Synthesis and Compilation of Carbonate Reservoir Architecture
and Petrophysical Quantification
The RCRL has
been in the process of describing, interpreting, and modeling
carbonate strata and petrophysical properties at the reservoir
scale and to a lesser extent at the exploration scale for 16 years.
During this time data have been collected that include everything
from basic outcrop measured sections, core descriptions, and petrophysical
data to complex 3D models, global transforms for porosity-permeability
models, and reservoir flow simulations. We have begun the process
of compiling and making some of these data available to sponsors
in the form of databases with tabular and image data. This process
has been hindered by the lack of a meaningful hierarchical organization
and format with which to display the enormous amount of data we
have collected over the past years with your support.
In 2005 we
plan to change direction toward synthesizing all available data
that we have into key topical areas and supplement these data
with selected information from the literature. We envision that
this will be a multiyear project starting in 2005 with a completion
target of 2007. Progress will be reported at each yearly review
meeting, and interim access to the results will be provided to
the sponsors at that time. We anticipate that the data will reside
on a server here at BEG with access password protected. In order
to protect our current and future sponsors, we have decided that
access to this database/synthesis product will be on a pay-as-you-go
basis starting with the 2005 sponsorship fee. There will be a
chargeback fee for those companies that join the RCRL in future
years or that skip a year.
have not yet been decided upon but will probably be based on geologic
age because that is the simplest approach. Special topics such
as facies dimensions carbonate petrophysics, etc., will be covered.
The synthesis will include search engines to find specific items
such as outcrops, reservoirs, lithology, and rock fabric, among
other parameters. Each outcrop and reservoir will have a brief
description of the geologic context, the basic data, and the interpretations
presented. The basic data will include core descriptions, outcrop
sections, core photos, petrophysical data, thin-section descriptions
and photos, typical wireline logs, and capillary pressure. Interpreted
data will include geologic maps and cross sections, 3D models,
and reservoir models with brief write-ups of the interpretation
methods used. This initial effort may evolve in several possible
directions on the basis of sponsors’ input and our experience.
and Petrophysical Characterization of Vuggy Porosity
and petrophysical characterization of vuggy porosity as related
to fluid flow and hydrocarbon recovery remains a critical problem
in the construction of carbonate reservoir models for reservoir
performance prediction. We have made considerable progress toward
converting wireline logs into matrix permeability profiles and
distributing these data within a sequence-stratigraphic model.
Little progress, however, has been made toward quantifying the
effect of vuggy porosity on recovery. Much of our understanding
is confined to vugs that are connected only through the interparticle
pore system called “separate vugs.” However, predicting
the distribution of separate vugs remains a major concern. Vugs
that form their own connected pore system, called “touching
vugs,” are poorly understood. Touching-vug systems are formed
by a complex combination of processes including dissolution, fracturing,
cavern collapse, and regional tectonism. Touching-vug and interparticle
porosity together form a bimodal pore system that is poorly understood
and difficult to simulate.
Characterizing, and Modeling of Bimodal, Touching-Vug Pore Systems
pore systems are commonly composed of both matrix and vuggy porosity.
Whereas there are several adequate methods for characterizing
matrix properties petrophysically using information gained at
the core scale, touching vugs are normally composed of large vugs
that cannot be adequately studied at the core scale. The scale
at which large vugs are connected is a key issue for understanding
the flow properties of touching vugs. Equally important is the
method by which the vugs are connected. Over the past few years,
the RCRL has developed a research program designed to investigate
the geology and flow properties of bimodal touching-vug pore systems.
This program has been focused on studying touching vugs at various
scales using material from carbonate outcrops and production data
from carbonate reservoirs.
The 1-ft scale
has been studied using a large outcrop sample of rudist rudstone
containing large rudist vugs and inter-rudist pore space. Todd
Arbogast and coworkers at the UT Institute for Computational and
Applied Mathematics have developed code for coupling Stokes flow
for the vugs and Darcy flow for the matrix using a high-resolution
CT scan of the outcrop sample. Efforts to characterize interconnections
and model flow in this sample continue using a pore-space model
based on a high-resolution CT scan. This project is funded by
the RCRL and a National Science Foundation grant with Dr. Arbogast
as principal investigator and will continue in 2005. As part of
this project we plan to investigate flow at the 10-ft scale by
drilling wells in the outcrop from which the large sample was
taken. Two wells will be drilled 10 ft apart, air injected in
one well, and pressure and flow rate measured in the other. This
procedure should provide us with a measure of effective permeability
at the 10-ft scale.
For the 100-ft
scale, we have initiated a study of a karsted Cretaceous road
cut that contain faults, fractures, solution-enlarged fractures,
and caves. The objective of this activity is to build an explicit
dual-porosity model showing the distribution of porosity in the
fractures, dissolution porosity, and matrix porosity. This model
will be used to experiment with flow properties at various scales.
In 2004 a relationship between tectonic fractures, stratigraphy,
matrix permeability, and dissolution at various scales was developed.
In 2005 we hope to complete the construction of a 3D model showing
the distribution of matrix and vuggy porosity suitable for flow
Mountain Karst Study
paleocaves (karst systems) are a significant player in the development
of touching-vug pore systems. During the last two decades, studies
from two divergent areas, oil exploration and development and
cave hydrology and geochemistry, have improved our understanding
of the processes of cave development and of paleocave systems.
However, our ability to develop robust geologic models and to
generate efficient and reliable resource predictions in these
systems remains inadequate.
In 2005 we
propose to focus on the characteristics and origins of paleocave
complexes and associated pore systems found in the mega karst
system exposed in the Ordovician outcrops, Franklin Mountains,
El Paso, Texas. The research will document 3D geometric patterns
of paleocave systems, identifying elements and processes that
control the patterns and developing geochemical and geomechanical
models for their origin. An important aspect of this research
will be to develop methods for forward modeling of the diagenetic
processes associated with cave formation.
At the reservoir
scale, in 2004 we analyzed injection data from several wells and
obtained permeability estimates from touching-vug intervals in
the Cogdell Pennsylvanian field in the Permian Basin. In 2005
we plan to initiate a reservoir study of the Hobbs San Andres.
Production data and core descriptions from the Hobbs San Andres
(Permian) field strongly indicate the presence of karsting and
touching-vug porosity. Numerous cores are available, and about
half the 600 wells have good suites of wireline logs. The field
is actively being converted to a CO2 flood, and modern production
data are also available. We are confident that a detailed reservoir
study of this reservoir will provide important insights into modeling
and predicting performance of touching-vug pore systems.
Modeling of Separate-Vug Pore Space
In 2005 we
plan to produce a 3D synthetic seismogram for the Lawyer Canyon
model focused on imaging the moldic porosity that has been mapped
in high-frequency cycle 7. In 2004 we showed that the velocity
changes induced by vuggy grainstone compared with non-vuggy grainstone
can generate a large enough impedance contrast to be detected
on seismic images. In 2005 we plan to study the petromechanical
characteristics of this vuggy grainstone by collecting numerous
samples and measuring their mechanical, velocity, and petrophysical
properties. The initial task will be to identify unique petromechanical
characteristics of the vuggy grainstone that will allow us to
unequivocally differentiate them from the nonvuggy grainstone.
Our initial review of the petrophysical literature shows that
vuggy grainstone might have a slightly different Vp/Vs ratio.
We will try to investigate further petrophysical properties such
as bulk and shear modulus or Poisson ratio. Eventually, we will
use these mechanical characteristics to build an elastic impedance
model that will be used to generate initially a 1D synthetic seismogram
at various offset angles using a full elastic wave equation method.
Ultimately, the AVO response of this vuggy grainstone will be
Modeling of Carbonate Reservoir Analogs
The RCRL continues
to be committed to extracting critical information on carbonate
reservoir heterogeneity from key outcrops in order to understand
scale and variability in three dimensions, most importantly in
the lateral, interwell dimension. In the past we have focused
on developing sequence-stratigraphic methods and detailed lateral
petrophysical variability, mostly in two dimensions. With the
advent of high-resolution ground-based lidar imaging technology
we are able to accurately image outcrop data in three dimensions
and provide 3D spatial statistics of various facies and geobodies.
We have used this type of information in our study of the Poza
Rica deep-water carbonate field in Mexico to construct a realistic
geologic model and are confident that collecting 3D spatial statistics
will provide our sponsors with tools to better construct geologic
In 2004 we
collected lidar data in Lawyer Canyon (Permian) and Painted Canyon
(Cretaceous), areas we have previously studied. In addition we
initiated work on the Virgil (Penn.) algal mounds in the Sacramento
Mountains, New Mexico, and the famous Windjana Gorge (Devonian)
in Australia. In 2005 we plan to finish work on the Virgil mounds
and initiate a project on the Mississippian crinoidal buildups
in the Sacramento Mountains, continue supporting a graduate student
studying the Windjana Gorge, and initiate a lidar study of the
Permian shelf margin in Last Chance Canyon, Guadalupe Mountains.
In addition we are supporting a Ph.D. thesis on slope carbonates
that will continue through 2005.
In 2005, we
will finish the geocellular model for Dry Canyon. Several regional
surfaces as well as the stratigraphic grid remain to be built.
The major challenge next year will be to populate the grid with
realistic facies. Especially, the siliciclastic interval and the
mixed siliciclastic and carbonate cycles will require some complex
krigging extrapolation or some other extrapolation methods. We
expect to test several different approaches to model the mound
distribution as well as the facies distribution. In 2004 we used
a stochastic approach to model the mound distribution in 3D. Next
year, we will try both an object-based approach and a multipoint
water outer-ramp buildups of the Sacramento Mountains as described
by Bachtel and Dorobek represent an excellent outcrop analog for
the complex Upper Pennsylvanian Cisco mounds that cap the Horseshoe
Atoll reservoirs at Sacroc and Cogdell. Because of the complex
interplay of steep buildup depositional topography, syndepositional
erosion, and sediment gravity flow processes, an actualistic model
for stratigraphic architecture and porosity/permeability distribution
at Sacroc and Cogdell has been difficult to develop. Although
the Lake Valley is Mississippian and the Cisco is Upper Pennsylvanian,
numerous similarities exist between these two occurrences. Most
important is the deeper water, crinoidal mud-mound facies that
predominates in both cases. Second is the interplay of steep buildup
topography and sediment-gravity-flow processes. Significant differences
between the examples other than age include the icehouse (Horseshoe
Atoll) vs. transitional (Lake Valley) style of stratigraphy and
the isolated atoll vs. land-attached ramp setting. While considering
these differences, we see the excellent outcrop quality of the
Lake Valley exposures, their direct application to addressing
issues of stratigraphic modeling of the Horseshoe Atoll reservoirs,
and the wider interest of complex outer-ramp to slope stratal
geometries of this prograding outer-ramp/buildup setting to provide
a valuable analog dataset.
we will generate for the Alamo Canyon area of the Lake Valley
system include a full 3D geologic model for the 3 ? 4 km area
of the canyon, including buildups, overlying prograding ramp grainstone
lobes, and enough lateral control to capture the along-strike
variability of the lobe to interlobe facies. The model will be
built using lidar, photomosaic mapping, and section measuring,
using available detailed mapping by Steve Bachtel of ExxonMobil.
We will also construct a Gocad model of the system and attempt
to populate this model with petrophysical properties from the
Horseshoe Atoll data for use in seismic modeling.
We are currently
in the process of acquiring data for a 3D stratigraphic model
of the Late Devonian Windjana Gorge outcrop. This area, located
in Western Australia’s Canning Basin preserves arguably
some of the world’s best examples of Paleozoic vertical
escarpment platform margins. We hope to address two key issues
here: (1) What is the spatial variability of depositional facies,
stratal geometries, and sequence architecture associated with
these vertical escarpment platform margins? (2) How does the observed
change from aggradational to progradational platforms in Windjana
Gorge affect fracture patterns in the reefal margin and marginal
slope? The 3D model generated by this study will integrate lidar,
detailed geologic mapping, section measurement, and existing conodont
biostratigraphy. The results of this study will contribute to
our understanding of the spatial complexity of platform margins.
The Last Chance
Canyon area of the Guadalupe Mountains will serve as one of our
key outcrop modeling areas for 2005. This work will be done by
two graduate students as the basis of their Master of Science
theses with completion scheduled for tfall 2006. Last Chance Canyon
is well known to both the U.S. and international communities as
one of the classic examples of sequence-stratigraphic architecture
and has been the focus of several very detailed and thorough studies.
Until recently, our understanding of the Last Chance Canyon geology
has been limited to idealized 2D dip-oriented cross sections.
With the newly developed 3D mapping and modeling technologies
using lidar, we believe that we will be able to add substantially
to our understanding of the nature of prograding systems in carbonates.
We expect to observe the scales of continuity of individual clinoforms,
the line- vs. point-source character of these clinoforms, and
the interaction of deeper-water buildups on the distribution and
source of down-slope sediments.
In 2005 we
plan to continue our investigation into the geology and petrophysics
of carbonate slope deposits by gaining access to several subsurface
datasets from the Permian Basin that are analogous to the Permian
toe-of-slope Victorio Canyon outcrop and by supporting a Ph.D.
student studying well-known slope outcrops. The goal of the subsurface
studies is to apply the knowledge gained from the Victorio Canyon
outcrop to subsurface analogs. The main goal of the Ph.D. thesis
is to produce general carbonate slope depositional models that
identify key relationships linking large-scale geometrical observations
to reservoir spatial heterogeneity. The data for this study will
come from the Upper Devonian strata of the Canning Basin, northwest
Australia, Middle Triassic strata of the Dolomites Alps, northern
Italy, and possibly Upper Permian strata of the Delaware Basin,
West Texas. Understanding of carbonate slope systems tract partitioning
with respect to (1) failure/shedding mechanism, (2) facies types
and distribution, (3) external geometry and internal architecture,
(4) spatial sediment storage capacity, and (5) syn- to post-sedimentary
diagenesis is critical for efficient hydrocarbon exploration and
performance prediction in these settings.
In 2005 we
will continue to generate 3D synthetic seismograms for our different
outcrop models. We are currently improving our modeling capabilities.
We are planning to use a 3D exploding reflector method based on
a one-way propagation of an acoustic wave. This algorithm has
been developed at BEG by geophysicist Sergey Fomel. We will be
able to produce more realistic synthetic volumes where we can
analyze geometrical artifact, amplitude tuning, and the response
of various seismic attributes, test inversion algorithm, and calculate
volumetric error. We plan to produce 3D synthetic seismograms
for the Dry Canyon, Painted Canyon, and Lawyer Canyon models.
Areas of Research
Modeling of the Dolomitization Process
In 2004 we
initiated work on diagenetic modeling by arranging for a graduate
student to model a Cretaceous dolostone bed in a tidal-flat-capped
high-frequency cycle where we can reasonably assume that the hypersaline
reflux dolomitization model applies. The focus is on flow rate,
which means that reasonable values of permeability at the time
of dolomitization are critical. Initial results indicate that
the 1-m-thick dolostone bed can be dolomitized in a few thousand
years, which suggests that hypersaline conditions were short lived.
We anticipate completion of this project in 2005 and plan to continue
this investigation in areas where the dolostone is much thicker
and contains multiple peritidal cycles.
of Internal Core-Plug Variability with CT Scanning
to research methods of integrating rock-fabric descriptions and
petrophysical properties. In the past we have relied on descriptions
of thin sections made from plug-end pieces, which may not accurately
represent the pore structure of the complete core plug. Quantification
of the pore-size distribution within a core plug was initiated
in 2004 by statistical analysis of 35 high-resolution CT scans.
Preliminary results suggest that the standard deviation ratio
is an indicator of average pore size. The standard deviation ratio
will range from zero to one and should indicate the average pore
size relative to the CT scan resolution. Smaller pores will produce
more uniform gray images with smaller standard deviation ratios.
Larger pores will have more variable CT images and larger standard
deviation ratios. Thus, the standard deviation ratio might be
an indicator of rock fabric that at least partially corrects for
the interdependence of mean porosity and porosity variance. We
plan to continue this research in 2005.
and Projects within the RCRL Umbrella
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. In 2004 we did mentoring projects
for Occidental Petroleum, Aramco, and Production Development Oman.
In 2005 we are planning to have mentoring projects with Occidental
Petroleum, Kinder Morgan, and Aramco. Sponsors are encouraged
to contact us with projects that could be mutually beneficial.