Track 7 shows the WGR in STB per million scf of produced
gas. This analysis indicates that there will be no water production
from completions in the reservoir intervals. Spikes in WGR occur
at very low porosity and at the base of Reservoir A. The conventional analysis also indicates that the water is bound and will not
be produced.
Well 4X in Fig. 18 has cased-hole test data obtained across
Reservoir A and part of Reservoir B. There was no water production. The test showed very high gas rate. Relatively small quantity
of condensate was also produced. As in Well 5X, the apparently
high water saturation is a computational artifact because of the low
porosity in the denominator (Eqs. 21 through 24). Consequently,
WGR spikes in Track 7 can be regarded as noise. Moreover, the
density-neutron overlay near the top of Reservoir A shows the
spiky intervals not to be gas sands. These are shaly or silty. In the
1023

Fig. 15Overlay of relative permeabilities calculated from capillary pressure data by the Brooks-Corey method with steady-state
relative permeability data.

reservoir sands, the saturation and WGR are well behaved and are
consistent with the test data.
Fig. 19 (Well 1X) has more-extensive test data. DST #1 across
the top half of Reservoir A produced all gas and no water. CH Test
#1 across the top 60 ft of Reservoir B produced all gas. Another

CH test (CH Test #2), across all of Reservoir B and approximately


5 ft below the free-water level, produced gas.
Discussion of Results
Through the HUs concept, saturation/height relationships have
been developed to honor differences in lithofacies or pore geometries. Generally, the Modified-FOIL model is the most direct and
the easiest to use. The respective HU-based Modified-FOIL models also performed the best in the wells analyzed. It was observed
the HU-1 FOIL model developed for reservoir B worked very well
in Reservoir A, which has predominantly HU-4-type rock. With
these observations, it became practical and necessary to take advantage of the reservoir tops and develop a hybrid model that is fit
for purpose in this field. The computer code was modified in a way
that allows the Modified-FOIL HU-1 model to be used for saturation calculations in Reservoir B, while the HU-1 FOIL model is
used to calculate saturation in Reservoir A, all in one pass without
intervention by the user. Both the FOIL and Modified-FOIL models do not require permeability in their applications. But application of the J function is constrained by the requirement that good
permeability models be available.
Water/gas relative permeability curves were developed for each
HU. These were then used successfully to calculate the potential of
water mobility in each reservoir. A very noteworthy observation is

Fig. 16Comparisons of relative permeability ratios calculated


from capillary pressure data by the Brooks-Corey method with
steady-state relative permeability ratio.
1024

December 2008 SPE Reservoir Evaluation & Engineering

Fig. 17Calculation of WGR and impact on water mobility in a well that was not tested.

that relative permeability calculated from capillary pressure data


agreed closely with steady-state relative permeability. This observation has important time and cost implications. Steady-state relative permeability measurements are time-consuming, whereas centrifuge capillary pressure data can be acquired in a relatively short
time at a significantly lower cost. Capillary pressure from this gas
field can now be used confidently to develop not only for saturation functions but also for relative permeability in this field.
Comparison between calculated WGR and well-test data confirm the robustness of the relative permeability-coupled saturation
December 2008 SPE Reservoir Evaluation & Engineering

models. Generally, WGR is less than 0.1 STB/million scf, except


where the porosity is low (less than the cutoff value) and near the
gas/water contact. At no point did the models contradict the test
results. These models can serve as guides to augment well tests and
can also be used to make well-completion decisions.
Conclusions
Saturation/height models from core capillary pressure (Pc) data are
an effective way to determine fluid saturation. This study developed saturation/height models by
1025

Fig. 18Water-mobility prediction vs. well-test results.

1) Linking depositional and diagenetic rock fabric to HUs.


2) Linking the HUs to zones with similar capillary pressure relationships.
3) Determining saturation/height functions for each HU with the
modified-FOIL function.
Improved fluid-saturation profiles derived from Modified FOIL
can be used to validate log models and parameters. Because this
method does not require permeability in field application, it can be
used in uncored wells.
HU-based relative permeability models have been developed
for each of the reservoirs in this gas field. Capillary pressure data
have presented a cost-effective way to develop both saturation and
relative permeability models in this gas field.
Coupled with HU-based relative permeability curves, the
saturation models will provide more-comprehensive petro1026

physical interpretation in gas-bearing formations and will predict


water producibility. The range of WGR indicates no water would
be produced in the tested intervals. This is validated by the
test results.
Nomenclature
a fitting parameter in relative-permeability-ratio/
water-saturation relationship
A fitting parameter in saturation functions
b fitting parameter in relative-permeability-ratio/
water-saturation relationship
B fitting parameter in saturation functions
Bg gas formation volume factor, RB/Mscf
December 2008 SPE Reservoir Evaluation & Engineering

Fig. 19Validation of mobility calculations with isolated cased-hole (CH) test results.

Bw water formation volume factor, RB/STB


C fitting parameter in saturation functions
Co conductivity of 100%-brine-saturated sand,
mhom1
Ct total conductivity of partially brine-saturated sand,
mhom1
Cw conductivity of brine solution, mhom1
F apparent formation-resistivity factor (1/m)
F* intrinsic formation-resistivity factor (1/m*)
H height above free-water level, ft
J(Sw) Leverett J-function
k permeability, md
krg relative permeability to gas
krnw relative permeability to nonwetting phase
krw relative permeability to water
m cementation exponent for clean sand
m* intrinsic cementation exponent
December 2008 SPE Reservoir Evaluation & Engineering

n saturation exponent for clean sand


n* intrinsic saturation exponent
Pc capillary pressure, psi
Pc-lab capillary pressure in laboratory fluid pair, psi
Pc-res equivalent capillary pressure in reservoir fluid pair,
psi
Pe capillary entry pressure, psi
Qv volume concentration of clay exchange cations,
meq/m?
Snwr irreducible gas saturation, fraction
Sw water saturation, fraction
S*w normalized water saturation
Swi irreducible water saturation, fraction
Swir, irreducible water saturation, fraction
contact angle, degrees
lab contact angle at laboratory conditions, degrees
res contact angle at reservoir conditions, degrees
1027


g
w
g
w

cos

coslab
cosres

pore-distribution index
gas viscosity, cp
water viscosity, cp
gas density, g/cm3
water density, g/cm3
interfacial tension, dyne/cm
product of interfacial tension and contact angle,
dyne/cm
product of interfacial tension and contact angle at
laboratory conditions, dyne/cm
product of interfacial tension and contact angle at
reservoir conditions, dyne/cm
porosity, fraction
pore-volume/grain-volume ratio

Acknowledgments
The authors thank the management of Saudi Aramco for their
support and for permission to publish this paper.
References
Ahmed, T. 2001. Reservoir Engineering Handbook, second edition, 298
301, 304309. Oxford, UK: Elsevier.
Amabeoku, M.O., Kersey, D.G., BinNasser, R.H., Al-Waheed, H.H., and
Al-Belowi, A.R. 2005a. Incorporating Hydraulic Units Concepts in
Saturation-Height Modeling in a Gas Field. Paper SPE 93763 presented
at the SPE Asia Pacific Oil & Gas Conference and Exhibition, Jakarta,
57 April. DOI: 10.2118/93763-MS.
Amabeoku, M.O., Lin, C., Al-Khalifa, A., Cole, J., Dahan, M., Jarlow, J.,
and Ajufo, A. 2005b. Use of Fuzzy-Logic Permeability Models to
Facilitate 3D Geocellular Modeling and Reservoir Simulation: Impact
on Business. Paper IPTC 10152 presented at the International Petroleum Technology Conference, Doha, 2123 November. DOI: 10.2523/
10152-MS.
Amaefule, J.O., Altunbay, M., Tiab, D., Kersey, D.G., and Keelan, D.K.
1993. Enhanced Reservoir Description: Using Core and Log Data to
Identify Hydraulic (Flow) Units and Predict Permeability in Uncored
Intervals/Wells. Paper SPE 26436 presented at the SPE Annual Technical Conference and Exhibition, Houston, 36 October. DOI: 10.2118/
26436-MS.
Archie, G.E. 1942. The Electrical Resistivity Log as an Aid in Determining
Some Reservoir Characteristics. Trans., AIME, 146: 5462.
Brooks, R.H. and Corey, A.T. 1966. Properties of porous media affecting
fluid flow. ASCE J. of Irrigation and Drainage 101: 8592.
Carmen, P.C. 1937. Fluid Flow Through Granular Beds. Trans., Institute of
Chemical Engineers London, 15: 150166.
Cuddy, S. 1993. The FOIL FunctionA Simple, Convincing Model
For Calculating Water Saturations In Southern North Sea Gas Fields.
Paper H1-17 presented at the SPWLA Annual Logging Symposium,
Calgary.
Ebanks, W.J. Jr. 1987. Flow unit conceptIntegrated approach to reservoir description for engineering projects. AAPG Bulletin 71 (5): 551
552.
Efnik, M.S., Dernaika, M., and Kalam, M.Z. 2006. Evaluation of water
saturation from laboratory to logs. Paper SCA2006-56 presented at the
International Symposium of the Society of Core Analysts, Trondheim,
Norway, 1216 September.
Eyvazzadeh, R.Y., Cheshire, S.G., Nasser, R.H., and Kersey, D.G. 2003.
Optimizing Petrophysics: The Ghawar Field, Saudi Arabia. Paper SPE
81477 presented at the Middle East Oil Show, Bahrain, 912 June.
DOI: 10.2118/81477-MS.

1028

Ibrahim, A., Bassiouni, Z., and Desbrandes, R. 1992. Determination of


relative permeability curves in tight gas sands using log data. Paper
SS presented at the SPWLA Annual Logging Symposium, 1417
June.
Kozeny, J. 1927. ber Kapillare Leitung des Wassers im Boden. Sitzungsberichte der Wiener Akademie des Wissenschaften, 136: 271306.
Leverett, M.C. 1944. Capillary Behavior in Porous Solids. Trans., AIME,
142: 152169.
Waxman, M.H. and Smits, L.J.M. 1968. Electrical Conductivities in OilBearing Shaly Sands. SPEJ 8 (2): 107122; Trans., AIME, 243. SPE1863-PA. DOI: 10.2118/1863-PA.

SI Metric Conversion Factors


bbl 1.589 873
E01 m3
cp 1.0*
E03 Pas
dyne 1.0*
E02 mN
ft 3.048*
E01 m
E02 m2
ft2 9.290 304*
psi 6.894 757
E+00 kPa
*Conversion factor is exact.

Maclean Amabeoku is Coordinator of Special Studies in the


reservoir description and simulation department at Saudi
Aramco. E-mail: maclean.amabeoku@aramco.com. In that
capacity, he formulates and oversees petrophysical studies,
with the sole objective of improving parameters and models
used in formation evaluation and minimizing uncertainties.
Amabeoku has done extensive work on core-log data comparison and integration, permeability prediction using innovative soft computing technologies, saturation-height modeling,
and relative permeability normalization. He holds a BS degree
from the University of Kansas, and MS and PhD degrees from
the University of Southern California, all in petroleum engineering. David Kersey is a senior petroleum engineering consultant
in the reservoir description and simulation department, Saudi
Aramco. E-mail: david.kersey@aramco.com. He currently
oversees strategic planning, training, and evaluation of new
technology. Kerseys previous position with Saudi Aramco was
Supervisor, Petrophysical and Special Studies Unit, Reservoir
Description Division. He did his undergraduate work at the University of Southern California and his graduate work at Harvard
University. Rami BinNasser is Supervisor of Special Studies in the
reservoir description and simulation department at Saudi
Aramco. E-mail: rami.binnasser@aramco.com. In that capacity, he sets future direction for petrophysics by evaluating new
logging and data processing technologies to improve formation evaluation. BinNasser has had several assignments as
petrophysicist in the Gas & Exploration Unit, production engineer and reservoir management engineer. He holds a BS degree in mechanical engineering from King Fahd University of
Petroleum and Minerals in Dhahran, Saudi Arabia. Ali Al-Belowi
is Supervisor of Udhailiyah area petrophysics in the reservoir
description and simulation department at Saudi Aramco. Email: ali.below@aramco.com. In this position, he ensures that
production and openhole logs are analyzed promptly, utilizing
the most complete and accurate petrophysical models. AlBelowi has extensive experience in petrophysical evaluations
of exploration and development fields. He holds a BS degree in
petroleum engineering from King Saud University in Riyadh,
Saudi Arabia.

December 2008 SPE Reservoir Evaluation & Engineering

You might also like