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 compiling and making some of these data available to sponsors in the form of databases populated with tabular and image data. In 2005 we initiated a multiyear project of synthesizing all available data and developing a database on a stand-alone server with a completion target of 2007.
In 2005 we established a new members-only data server with a Web portal available for all current members with appropriate password security measures (http://begsql.beg.utexas.edu/rcrl/login.aspx ). Data loading has been tested using data from Lawyer Canyon outcrop and Seminole San Andres reservoir using lidar, geological models, photomicrographs, porosity-permeability tabulations, wireline log data, and publication lists. The initial testing of the system is complete, and a number of sponsors have logged on.
In 2006 we will begin to add data, giving preference to study areas that have the most complete suite of data and adding older studies that best complement the current study areas. The data types will be expanded to include current and past RCRL presentations and posters. We plan to develop a method this year to integrate new data into the database as part of our regular workflow. This process will promote regular updates to the database as we complete various phases of a research project. This effort will help keep the flow of information from the RCRL research projects to our sponsors steady and timely while offering the sponsors a way to interactively search the entire database using basic search criteria.
Geological, Petrophysical and Seismic Characterization of Vuggy Porosity
Geological and petrophysical characterization of vuggy porosity as it relates 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 large vugs 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 regional tectonism, fracturing at many scales, hydrologic geochemistry, dissolution, and cavern collapse. Touching-vug and matrix porosity together form a coupled pore system that is poorly understood and difficult to simulate.
Touching vugs are normally composed of large pores that cannot be adequately studied at the core scale. Over the past few years, the RCRL has developed a research program designed to investigate the geology and flow properties of touching-vug pore systems. This program has been focused on studying touching vugs at various scales using samples and shallow well tests from carbonate outcrops and production data from carbonate reservoirs.
Pipe Creek Outcrop Study
The 1-ft scale has been studied using a A Pipe Creek outcrop sample of rudist rudstone containing large rudist vugs and interrudist pore space has been studied at the 1-ft scale. High-resolution CT scans and laboratory fluid-flow experiments have been conducted with this sample to determine vug-size statistics, vug-connection geometry, permeability, and tracer transport properties. This work is the subject of the Petroleum Engineering Ph.D. dissertation of Liying Zhang and two Society of Petroleum Engineers papers. As a result of this work, published in 2005, we now have a collection of techniques for quantifying vug-size statistics and vug-connection geometry from CT scans that we believe think will have general applicability in vuggy samples as much as 1 ft in diameter. In 2006 we plan to test these methods on other touching-vug samples.
In 2005 we initiated a larger scale study at the Pipe Creek outcrop using two 25-ft wells drilled 5 ft apart. A preliminary air-extraction test with these wells indicates connected vugs at the 5-ft scale and an effective permeability of 1 darcy, consistent with whole-core analysis of four core samples from these two wells. We plan additional well tests for 2006, including cross-well ground-penetrating-radar (GPR) tomography to determine porosity at the 5-ft scale and tracer tests to determine the amount of pore space contributing to fluid flow.
Lake Medina Outcrop Study
For the 100-ft scale, we are studying karst exposed in a Cretaceous road cut near Lake Medina, Texas. This exposure contains small faults, fractures, solution-enlarged fractures, and caves that are exposed on road cuts on both sides of the highway. The objective of this study is to build an explicit dual-porosity model showing the distribution of solution-enlarged fracture porosity, large vugs (including caves), and matrix porosity. This model will be used to develop improved methods for modeling flow in dual-porosity systems. In 2005 we collected detailed lidar data of fracture pore space and large vugs. We are able to image pore space down to about 1 cm in width/diameter. The lidar images show the distribution of large pore space along fracture traces and the spatial orientation of fracture and large-vug pore space. The challenge for 2006 is to find a means for constructing a 3D image of this pore space.
Franklin Mountain Karst Study
Karst processes are significant 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 performance predictions in these systems remains inadequate.
In 2006 our research will focus on detailed diagenetic field mapping of the uppermost El Paso Group (Lower Ordovician), just below the regional unconformity (30 m.y. time gap) that marks the top of the Sauk C second-order supersequence set exposed in the Franklin Mountains. Mapping this unit will help us understand the paleocave system at the time the overlying Montoya (Upper Ordovician) was deposited and give us a better understanding of how the cave system evolved and what impact it has on the touching vugs found in the overlying Montoya Formation. As part of this mapping we will collect samples from the breccia matrix for conodont analysis in an effort to date the time of cave formation.
Hobbs San Andres Karst Reservoir Study
At the reservoir scale, we have initiated a study of the Hobbs San Andres reservoir, New Mexico. Production data and core descriptions from the Hobbs reservoir strongly indicate the presence of a diffusion-flow paleokarst system and associated touching-vug pore network. We have described depositional facies and touching-pore types from seven cores and are in process of describing four new cores. In addition, we have described 52 thin sections from two wells. These data will form the basis for an extremely robust sequence-stratigraphic rock-fabric framework. Pore-type descriptions, core analysis, and well tests have provided an initial view of the permeability problem. The core analysis permeability values are generally a factor of 3 to 5 times greater than the matrix values estimated using the rock-fabric method, and permeability calculated from pressure transient analysis is 10 to 100 times higher than that from core measurements. Core descriptions and injection profiles have provided an initial insight into the distribution of the touching-vug pore space that caused these large permeability-enhancement figures. Core descriptions suggest that karsting from meteoric waters forms touching vugs, and thin-section descriptions suggest that anhydrite and gypsum dissolution may contribute to permeability enhancement.
We will continue this study in 2006 by describing the remaining cores and thin sections as needed. Image logs will be calibrated with core descriptions to enhance their value in uncored wells. Injection profiles appear to be the best source of data for locating high-permeability zones in uncored wells, and this extensive body of data will be incorporated into our 3D model. We plan to investigate fracture modeling because we think that the karst hydrology is, to some degree, controlled by preexisting fracture patterns.
We have worked with the existing 3D seismic data and have found serious problems with the melding of the three 3D surveys to the extent that we are unable to image major karst features in the San Andres. However, amplitude analysis may be useful for mapping high porosity zones for stratigraphic control. A combination of high-resolution sequence-scale seismic interpretation coupled with attribute analysis and inversion will be generated in 2006 to attempt to define the representative seismic fingerprint of this diffuse-flow paleokarst network.
Seismic Modeling of Separate-Vug Pore Space
Our goal is to develop seismic processing and interpretation strategies to identify and map moldic porosity from 3D seismic data. It is well known that separate-vug pores add little to permeability, so mapping the distribution of this pore type seismically is a necessary first step in estimating permeability. In 2005 we studied seismic response to vuggy porosity in high-frequency cycle 7 of the Lawyer Canyon outcrop. The study showed that the difference in velocity between the separate-vug grainstone and nonvuggy grainstones can generate a large enough impedance contrast to be detected on poststack seismic images. Initial results from modeling amplitude variation with offset (AVO) of the cycle 7 grainstone showed that this impedance difference might be detected and mapped using AVO techniques.
In 2006, we will continue research on characterizing and quantifying the petrophysical signature of moldic grainstones using the Lawyer Canyon outcrop. A statistically significant number of acoustic and elastic rock property measurements will be made from this outcrop to build a more realistic elastic model and synthetic seismogram of the Lawyer Canyon outcrop. In addition, we will search for a subsurface data set from an oomoldic reservoir having cores, modern wireline logs, and 3D seismic data.
3D Modeling of Carbonate Reservoir Outcrop Analogs
Now that the methodology of construction of realistic 3D models of classic carbonate outcrop analogs is well in hand, we will shift our emphasis in 2006 to using quantitative parameters of facies proportions and dimensions to improved 3D modeling of analogous reservoirs. Two main areas of advance that we will pursue are (1) improved methodologies for 3D facies modeling, and (2) seismic imaging of these geologically realistic models. Initial results from the Hobbs model and the Lawyer Canyon seismic imaging show the potential to generate significantly more realistic reservoir models. A more focused effort will be made in Hobbs field once a refined correlation scheme for the upper San Andres is developed in this reservoir. We will draw on data from mid-upper Guadalupian outcrop examples to form the basis of this modeling.
Using outcrops to model seismic response continues to be an important research direction for the RCRL. The primary goal is to determine what geologic features can be imaged with seismic and at what resolution. We have a number of robust outcrop stratigraphic models from a number of facies tracts that are suitable for seismic quantification and continue to develop others. This data set provides the basis for our goal of improved geologic interpretation of seismic images from carbonate reservoirs.
We are confident that collecting 3D spatial statistics and modeling seismic response from outcrops will provide our sponsors with tools to better construct geologic images.
New Approach in 3D Outcrop Modeling
In 2005, three 3D outcrop models (Victorio Canyon, Lawyer Canyon, and Painted Canyon) were generated using lidar images and both deterministic (categorical kriging) and stochastic (Gaussian simulation) approaches to model the facies distribution. This year will refine these techniques and investigate alternative modeling strategies to realistically populate these 3D models with facies information. Currently we are using kriging to model laterally continuous facies, and smaller discontinuous sedimentary bodies are mapped either deterministically or modeled using Gaussian stochastic simulation. We believe think that we have developed a very reliable workflow for constructing basic chronostratigraphic surfaces.
Filling the reservoir volume between control points (whether outcrops, measured sections, or wells) with discontinuous sedimentary bodies is the principal problem. In 2006, we plan to develop efficient methods of integrating deterministic forward stratigraphic modeling into the process of building 3D geocellular models using Dioniso, a diffusion-based forward stratigraphic modeling package developed by IFP. Initially, we plan to reproduce the 3D stratigraphic architecture of our existing outcrop-based detailed 3D models. Then, we are planning to use Dioniso to generate generic 3D models using stratigraphic architecture and facies geometries extracted from the 3D outcrop mapping and use these generic models as training images for multipoint statistic extrapolation of the outcrop data to generate a 3D geocellular model.
Slope depositional systems and their reservoir architectures will continue to be an element of the RCRL research in 2006 centering on three systems, the Devonian of Australia, which carries on from 2005, the steep-rimmed margins of the Capitan foreslope in the McKittrick Canyon to Pine Canyon area of the Guadalupe Mountains, and the low-angle (2-6 degree) slope systems of the prograding Victorio Peak to Bone Spring transition along the Western Escarpment of the Guadalupe Mountains. We have delineated a series of styles of slope sedimentation and reservoir development that looks at a process/product classification of platform to basin transitions. We now need to test the applicability of this scheme for these classical exposures and to highlight additional sources of data in the slope setting. One possible source of data will be from shallow seismic profiles of ramp-to-rimmed shelf transitions such as observed along the West Florida margin and in the Persian Gulf.
Outcrop Seismic Modeling
In 2006, we will continue our investigation of the seismic response of outcrop models. We will use the BEG collection of advanced forward seismic modeling methods to produce more realistic synthetic volumes to analyze geometrical artifacts, amplitude tuning, volumetric error, and the response of different seismic attributes and to test inversion algorithms. We are planning to use this approach for the Dry Canyon, Painted Canyon, Last Chance Canyon, and Lawyer Canyon models.
In addition we are planning research into the spatial variability of carbonate acoustic properties and the effects of this variability on seismic modeling. Carbonate rocks exhibit significant variability in their petrophysical and acoustic properties even for samples having very similar petrographic characteristics. Past studies have shown that most of the total variance in petrophysical properties often occurs at scales of 10 ft or less. Other studies have demonstrated that significant variability in carbonate acoustic properties can be caused by porosity and pore-shape variations within a single rock-fabric class. In our Victorio Canyon seismic modeling spatially correlated impedance noise created a significant change in the synthetic seismogram and resulted in more realistic seismic images. However, very few studies have investigated the spatial distribution of acoustic properties within sedimentary bodies. Therefore, in 2006 we plan to conduct outcrop measurements to evaluate the spatial variability of acoustic properties in carbonates and to begin seismic modeling studies to investigate the effects of this variability on the seismic response of carbonate systems.
Subsurface Reservoir Modeling
In past years the RCRL has developed a systematic method for integrating high-resolution sequence stratigraphy, rock fabrics, core data, well-log data, geostatistics, and scaleup approximations to produce superior 3D models of petrophysical stratification common in shallow-water platform carbonate reservoirs. These methods were demonstrated in the construction of a model for a 1-mi 2 area of the South Wasson Clear Fork reservoir in West Texas. We think that these model construction methods can be adapted for improved 3D petrophysical modeling of any stratified carbonate system, and we are actively seeking opportunities to test the methods in other reservoirs, particularly Cretaceous examples.
Because of the small size of the South Wasson Clear Fork model, only moderate lateral petrophysical trends were encountered. Therefore, in 2006 we plan to test our petrophysical model construction methods in a larger reservoir having more significant lateral petrophysical trends and to include seismic methods for filling the interwell volume. We have selected the Fullerton Clear Fork reservoir in West Texas because of its size (55 mi 2 ), existing high-resolution sequence-stratigraphic and petrophysical framework, and existing database of core data (25 wells) and quality-controlled well logs (850 wells). The BEG has recently completed a detailed geological and petrophysical model of this reservoir. Our approach will be to test our more advanced methods of distributing properties within the existing stratigraphic framework and using existing petrophysical properties calculated from wireline logs using the rock-fabric method.
An important part of the Fullerton Clear Fork project will be investigation of advanced forward seismic modeling of subsurface reservoir models to investigate the sensitivity of seismic response to interwell stratigraphic and petrophysical uncertainty and to provide fundamental insights for improved seismic inversion.
Diagenetic Modeling of the Dolomitization Process
Prediction of dolostone patterns requires knowledge of the dolomitizing hydrology. Dolomitization patterns linked with hypersaline density flow are predictable because the source of the dolomitizing water is known to be evaporitic tidal flats and a hydrologic model can be constructed on the basis of tidal-flat-related paleotopography. In 2004 a graduate student accepted the challenge to construct a hydrological model of reflux dolomitization using a dolomitized tidal-flat cycle exposed in the Cretaceous outcrops near Austin, Texas, as data. The focus is on flow rate, which means that reasonable values of permeability at the time of dolomitization are critical. The student has completed the study, and the results indicate that the 1-m-thick dolostone bed can be formed in 500 years, an amazingly short period of time. This finding suggests that reflux occurs periodically, perhaps related to tidal flooding events, and that the 500 years is spread over thousands of years.
We are supporting a doctoral student to continue this research. Her research problem is to model dolomitization of the Algerita Escarpment, assuming a reflux dolomitization model. We hope that this project will produce reliable forward-modeling methods that can be used to predict the distribution of reflux dolostone.
Mentoring 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. In 2005 we did mentoring projects for Aramco, Kinder Morgan, and Occidental Petroleum.