Outcrop and Subsurface Characterization
of Carbonate
Reservoirs for Improved Recovery of Remaining
Hydrocarbons
Executive
Summary
The
Reservoir Characterization Research Laboratory (RCRL) for carbonate
studies is an industrial research consortium run by the Bureau of
Economic Geology (BEG), John A. and Katherine G. Jackson School of
Geosciences, The University of Texas at Austin (UT). RCRL’s
mission is to use outcrop and subsurface geologic and petrophysical
data from carbonate reservoir strata as the basis for developing new
and integrated methodologies to better understand and describe the
3-D reservoir environment.
The
RCRL program has run continuously since 1987 and has produced more
than 40 external publications, as well as BEG publications, on carbonate
reservoir characterization, sequence stratigraphy, petrophysics, geostatistics,
and petroleum engineering. We provide research results to our sponsor
companies through annual review meetings, CD’s, preprints of
publications, short courses on geologic and engineering aspects of
our research, and a mentoring program in which we work hands-on with
industry staff using their datasets. In addition, our results are
posted in a password-protected part of our Website (www.beg.utexas.edu/indassoc/rcrl/index.htm).
RCRL
has maintained a membership of between 13 and 18 companies per year
(see list of 2003 sponsors at end of proposal). The sponsorship currently
has strong interests in the Permian Basin as well as Middle Eastern
carbonate reservoirs. This enrollment, supplemented by government
grants, supports between three and six professional staff members
and varying numbers of graduate student research associates, plus
strong computer, editing, and graphics services.
Three
geologists, a geological engineer, two reservoir engineer/modelers,
and a geophysicist/interpreter form our core group:
Dr.
Charles Kerans, Geologist Dr. Xavier Janson, Geologist Mr. Jerome A. Bellian, Geologist Mr. F. Jerry Lucia, Geological Engineer Dr. James W. Jennings, Jr., Reservoir Engineer/Modeler Dr. Fred Wang, Petroleum Engineer Dr. Hongliu Zeng, Geophysicist
All
staff members have extensive industry experience or have worked closely
with industry and are well aware of the challenges and questions facing
development geoscientists and engineers. We are also very proud of
our graduate student staff, which has included several award-winning
students, many of whom are now working in the industry.
Each
year we combine industry input with our own ongoing research plans
to develop a set of key geological and engineering research topics.
Plans for 2004 are focused on the following areas:
•
Mapping,
characterizing, and modeling touching-vug pore systems.
•
3D
modeling of geologic facies and petrophysical rock-fabric elements
using outcrop and subsurface data
•
CT
imaging for characterizing heterogeneity at the core-plug scale
•
Limestone
recrystallization and its effects on petrophysical properties
•
Expansion
and integration of the rock-fabric and core databases
•
Continuation
of field-based studies of carbonate slope systems
•
Initiation
of regional sequence analysis of the Arab-D
If
you have any questions on any aspects of the RCRL Carbonate Reservoirs
Research Program, please contact Charlie Kerans (512-471-1368 or charles.kerans@beg.utexas.edu)
or Jerry Lucia (512-471-7367 or
).
Funding
With this proposal, we invite you to participate in the continuation
of the RCRL Carbonate Reservoirs Research Program. A list of sponsors
during 2003 can be found at the end of this proposal. In 2004 the
annual RCRL Industrial Associates contribution to the program will
continue to be $45,000 per year. Our industrial sponsors will continue
to receive research results at annual review meetings, in short courses,
during mentoring activities, in publications and CD’s, and on
our Website.
Research
Directions for 2004
Mapping,
Characterizing, and Modeling Touching-Vug Pore Systems
One
of the single most pressing issues in the characterization and modeling
of carbonate reservoir strata that remains to be effectively addressed
is the geologic and petrophysical characterization of touching-vug
pore systems, and the integration of these data into fluid-flow models.
We have outlined two areas of research where touching vugs of depositional
and diagenetic origin are seen to be significant and are planning
to model these datasets as a contribution to this important area of
research.
Multi-Scale
Characterization and Flow Modeling at Pipe Creek
Outcrop exposures of Cretaceous rudist mounds and their associated
deposits near the town of Pipe Creek in Central Texas exhibit a
variety of heterogeneities throughout a range of scales including
(1) decimeter-scale separate-vug pores within mound-core caprinid
fossils standing upright in porous sediment, (2) decimeter-scale
touching-vug pores within and between caprinid fragments in steeply
dipping flank beds, (3) an unknown scale and degree of connectivity
between these pore systems, (4) a heterogeneous arrangement of mounds,
debris deposits, and intermound sediment capped by discontinuous
beaches at lateral scales of tens of meters and larger, but too
small to be mapped between wells, and (5) the organization of these
features in kilometer-scale shore-parallel belts stacked in depositional
cycles as much as 10 m thick.
Understanding and modeling spatial variability and fluid flow in
systems such as this presents numerous challenges. Perhaps the most
important of these challenges are those having to do with flow in
the touching-vug pore network at multiple scales. Advances in this
area will of course have significant applications to analogous reservoirs
in the Middle East, but the methodologies developed to address this
problem will probably have even more important applications to other
touching-vug systems as well.
In past years we have studied touching-vug pores in a roughly 25
cm ? 25 cm ? 35 cm caprinid debris sample by using CT scanning (in
collaboration with Richard Ketcham and coworkers at the UT High-Resolution
X-ray CT Scanning Facility), numerical flow modeling (in collaboration
with Todd Arbogast and coworkers at the UT Institute for Computational
and Applied Mathematics), and physical experiments (in collaboration
with Steve Bryant and coworkers at the UT Department of Petroleum
and Geosystems Engineering). The results of this study to date suggest
that there is a connected pathway spanning a large fraction of the
sample that can sustain very large flow rates equivalent to permeabilities
on the order of 100 Darcys or more. However, the behavior of the
sample is highly dependent on the boundary conditions, and experiments
on larger samples are needed to adequately characterize the flow
patterns.
In 2003 we constructed a stochastic simulation of mound cores, debris,
inter-mound sediment, and beach deposits each having realistic shapes
in a 1-km-scale fluid-flow model using data from the outcrop and
several cored wells. In 2004 we may be able to refine the geometries
in that model by using ground-penetrating (GPR) data and interpretations.
However, the most important factor in fluid-flow prediction seems
to be the unknown magnitude, scale, and orientation of the high-permeability
connections in the debris deposit.
Therefore, in 2004 we hope to conduct a field experiment flowing
air with gas-phase tracers between newly drilled shallow wells to
gain some insight on effective permeability at the 1- to 10-m scale.
We plan to continue this approach by studying large pore systems
in other large outcrop samples and improving our methods of visualization,
numerical modeling, and laboratory experimentation to obtain a better
understanding of the magnitude, scale, and orientation of the high-capacity
flow paths.
Geology,
Petrophysics, and Modeling of Touching-Vug Pore Systems
We have made
considerable progress toward converting wireline logs into matrix
permeability profiles and distributing these data within a sequence
stratigraphic model using geostatistical methods. A key aspect in
this conversion is linking rock fabrics to petrophysical properties
through petrophysical class and rock-fabric number. In the past we
have presented methods for determining the petrophysical class and
rock-fabric number from porosity, acoustic, and resistivity logs.
In 2003 we demonstrated how the rock-fabric method could be used to
construct permeability models when only porosity logs are available
by integrating rock-fabric numbers with stratigraphic information.
Characterizing and estimating permeability of touching-vug pore systems
remains a critical problem that is unresolved. Touching-vug systems
are formed by a complex combination of processes including dissolution,
fracturing, and cavern collapse, as well as regional tectonism, which
in part accounts for the difficulty in constructing a predictive model
of this important pore type. We have gathered a considerable volume
of information on touching-vug pore systems and have concluded that
vug interconnection is the most important character to model and that
the fracture aperture and surface roughness is less important. Vug
interconnections are best studied at the outcrop scale because the
important scale of interconnection is typically larger than core scale.
Equally important is the problem of constructing flow simulations
of touching-vug pore systems with or without matrix permeability.
We already have experience modeling coupled flow between a permeable
matrix and thin fractures in standard finite difference simulators
using both explicit fracture representation and nonneighbor connections.
Flow in larger vugs can be simulated with coupled Stokes flow for
the vugs and Darcy flow for the matrix using specialized code developed
by Todd Arbogast and coworkers at the UT Institute for Computational
and Applied Mathematics. Complicated systems having fractures and
vugs at several scales may require a combined approach.
In 2004 we propose to construct a touching-vug model of a Cretaceous
road cut located in Kerr County, Texas, and use this model in flow
simulation studies. An initial description of this road cut has been
made by Susan Hovorka and others as part of a study of the Edward
aquifer, and that study will be used to place the road cut in a stratigraphic
and structural context. The road cut contains a variety of touching-vug
types, including fracture porosity, solution-enlarged fracture porosity,
cavernous porosity, a small cave, and numerous dissolved anhydrite
nodules. It is located just north of the Balcones Fault Zone. We intend
to (1) map the touching-vugs on both walls of the road cut using photo
pans and ILRIS imaging, (2) extend fractures and caves from one side
of the road cut to the other where possible, and (3) distribute other
features stochastically within the 3D volume. Although modeling matrix
petrophysical properties will not be a prime consideration, we have
core plug measurements that indicate a matrix having an average porosity
of 12 percent and permeability of about 5 md. After this static model
is constructed we plan to (1) study methods of flow-model construction
for various sizes of touching vugs and (2) use the resulting models
to study the effect of fracture connectivity on fluid flow.
3D
Modeling of Carbonate Reservoir Analogs
The second
area we see as a major barrier to building realistic 3D models of
carbonate reservoirs is in populating our models with petrophysically
significant, geologically reasonable facies distributions. We believe
that our workflow and methods for constructing the basic chronostratigraphic
framework have advanced greatly and are reliable. However, “filling
the space” with facies distributions that represent realistic
3D depositional systems is still a major problem. Toward improving
this process, we continue to build the new generation of 3D laser-based
outcrop analog models to better understand the process of realistic
facies modeling. The Dry Canyon dataset will be our major thrust,
but we will attempt to pull into the 3D world two previously generated
datasets from Lawyer Canyon and Painted Canyon.
Dry
Canyon 3D Model
We are planning
in 2004 to complete a 3D stratigraphic model of the Dry Canyon outcrop.
This effort will entail (1) completing the field mapping, (2) acquiring
lidar for the main Beeman Canyon and associated side canyons, (3)
generating detailed descriptions of the three EPR cores and interpreting
the associated well-log suites, and (4) collecting additional sample
profiles in Yucca Canyon and Beeman Canyon to provide fabric/property
data for population of properties within the 3D model. We will refine
our initial sequence stratigraphic framework by incorporating the
stratigraphic architecture exposed in Beeman Canyon. By integrating
the stratigraphy exposed in these multiple canyons in a single 3D
geologic model we hope to answer the following questions:
•
What
is the exact timing and spatial distribution of mound growth
within the sequence stratigraphic framework?
•
Is
mound size and position on the shelf predictable within a sequence
stratigraphic framework?
•
What
are the time-stratigraphic and spatial relationships between
carbonate and siliciclastic facies?
•
What are the controls on their position in time and space, extent,
and relative importance? Antecedent topography? Position within
the large scale sequences? Higher frequency accommodation changes?
Our initial
results suggest that buildup distribution is controlled by paleotopography
and changes in accommodation. These parameters also control where
and when siliciclastic and carbonate sediment are accumulating on
the shelf. We will test these observations against the stratigraphy
exposed in Beeman Canyon. In addition, we are planning on using
GPR to fill the gap between exposures where stratigraphic relationships
between sedimentary features such as mound and intermound areas
are poorly exposed.
Building a 3D model of the Dry Canyon area will improve our workflow
and methodology for populating facies and rock properties within
the 3D stratigraphic framework. We will build on knowledge gained
from the Victorio Canyon example and apply new algorithms, particularly
for modeling discrete buildup patterns. We expect to use a combination
of stochastic and deterministic methods to properly model the distribution
of small phylloid-algal and leopard-rock mounds. A range of deterministic
and object modeling methods will be used to recreate the siliciclastic
channels.
To build a realistic reservoir model, we will collect data from
subsurface analogs where mechanical and petrophysical properties
can be extrapolated. We have been actively studying a series of
Canyon-Cisco-age reservoirs along the Horseshoe Atoll trend that
are partly time-equivalent and have numerous facies that compare
favorably with the outcrop facies. In addition to building the 3D
model, we will calculate 3D synthetic seismograms and compare the
seismic signature of the Dry Canyon model to these subsurface analogs.
3D Digital
Documentation of Classic Outcrops
Addition of the ILRIS 3D and associated tools to the RCRL workflow
has opened up capabilities that were previously not available for
characterizing classic carbonate outcrop analogs. A workflow of
capturing outcrops using lidar, Global Positioning System, and limited
GPR, coupled with standard sequence stratigraphic and petrophysical
approaches will permit us to do a better job of creating realistic
analog models for facies modeling and reservoir simulation experiments.
Two areas previously studied that will benefit greatly from this
3D model construction approach are Lawyer Canyon and the Pecos River
(Painted Canyon). Both these areas have detailed 2D geological models,
have varying amounts of site-specific petrophysical data, and have
undergone some level of reservoir simulation experimentation using
2D windows. Lawyer Canyon still serves as the reference section
for understanding detailed flow unit layering and simulation in
shallow-water carbonate reservoirs. Petrophysical data collected
for Lawyer Canyon comes from a series of 3D canyon walls and then
is collapsed onto a single 2D plane. Placing data in a more realistic
3D setting will most likely improve our understanding of flow by
moving this 3D problem (reservoir heterogeneity and sweep efficiency)
into a 3D context.
The prograding grainstones of Painted Canyon on the Pecos River
provide a superb example of Cretaceous prograding highstand grainstones.
The complex architecture of caprinid debris rudstones and associated
grainstones is well displayed in the 90-degree arc of the main Pecos
Canyon, as well as in the side canyon (Painted Canyon). These exposures
combined illustrate differences in development of rudist reservoirs
in environments from beach-foreshore to lower shoreface. As yet,
this outcrop is represented only by a simplified 2D panel. Greater
control from lidar will allow a full 3D model to be constructed
that will aid in better understanding of this important facies association.
Products from these studies will be full Gocad models that will
allow examination of these classic outcrop analog datasets in a
range of workstation and VR settings.
Additional
Areas of Research
Quantification
of Internal Core-Plug Variability with CT Scanning
Understanding the flow properties of core samples has always been
hampered by inadequate characterization of their internal heterogeneities.
Thin sections provide critical insight into the internal structure
of a sample, particularly at the pore scale, but the information is
often unrepresentative because of the small volume of a thin section
relative to that of the sample as a whole. High-resolution X-ray CT
scanning may bridge the gap. Although the resolution of these instruments
does not yet equal that of thin sections, they can quantify bulk density
on a tightly spaced 3D grid filling the entire sample volume. The
cost of scanning a sample is small enough to consider making it a
regular part of routine plug analysis, and large quantities of these
data may become available in the coming years. Methods to analyze,
summarize, and relate the data to other petrophysical measurements
will be desirable.
We have CT scans of 50 plugs samples from the South Wasson Clear Fork
reservoir and another 30 from the Fullerton Clear Fork reservoir.
In addition, we have routine porosity, permeability, and grain density
measurements for all these samples, thin sections for most of them,
and various special core analyses for many of them. In 2002 we used
detailed flow modeling for one of these samples to study the effect
of patchy anhydrite on permeability. In 2004 we plan to use this collection
of CT scan data to develop methods for routine quantification of internal
plug variability, and to relate that variability to porosity, permeability,
rock fabric, and if possible, other special core analysis parameters.
Porosity should be related directly to the average bulk density measured
by each CT scan, whereas rock-fabric, permeability, and other flow-related
properties may be related to the variances, correlation lengths, and
correlation anisotropies of the internal heterogeneities.
Limestone
Recrystallization: Effects on Petrophysical Properties
Petrographic/petrophysical studies of ooid grain-dominated fabrics
from the Sacroc and Cogdell fields of Pennsylvanian age and the
Fullerton Clear Fork field of Permian age have shown relationships
that are very different from predictions based on previous work.
Oomoldic grainstones typically are characterized by low permeability
and poor capillary properties compared with ooid grainstones, Archie
m values in the range of 2.5 to 4 compared with an average value
of 2 for ooid grainstones, and total porosity values much larger
than porosity derived from acoustic logs. The classic explanation
for these characteristics is that (1) oomoldic porosity contributes
much less to permeability than intergrain porosity, (2) pore space
is composed of large oomolds connected by very small pores found
in the matrix and ooid rings, and (3) electric and sonic waves are
controlled more by the small matrix pores than by the large oomolds
in an oomoldic grainstone.
Petrographic studies completed in 2003 revealed that some oomoldic
grain-dominated limestones have petrophysical characteristics more
like intergrain porosity than moldic pore space. There appears to
be complete spectrum from grain molds to recrystallized ooids with
pore space between calcite spar of varying sizes, and only grainstones
with pure grain molds appear to have characteristics typical of
ooid grainstones. However, most of the “ooid grainstones”
in these three fields have a minimal volume of grain molds and are
best described as recrystallized grainstones, and the petrophysical
properties are related to the size of the calcite spare rather than
the size of the ooid grains.
In 2004 we proposed to further investigate the processes by which
an ooid grain-dominated sediment is recrystallized and how the resulting
changes in pore structure affect the petrophysical properties. Key
questions include 1) What is the porosity history of an ooid? (2)
How are the changes related to original mineralogy and diagenetic
environment? and finally (3) Can the diagenetic changes be linked
to depositional environment, paleotopography, and sequence of events?
Most of the reservoir data will come from the three fields mentioned
above. We will gather fabric and porosity data from ooids of different
ages starting with modern ooids to investigate the history of diagenetic
change.
Slope
Carbonates
Carbonate slope and toe-of-slope depositional settings have received
far less investigation compared with extensively studied shallow-water
carbonate deposits. Our understanding of the processes and mechanisms
involved in carbonate slope sedimentation, as well as the resulting
sedimentary geometries and stratigraphic architecture, is limited
compared with similar deposits in siliciclastic settings. Yet there
is a significant opportunity to leverage learnings from the past
10 years of extensive research in deep-water clastic systems to
better understand carbonate systems.
The RCRL just finished an integrated study of the classic Albian
slope/toe-of-slope Poza Rica field in Mexico. In addition, last
year we completed a study of Permian toe-of-slope to basin redeposited
carbonates in Victorio Canyon. Though many of the pore types are
similar to those associated with the shallow-water, many other aspects
differ significantly. We have begun to develop an initial conceptual
model for the stratigraphic architecture of these redeposited sediments
in the lower slope, toe-of-slope, and basinal depositional environment.
We see a critical need for comprehensive investigations that will
provide the fundamental depositional mechanisms and the time/space
architecture of redeposited carbonate sediments in deep-water settings.
Though some degree of prediction has been gained through better
understanding of the sequence framework of the slope systems, a
great deal is yet to be understood with regard to lateral continuity
of turbidites and debris flows and what controls this continuity.
We are searching for additional outcrops that expose platform to
basin deposits in order to better understand the sequence controls
on depositional styles, and to start to quantify flow-unit continuity
along both strike and dip. Last year we visited the spectacular
Gorbea platform outcrop in northwestern Spain that shows good exposures
of Lower Cretaceous (Albian) platform-margin clinoforms and toe-of-slope
megabreccia. In 2004, we plan to investigate other potential candidates
for a 3D outcrop-based reservoir modeling. Possible candidates include
the northern part of the Maiella platform in Italy, the slope of
the Triassic isolated platforms in the Dolomites, Cretaceous outcrops
in the Spanish Pyrenees and in the Vercor of Southeastern France,
and the Jurassic outcrop in Oman. We are particularly interested
in potentially interesting slope/toe-of-slope outcrop suggestions
from sponsor companies.
Rock Fabric and
Core Descriptions Database
In 2003 we initiated work on an Access database to include core
descriptions from the numerous fields the RCRL has studied by compiling
core descriptions from Shuaiba fields in Oman. In addition, the
rock-fabric database that was presented in 2002 was improved. In
2004 we plan to continue this effort by (1) adding more core and
rock-fabric descriptions and (2) combining the core and rock-fabric
databases where feasible.
Mentoring
and Projects within the RCRL Umbrella
Direct contact
with the technical staff of our sponsoring companies is an important
ongoing aspect of our program. This interaction allows us to test
our concepts and methods on real problems while assisting sponsors
in developing new reserves. Sponsors are encouraged to contact us
with projects that could be mutually beneficial.
In 2003 we worked with Oxy Permian constructing a geologic and porosity
model of the Cogdell (Pennsylvanian) field in West Texas. We plan
to continue this work into 2003 integrating matrix and touching-vug
permeability into the model. Matrix permeability will be based on
rock-fabric/porosity transforms, and touching-vug permeability will
be based on production and injection profiles as well as other production
data.
In 2004 we will initiate
a feasibility study for carrying out a core-based study of the Arab
Formation throughout Saudi Arabia, Qatar, and the U.A.E. The Arab
Formation is the world’s most prolific hydrocarbon system.
It is well developed in several countries and states throughout
the Middle East, including Saudi Arabia, Qatar, and the U.A.E. This
study would last for the next 3 years and would focus on developing
an understanding of the larger scale basin patterns and sequence
hierarchy of the Arab. Countries contributing data will get a strong
say in where the study will be focused and the number of cores examined
in particular regions. Other RCRL members will receive digital core
descriptions and log suites for selected wells, with full documentation
of key facies and sequence/cycle analysis. It is planned that over
the course of the next 3 years a sufficient database will be collected
to add significantly to our understanding of the regional stratigraphic
architecture of the Arab. Recruiting QP and ADCO to the program
and engaging the temporary assistance of a biostratigrapher will
be important to the success of this project.
List
of 2003 Sponsors
Anadarko
Aramco
BP
Chevron/Texaco
Conoco/Phillips
ExxonMobil
Great Western Drilling
Kinder Morgan
Landmark
Marathon
Occidental Petroleum
Petroleum Development Oman
Shell International
Statoil
TOTALFinaElf
