SPE 78529

Geological Modeling of a Tight Carbonate Reservoir for Improved Reservoir

Management of a Miscible WAG Injection Project, Abu Dhabi, U.A.E.

J .S. Gomes, M.T. Ribeiro, P. Fouchard, B.N.Twombley, S. Negahban, S. Al-Baker
Copyright 2002, Society of Petroleum Engineers Inc.

This paper was prepared for presentation at the 10
th
Abu Dhabi International Petroleum
Exhibition and Conference.

This paper was selected for presentation by an SPE Program Committee following review
of information contained in an abstract submitted by the author(s). Contents of the paper,
as presented, have not been reviewed by the Society of Petroleum Engineers and are
subject to correction by the author(s). The material, as presented, does not necessarily
reflect any position of the Society of Petroleum Engineers, its officers, or members. Papers
presented at SPE meetings are subject to publication review by Editorial Committees of the
Society of Petroleum Engineers. Electronic reproduction, distribution, or storage of any part
of this paper for commercial purposes without the written consent of the Society of
Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an
abstract of not more than 300 words; illustrations may not be copied. The abstract must
contain conspicuous acknowledgment of where and by whom the paper was presented.
Write Librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-
952-9435.

Abstract
This paper presents the methodologies adopted to model a
tight carbonate reservoir located in Abu Dhabi and to better
predict its performance (completed with horizontal wells)
under a water-alternating-gas (WAG) process. The model is
built through integrating geological, geophysical,
petrophysical, geomechanical, and geostatistical
information.

The large-scale reservoir framework is built by integrating
horizontal wells and 3D seismic data. Horizontal well results
are used to improve the velocity modeling and depth
conversion. The fine scale reservoir zonation is based on
lithostratigraphic correlations derived from the porosity and
micro-resistivity logs. Stylolitic intervals are used as
stratigraphic markers to guide the reservoir zonation. The
Porosity model is derived from a high resolution stochastic
seismic inversion, and the permeability model is generated
using cloud transforms with P-Fields applied by reservoir
rock types. High-pressure mercury injection data is used to
define reservoir rock types. Lorenz plots have been applied
and found to be a useful technique for capturing the
heterogeneity of the reservoir and determining the main
flow units. Fracture analysis is conducted using cores and
image logs (FMI). A geomechanical study is performed to
assess the orientation of the horizontal wells in the field. A
discussion on the orientation of the horizontal wells with
respect to maximum principal stress versus
productivity/injectivity is also addressed in this paper. A
mechanistic compositional flow model is built to perform
sensitivity analyses on various WAG schemes (cycle, ratio,
etc). A full field compositional model is subsequently built
to evaluate the field performance under various development
scenarios. The field is scheduled to come on stream by
December of 2005.

Field History
The field was discovered in 1969. Reflection seismic had
defined a number of structure closures at several
stratigraphic levels, and the discovery well, W-1, was drilled
to test one of these structures. Oil and gas shows were
recorded when the well penetrated the main reservoir
interval. Subsequent tests proved the commercial viability of
the structure. In 1995 a 3D seismic survey was acquired and
a new re-interpretation performed. The combined evaluation
of OH logs, well test and 3D seismic results provided some
encouragement, and in 1999 the field was declared
commercial.
Between 1994 and 1999 ADCO implemented an Early
Production Scheme (EPS) to evaluate the well performance,
with vertical and horizontal wells. Based on the successful
results of horizontal well performance a field development
plan was devised consisting of a line drive with 20
horizontal producers and 15 horizontal injectors under a
WAG process (Fig 1). The development plan calls for
20,000 BOPD by end of 2005, maintaining the plateau for a
specified period of time, and achieving a high recovery
efficiency. Production forecasts were based on
compositional models which are used to predict and monitor
the reservoir performance during Phase-1 and also to assist
the subsequent phases of development beyond 2015. The
latter was based on detailed static (geological) modeling, the
subject of this paper.

Seismic data
Three generations of seismic were acquired between 1962
and 1995, with the latest being a 3D survey recorded in
1994 and 1995 by Western Geophysical. This survey covers
1334 sq. km over the field area with a 25 x 25 m. bin size
that translates into 2.1 million traces providing a dense grid
of information of the subsurface.
The original processing was completed in 1997 by Western
Geophysical and the seismic data quality was good enough
to carry out accurate structural interpretation. However,
despite favorable surface conditions, the seismic data
suffered from stacking velocity dispersion that had a
significant impact on amplitude values.
Partial reprocessing was then undertaken in 1999 with
special attention to stacking velocity picking. A new seismic
2 J .S. Gomes, M.T. Ribeiro, P. Fouchard, B.N.Twombley, S. Negahban, S. Al-Baker [SPE 78529]
cube was produced for reliable acoustic impedance
inversion and seismic reservoir characterization study.

Seismic interpretation
The data quality is good and the well to seismic calibrations
provide an excellent match that allows both top and base of
the reservoir to be picked and autotracked with confidence.
Seismic interpretation and the structural time maps highlight
that the Field lies on a broad anticline trending NE-SW,
characterized by several low relief structures separated by a
complex fault pattern (Fig 2). As the structure is very flat
(25 ms TWT vertical closure at the crest), time to depth
conversion is critical as 25 ft/sec. error in the average
velocity (less than 0.5%) from surface to reservoir level
gives 30 ft error on the depth map. As such, even if it looks
small, the impact on the closed surface and the volumetric
estimates is dramatic. To minimize uncertainties in velocity
modeling and depth conversion, horizontal well penetration
points at EP and TD have been used.
The deeper zones that have a very poor seismic reflector are
built in depth by stacking an isochore map to the main
reservoir depth map.
The other main issue comes from the picked time itself that
is corrupted by unsolved time statics located near the
surface, generating some shifts upwards or downwards at
reservoir level. For example by smoothing a shallow horizon
TWT map and producing the residual between the smoothed
and raw time maps, Thin and Thick anomalies are
observed that may be related to long period statics.
This is highlighted by problems encountered at various well
locations. These wells came deeper than expected because
the time was too short (5ms TWT is equivalent to 30ft). The
opposite case was observed where doubtful lows are present
in time that corresponds to static anomalies. The new
seismic reprocessing pilot has almost removed these lows
(which support this explanation).
On the other hand, it is very dangerous to entirely believe
that every shallow high or low is not reality because by
doing so real geology can be removed. This is particular
important in very gentle structures which is the case of this
field. The ongoing full reprocessing together with new deep
upholes and new static model will bring a global solution to
this serious issue.

Fault Interpretation
The 3D seismic of the area revealed two different fault
trends (N75W and N45W), composed of arrays of smaller
en echelon faults (Fig 3). These fault trends are interpreted
as conjugate shears. The en echelon fault segment lengths
range from 0.5 to 4 km. The sense of overstep between
segments suggests that the two sets of conjugate faults have
opposite senses of lateral displacement. The N75W trending
set shows clear evidence of dextral shear while the N45W
trending set is apparently sinistral. Vertical and lateral
throws are present in individual fault segments.

Both the N75W and the N45W fault trends are well imaged
with 3D seismic, especially with the use of azimuth displays
(Fig 4). The N75W dextral strike-slip fault zone is
characterised by pop-up structures indicative of a
compressive component (transpressional), while the N45W
fault trends are characterised by small elongated grabens, as
a reaction to an extensional regime (transtensional). The
conjugate directions of these faults define a N60W
maximum horizontal stress associated with the Late
Cretaceous compression as a result of the Ophiolite
obduction in Oman.

The N75W fault trend cuts the Phase-I development area
into northern and southern blocks (Fig 1). Fault offset and
intensity is greater in the Phase-I area (at crestal location)
than in the downflank areas.

The faults tend to lose their throw and die out in less
competent units above and below the main reservoir
intervals. The vertical throw of these faults also diminishes
from 40-50 feet near the crest to zero off structure. The fault
trends and throws in this field exhibit a remarkable
similarity with the faults of a nearby field to the west (Fig
3).

Several parallel narrow grabens can be seen on the field
along the N45W trend faults and it is believed they have a
significant impact on the rock property distribution. Seismic
amplitude maps at reservoir levels highlight abrupt
amplitude variations across the graben trends indicating a
possible barrier to fluids and a control of the diagenesis
mechanism.

Because of the en echelon nature of these faults complete
sealing efficiency, at the field scale, is unlikely. Although
small scale (localised) barriers or "fluid flow baffles" are
present, at the field scale fluid communication between the
northern and southern block is expected.

Geostatistical Seismic Inversion
The methodology uses J ason software that honors seismic,
petrophysical and geological data. The generation of
multiple equi-probable models allows rock property
uncertainties to be addressed.
A previous Deterministic Acoustic Impedance inversion
gave valuable information about the lateral variation in rock
properties for the main reservoir. However the result
consists of a single volume at seismic resolution that cannot
resolve the very thin reservoir units, and the mean porosity
prediction has not enough accuracy to be used in geological
modelling. The reservoir interval is characterised by tight
non-reservoir carbonate at both top and bottom and also thin
stylolitic streaks of tight non-reservoir carbonate within the
porous intervals.

A geostatistical inversion approach was therefore launched
integrating well, seismic and geological data in order to
resolve the vertical heterogeneity of the reservoir. In this
process, acoustic impedance inversion is run with
simultaneous lithology simulation that lead to multiple equi-
probable 3D property models (AI and lithotype). In a second
step, multiple equi-probable Porosity models are simulated
in the time domain and then time-depth converted. Acoustic
impedance and porosity cross-section models are exhibited
in Figs 5 & 6. The net pore volume above OWC is used to
[SPE 78529] Geological Modeling of a Tight Carbonate Reservoir for Improved Reservoir Management of a Miscible 3
WAG Injection Project, Abu Dhabi, U.A.E.
rank the various models, so that the P15, P50 and P85
models were exported to geological modeling packages.

Emphasis was given to the need to perform the geostatistical
seismic inversion within the same framework as the one
used in the geological model in order to preserve the
reliability of the spatial correlation.

A 3D geological model derived without the seismic
constraint compared with those integrating either
deterministic or stochastic inversion results, highlights the
improvement (Compare Figures 5 and 6a). This revealed
that geostatistical acoustic impedance inversion clearly
improves the vertical and lateral resolution of the rock
property distributions and provides a better way to quantify
uncertainties, producing therefore a more reliable geological
model.

Geological Framework
The reservoir, about 140 feet thick, represents the HST of a
Third Order stratigraphic sequence, and shallows upwards
from lime mud dominated sediments in the lower 50 feet, to
grainstone dominated sediments in the upper 90 feet (Fig 7).
Porosity is moderate, ca. 20%, with core permeability values
generally less than 10mD. Facies and rock properties are
comparable with those described in Kirkham et al.[1] from a
similar transition zone accumulation.

In the lower section, matrix micro-porosity is dominant and
very low permeability results. Heterogeneity is introduced
however in the form of discontinuous dolomite and
calcareous dolomite beds, up to 4 feet thick, in which poorly
developed intercrystalline and biomouldic porosity locally
provides core permeability to 20mD.

The upper section is characterized by fining upward poorly
sorted intraclastic, bioclastic grainstone beds that can be
very coarse grained and locally pebbly, and often with
irregular and abrupt bases. These beds or parasequences, <2
to 5 feet thick, and arranged in shallowing upwards
parasequence sets, Fig 7, are packaged between major field-
wide and cemented stylolite horizons that define the
reservoir sub-zonation, and facilitate a coarse scale reservoir
layering (Fig 8) that is captured in stochastic seismic
inversions.

The upper subzone is characterized by a change to rudist
and coral debris floatstone textured sediments, with
alternating fine packstone and coarser grainstone matrix,
arranged in thin lenticular biostromal units. A widespread
unit of fine to medium grained, moderately well sorted
peloidal and foraminiferal grainstones deposited in very
shallow but low energy conditions are the final sediments of
the HST.

The sequence terminates at an abrupt and irregular contact
between cream coloured, miliolid-rich sediments and
overlying dark grey, more argillaceous and orbitoliniid-rich
sediments of the dense limestone caprock. The contact is
interpreted as a marine hardground developed as a result of
major transgression of the Thamama carbonate platform.

The textural heterogeneity seen in the reservoir would
normally give rise to large core permeability differences, but
due to severe diagenetic overprint this has been strongly
suppressed. From fluid inclusion studies it is apparent that
very intense cementation by calcite had occurred prior to
substantial entrapment of oil and was extensive even within
the palaeo-crestal area, now located northeast of the present
day crest.

It proved impossible to establish a reservoir rock type
scheme based on facies or depositional texture, and only
cemented stylolitic units, facies boundaries and dolomite
beds can be differentiated from both core and log response.
These geological characteristics are utilised as the
framework for purposes of geological modeling to constrain
distribution of rock properties.

Fluid Inclusion Study
An integrated fluid inclusion (FI) and petrographic study of
60 samples from across the field is in progress. Preliminary
results are that the oil inclusions are consistently heavier (ca.
30
o
API) than current reservoir oil (ca. 40
o
API) and that the
most abundant oil inclusions are located to the northeast of
the present structural crest. This, together with enhanced
preservation of remnants of the primary intergranular pore
system, defines the area of the palaeo-crest at time of initial
hydrocarbon entrapment.
Cathodoluminescence (CL) clearly demonstrates the
evolution of pore water chemistry during burial and the
presence of 4 CL zones within the calcite cement.
Integration of the CL data and microthermometry from the
FI work shows that only zone C4 is associated with oil
inclusions.
The aqueous inclusions indicate two main phases of
cementation, but the earliest finely crystalline grain rim
cements (CL zones C1 and C2) are generally inclusion free,
but rare monophase inclusions indicate salinity close to
seawater (ca. 3.5 wt % NaCl equiv.).
Later burial medium to coarse crystals with inclusions
having relatively low salinity around 5-7 wt% (CL zone C3)
contain inclusions showing an overall increase in salinity
with temperature (from ca. 5 to ca. 20 wt%) up to 100
o
C.
The latest cement (CL zone C4), associated with oil
inclusions, has salinity close to 20 wt % and precipitated at
higher temperatures.
When work is complete, a detailed cementation and burial
history model, integrated with structural evolution, will
provide a refined diagenetic model to constrain and facilitate
the distribution of rock types.

Reservoir Rock Type scheme
Permeability is now substantially controlled by connected
micro-porosity, enhanced locally by grain dissolution and
partial dolomitisation. In the palaeo-crestal area,
disconnected remnants of intergranular porosity provide
slightly elevated permeability values in routine core
analyses, but no laterally continuous higher permeability
streaks are identified. Permeability values within the
4 J .S. Gomes, M.T. Ribeiro, P. Fouchard, B.N.Twombley, S. Negahban, S. Al-Baker [SPE 78529]
grainstones rely upon tortuous grain to grain connectivity
via grain-mouldic pores and the extensive intragranular
micro-porosity seen in impregnated thin sections.

Subtle variations of pore size distribution (PSD) and
Capillary Pressure (Pc) curves derived from the analysis of
high pressure mercury injection (MICP) data enabled a RRT
scheme to be established comprising 4 porous and 2 dense
or stylolitic types based on porosity-permeability criteria.
The initial scheme, seen in Figure 9, was based on 55
samples from two wells, and was subsequently validated by
28 additional samples from higher elevations. A further 180
MICP data analyses from ongoing SCAL work have
demonstrated the robustness of the scheme.

Lorenz Plots and Flow Units
Following reservoir rock type classification using the high-
pressure mercury injection data, the final validation of the
defined rock types has been completed using integrated
petrophysical methods to identify reservoir flow units.
Stratified modified Lorenz plots (2) have been generated to
assess the flow-capacity versus storage-capacity for each
layer and for each rock type (Fig 10).

RRT Validation and Estimation in Uncored Intervals
The use of supervised neural networks (SNN) within
GeoFrame
TM
allowed the use of the core petrophysical data
in addition to log data in order to classify in a reliable way
cored intervals for new wells.
The same procedure was used to estimate the RRTs in
uncored intervals (no permeability information) using the
available set of logs common to all the wells, and the RRTs
identified in cored intervals in the training dataset. As a
result a RRT log was created which then can be used to
complete the hard dataset for the 3D geological model.
The defined RRT scheme is used to condition stochastic
modeling of porosity, permeability and water saturation,
utilizing stochastic seismic inversion to provide trends for
inter-well extrapolations of properties.

Geological Modeling
Subsequent to stochastic seismic inversion, RRT
classification and flow unit identification, the next step was
to create a geological model integrating the results of these
studies.
The framework was built using interpreted seismic horizons
and well markers and isochores for the other reservoir
subzones. A fault model was also built, which incorporates
around 50 seismic faults.
The model cell dimension is 125 meters in X and Y and the
vertical zonation comprises 70 layers distributed
proportionally within the 15 model subzones and according
to the reservoir heterogeneity (Fig 8). Such fine layering
allowed the extraction of high-resolution models for flow
simulation purposes.
The main objective was to build different 3D subsurface
property models for the reservoir using various geostatistical
techniques, including Sequential Indicator Simulation (SIS),
Sequential Gaussian Simulation (SGS) and cloud transforms
with P-Fields, making wide use of the geostatistical seismic
inversion results to constrain the geological modeling.
AI data, a soft constraint, were not used on the non-porous
stylolitic intervals, so the simulation results for these zones
are based only on the hard data and variograms. As few
wells have been drilled, variograms are difficult to define
and so some assumptions had to be made during the
modelling process.

The modelling process was conducted according to the
following workflow (Figure 11):

Use the seismic AI inversion results as soft data and
well data as hard data.
Use the AI as a soft constraint to generate RRT models
using SIS with collocated co-kriging. This was done
using proportion curves, since a relationship exists
between AI and RRT. However, for the case of
dolomite (RRT5) it was necessary to perform a second
step where this RRT was lumped with RRT1. Using
probability maps from an indicator kriging process,
replace the modeled RRT in the first step by RRT5 but
maintaining the proportion of the RRTs for that specific
subzone.
The RRT model was populated with porosity using SGS
with collocated co-kriging with AI.
A permeability model was generated from the porosity
model constrained by AI and RRT using a cloud
transform with P-Fields method. This permeability
model was subsequently conditioned to the well
permeability data by collocated co-kriging.
Integrate results of fracture study in the permeability
model(s)
Adjust the permeability model(s) with multipliers based
on fracture intensity maps from the previous step and
well test data
Calibrate Pc curves per RRT with log-derived
saturation profiles and use them to build a water
saturation model.
Evaluate uncertainties and rank realizations using
connectivity analysis and/or streamline simulation
technology.
Upscale 3 models (low, average, high cases) for
numerical simulation.

Figures 12a-12c show for a reservoir layer and for one
single realization, the RRT, porosity and permeability
simulation results.

Mechanistic Model
Using the local grid refinement technique, an area in the
northern block of 5 by 2.5 km, including an existing
horizontal well and two proposed horizontal wells, was
extracted using cells of 50 by 50 meters, and exported to a
compositional flow simulator.

The purpose of the mechanistic flow model was to address
three main issues:

1. Honour detailed reservoir characterization to evaluate
the impact of thief zones (if any), directional
[SPE 78529] Geological Modeling of a Tight Carbonate Reservoir for Improved Reservoir Management of a Miscible 5
WAG Injection Project, Abu Dhabi, U.A.E.
permeability, stylolite development and continuity in
the overall sweep efficiency of the reservoir.
2. Evaluate the physics of the WAG process such as
trapping, relative permeability, hysteresis and water
blocking, WAG cycle/ratios, and flow instabilities (gas
fingering and gravity segregation).
3. Assess the upscaling.

Fracture Modelling
A fracture study was conducted due to the presence of a
major seismic fault in the middle of the development area.

The input data was fracture counts based on core and FMI
data, used as fracture intensity indicators and two main
drivers such as:
Geological structure (curvature and slope), layer
thickness, porosity model, overburden and faults
(distance to faults).
Seismic attributes - reflection strength, AI, maximum
curvature, azimuth, amplitude etc.

The methodology adopted makes use of the artificial
intelligence tools. It is based on the assumption that there is
a complex relationship between a large number of drivers
(geological and seismic) and the presence of fractures from
cores and FMI.

The first pass was to rank the drivers using a fuzzy neural
network that evaluates the hierarchical effect of each driver
(Fig 13). The second step was to create stochastic models of
fracture intensity using a back-propagation neural network.
Figures 14a and 14b exhibit two fracture models.

The third step was the evaluation of the results on a model
subzone base to define a fracture intensity map capable of
being combined with the upscaled geological model to
derive a permeability multiplier map.

The main findings of this study showed that the reservoir is
not a fractured reservoir and that the presence of few
fractures is highly dependent upon the distance to major
seismic faults. This is in line with the data from cores, FMI
and dynamic data.

Geomechanics
Prior to drilling the horizontal wells, the Team identified the
need to conduct a Geomechanical study to address the
following:

1. Orientation of the maximum horizontal principal stress
in the Field area (near faults and away from fault
zones). This would assist the orientation of the
horizontal wells (producers and injectors).
2. Thermal induced fracturing. This would assess if
thermal induced fractures would occur and how far they
would propagate (and direction) as a result of water
injection.
3. Stress dependency of permeability and porosity.

In order to address the above issues in a short time frame,
the Team decided to conduct this project on a phased
approach. The initial Project was split into two phases as
follows:

Phase-I: Mapping the maximum horizontal stress fields
across the field region using borehole breakout data,
dipmeter and FMI logs (mapping the orientation of
drilling induced fractures).
Phase-II: Laboratory tests on selected carbonate core
samples with different rock competencies.

Stress Field Mapping
The Stress Field mapping was completed as part of Phase-I
and the main findings can be summarized as follows:

The orientation of the maximum horizontal stress is not
constant across the Field area (Fig 15).
Wells farther from the faults show a more regional NE-
SW trend for the maximum horizontal stress.
Wells near the fault show a variable stress field
orientation.

Geomechanical Tests (Lab)
Phase-II of the project consisted of geomechanical tests on
core samples to characterize in-situ strength of key reservoir
layers that would provide basic data needed for a spectrum
of studies, including wellbore stability, fracture propagation
and reservoir simulation studies. The objective of the testing
was to construct the failure envelopes of the main reservoir
rocks encountered in the Field. Two oriented cores were
selected and the following tests were conducted:

4 Circumferential Velocity Analysis (CVA) tests to
measure compressional sound waves propagation
12 Triaxial tests to measure static elastic properties
(Youngs modulus, Poisson's ratio) and construct failure
envelopes.
6 Triaxial tests with compressional and shear wave
velocities (Vp, Vs). The results indicate that sonic
velocities did not vary significantly with confining
pressure, indicating the absence of microcraks in the
specimens.
5 Thermal Expansion tests to determine the thermal
expansion coefficient of several reservoir intervals.
Mineralogy measurements were performed to
characterize the core specimens. Thermal tests under
atmospheric pressure (temperature range of 80
o
to 170
o
F) gave very high values (in the range of 21 to 24 x
10
-6
/
o
F) indicating that thermal micro-fracturing took
place. In-situ linear expansion coefficient was estimated
in the range of 5 to 6 x 10
-6
/
o
F which are representative
of the un-fractured formation.
36 Brazilian Tests to determine tensile strength.
30 Porosity/permeability stress dependency tests.

The results of Phase-I and Phase-II studies will feed the
third phase of the project (3D Geomechanical Modelling),
which is being conducted at the moment.
6 J .S. Gomes, M.T. Ribeiro, P. Fouchard, B.N.Twombley, S. Negahban, S. Al-Baker [SPE 78529]

3D Geomechanical Model
The aim of the 3D geomechanical modeling is use data
gathered in Phases I & II of the ongoing Geomechanical
Study and dynamic production data to address the following
main objectives:

1. In-Situ Stress Refinement (e.g. finite element analysis
to construct 3D geomechanical model).
2. Fracture Extension Rate while Injecting Water.
3. Impact of joints/faults as permeability enhancements
due to pressure and/or temperature gradients.

In-Situ Stress Refinement
Starting with the estimation of minimum and maximum in-
situ stress data provided by Phase-I, the aim is to determine
the in-situ stress variation in the vicinity of the faulted areas.
A full 3D finite element model is in the process of being
constructed for this purpose. It will integrate representative
geomechanical data from laboratory measurements (Phase-
II) and fault traces from the 3D seismic. The finite element
model will be extracted from the 3D-geostatistical model
and subsequently loaded at its boundaries with a horizontal
stress averaged from measurements at individual wells
obtained from Phase I of the study. The geomechanical
simulator will calculate the stress equilibrium in the
geological model. Resultant variations in stress are the
results of slips on faults, folding of the structure and
heterogeneity of mechanical properties.

The final aim is an estimation of in-situ stress in the vicinity
of the faulted areas within the field development area.

Fracture Extension Rate while Injecting
A study will be performed for fracture growth during cold
water injection. The major processes that must be
considered for this study are: porous flow, heat convection
and conduction, changes in in-situ stresses due to the
presence of thermal stresses, and fracture growth due to the
changing stress field. The study should involve three major
components as follows:

1. The first component consisting of performing a
standard reservoir injection/production pattern study
using a flow simulator to model pattern performance for
a simulation containing a hydraulic fracture where the
fracture is allowed to grow with time.

2. The second aspect of the study should involve
calculating a temperature distribution within the
reservoir as a function of time. The temperature
distribution will be affected by the injection rate and by
the presence of a growing hydraulic fracture. The
temperature profile can be calculated analytically using
a simplified energy conservation calculation that
assumes a constant temperature elliptical region around
the injection well at any given time and can also be
calculated using the simulator that can calculate
temperature profiles.

3. The analytical approach probably the preferred
approach, can be used for the majority of the study if
the results are consistent with the more rigorous
approach. The third aspect of the study involves using
the approach for calculating fracture will be the
analytical solutions. This aspect of the study will
provide the average stress changes due to thermal
stresses and will also provide the fracture profile as a
function of time.

All three components of the study should be physically
connected and a proper solution will require iterating
between the three components to obtain a converged result.

Impact of joints/faults as permeability enhancements
due to pressure and/or temperature gradients.
Although hydraulic fractures may be created around the
injectors (enhancing the permeability), it is the joints and/or
faults that may induce both matrix and fault conductivity
enhancements in the short and long term. For this reason the
study will address the impact of joints and/or faults in
permeability enhancements due to pressure and/or
temperature gradients.

Risk Assessment & Management Strategy
To arrive at best decisions with regard to business,
management and technical issues associated with Phase I
development plan, the team has embarked on a
comprehensive Dynamic Risk Assessment & Management
Strategy (DRAMS). Similar strategy is being applied to a
nearby field (3). DRAMS will make the Phase I project
successful by reducing risk associated with delivering the
required 20,000 BOPD production by end 2005, maintaining
the plateau for a specified period of time, and achieving a
high recovery efficiency. In addition, the information
obtained from DRAM during Phase I development will
provide very valuable information to risk mitigate the
subsequent phase of development beyond 2015. The
proposed DRAMS will achieve the Phase I project
objectives by focusing on important issues, considering long
time horizons, accounting for uncertainties, and effectively
resolving conflicts by providing contingency plans.

Conclusions
The geological modeling work presented in this paper
covered seismic constrained porosity and permeability
modeling, both with and without rock typing. Various
geostatistical techniques were used, including sequential
Gaussian simulation (sGs) for the porosity, sequential
indicator simulation (SIS) for the rock types and cloud
transform with P-Fields for the permeability, with emphasis
on the use of seismic data to constrain the modeling process.
Stochastic seismic inversion, when conditioned to the
reservoir framework, played a major role in the modeling
process. Detailed rock and flow unit characterization were
the key elements for the petrophysical modeling. The
judicious integration of cores and image data were vital for
the assessment of the fracture intensity. Geomechanical
studies coupled with structural analyses were extremely
useful for the assessment of the horizontal well orientations.

[SPE 78529] Geological Modeling of a Tight Carbonate Reservoir for Improved Reservoir Management of a Miscible 7
WAG Injection Project, Abu Dhabi, U.A.E.
Acknowledgements
The authors wish to thank the management of Abu Dhabi
National Oil Company (ADNOC) and Abu Dhabi Oil
Company for Onshore Oil Operation (ADCO) for their
permission to publish this work.
The authors also wish to acknowledge the support of their
colleagues in ADCO. The contributions of specialist
consultants Norman Oxtoby and Dr J Neilson are
acknowledged for FI and CL work respectively. We also
thank the contributions provided by Zissis Moschovidis and
J ohn Cameron from PCM Technical Inc and M. Akbar from
GeoQuest for the geomechanical studies and stress field
mapping respectively. Acknowledgments are extended to
Alin Huerlimann from J ason for the stochastic seismic
inversion work and to Ian Golliferand Abdel Zellou from
RC2-Veritas Group for their support in geological/fracture
modeling.


References
1. Kirkham, A., J uma, M.Bin, McKean, T.A.M., Palmer,
A.F., Smith, M.J ., Thomas, A.H., and Twombley, B.N.
(1996) - Fluid Saturation Predictions in a Transition
Zone Carbonate Reservoir, Abu Dhabi, GeoArabia,
Vol. 1, No. 4.
2. Gunter, G.W. et al. (1997) - Early Determination of
Reservoir Flow Units Using an Integrated Petrophysical
Method, SPE 38679, pages 373-380.
3. Ayoub, M., Gomes, J .S., Negabhan, S., Al-Baker, S.
(2002) - "Identification, Quantification and Analysis of
Subsurface Uncertainties associated with Reservoir
Description. Assessment of its Impact on the
Development of a Miscible WAG Injection Project with
Horizontal Wells", Abu Dhabi, Poster presentation at
ADIPEC 2002.







8 J .S. Gomes, M.T. Ribeiro, P. Fouchard, B.N.Twombley, S. Negahban, S. Al-Baker [SPE 78529]








Fig-1 Fig-2
Fig-3 Fig-4
Fig-5
Fig-6
Fig-6a
[SPE 78529] Geological Modeling of a Tight Carbonate Reservoir for Improved Reservoir Management of a Miscible 9
WAG Injection Project, Abu Dhabi, U.A.E.




2
grainstone
para-
sequence
sets
Rudist
biostromes
dolomite
dolomite
1
3A
4A
3B
5
4B
6
Lime mud
dominated
sediments
Cemented subzone
Stylolite cementation
CAPROCK
Schematic reservoir sequence with
shallowing-up parasequence sets of
coarse to fine grainstones in subzone 3.
Dolomite beds confined to the sub-tidal
lime mudstone to packstone dominated
lower subzones.



Reservoir
Reservoir
RESERVOIR MODELLING
A
I
S
u
b
MSZ
2
3
3
4
Top Sei
Base Sei
Layers
1
1
1
6
Fig-7 Fig-8
10 J .S. Gomes, M.T. Ribeiro, P. Fouchard, B.N.Twombley, S. Negahban, S. Al-Baker [SPE 78529]



A
Stratigraphic Modified Lorenz Plot - B-1
Core Data
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
0.00 0.20 0.40 0.60 0.80 1.00
Storage Capacity (phi*h)
F
l
o
w

C
a
p
a
c
i
t
y

(
k
*
h
)
SubZone1 2 Dense 3A Dense
3B Dense 4A Dense 4B
Dense 5 6
Stratigraphic Modified Lorenz Plot - B-1
Core Data
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
0.00 0.20 0.40 0.60 0.80 1.00
Storage Capacity (phi*h)
F
l
o
w

C
a
p
a
c
i
t
y

(
k
*
h
)
RRT 1 RRT 2 RRT 3 RRT 4 RRT 5 RRT 6
B
Inflexion
Fig-9
Fig-10
Fig-11

[SPE 78529] Geological Modeling of a Tight Carbonate Reservoir for Improved Reservoir Management of a Miscible 11
WAG Injection Project, Abu Dhabi, U.A.E.










Fig-12b
Fig-12a
Fig-12c
Fig-13
Fig-14a
Fig-14b