Abstracts from 2005 Midcontinent AAPG Meeting in Oklahoma City
Martin K.Dubois, Alan P. Byrnes, Timothy R. Carr, Geoffrey C. Bohling, Saibal Bhattacharya, John H. Doveton, and Nathan Winters
The Hugoton Asset Management Project (HAMP) is a two-year industry-Kansas Geological Survey study of the reservoir systems in the Hugoton Embayment of the Anadarko Basin with modeling the Permian gas reservoir systems and developing a digital field catalog for the pre-Permian reservoirs as primary objectives. The project is a collaboration between the Kansas Geological Survey and nine industry partners designed to provide the knowledge and technical base required for intelligent stewardship, identification of new opportunities, and continued improvement in recovery strategies.
The Hugoton and Panoma Fields, North America's largest, produce from the Wolfcampian Chase and Council Grove groups, respectively, and have yielded 34 TCF gas in Kansas and Oklahoma since the 1930's, an estimated 67% of original gas in place. Remaining gas in this giant stratigraphic trap is mostly in lower permeability pay zones of the 550-foot thick, layered reservoir system consisting of thirteen fourth-order marine-nonmarine sequences.
Direct estimates of water saturation by electric logs are not possible due to deep filtrate invasion. Lithofacies-controlled petrophysical properties dictate gas saturation and accurate discrimination of lithofacies reduces error in predicted permeability and gas volume. The use of neural networks to predict lithofacies at wells, automation and stochastic modeling make it possible to develop robust geologic models for the giant reservoir. Integration of core and log petrophysics with the geologic model provides an accurate static engineering model. Numerical reservoir simulations validate the static model and help identify higher pressure, under produced intervals in the layered reservoir system and forecast future production rates.
Geoffrey C. Bohling, Martin K. Dubois, John H. Doveton and Alan P. Byrnes
The Hugoton Asset Management Project has focused on the development of a geomodel for the Hugoton and Panoma fields. This process has required automated processing of large data volumes at several steps, including prediction of lithofacies from geophysical well logs in numerous wells based on a neural network trained on log-facies associations observed in cored wells, generation of geologic controlling variables (depositional environment indicator and relative position in cycle) from a tops dataset, and computation of porosities corrected for mineralogical variations between facies and for washouts. In addition, we have developed code for batch processing the predicted facies and corrected porosities at the wells to estimate water saturations and original gas in place using petrophysical transforms and height above free water level, providing a quickly computed measure of the plausibility of the geomodel. The neural network code, including batch facies prediction based on logs from a large number of LAS files, has been added to an earlier Excel add-in for nonparametric regression and classification. However, the computationally intensive task of determining the optimal neural network parameters through crossvalidation has been accomplished using scripts in the R statistical language. The remaining tasks have been implemented in Excel, with the controlling parameters for each process specified in a simple spreadsheet layout. Due to the data volume involved, automation of these procedures has been crucial to the development of the Hugoton-Panoma geomodel.
Relations between Lithofacies and Porosity, Permeability, Capillary Pressure, and Relative Permeability in the Chase and Council Grove groups, Hugoton Embayment, Kansas
Alan P. Byrnes and Martin K. Dubois
Fundamental to reservoir geomodel construction is the population of cells with basic lithofacies and their associated porosity, permeability, and fluid saturation. Petrophysical properties vary between ten major lithofacies classified. Mean and maximum porosities increase from mudstones to grainstones. In situ porosities (fi) correlated with routine porosity (froutine) using: fi = 1 x froutine-0.68. Equations developed to predict permeability and water saturation use porosity as the independent variable because porosity data are the most economic and abundant, and are well correlated with other variables for a given lithofacies. In situ Klinkenberg (high-pressure gas or liquid-equivalent) gas permeability (k) exhibits a power-law relationship with porosity though the relationship changes in some facies at porosities below ~6%. Each lithofacies exhibits a relatively unique coefficient and exponent in k-f relationships of the form:
k = 10 B fiA
At fi > 6% permeability in grainstone/ bafflestones can be 30X greater than mudstones and >100X greater than marine siltstones of similar porosity. Differences in permeabilities between nonmarine silt/sandstones and shaly siltstones range from 3.3X at 12% porosity to 7X at 18%. Full-diameter cores frequently exhibit permeabilities as great as 50X plug permeabilities due to stress relief fracturing.
Capillary pressures and corresponding water saturations (Sw) vary between facies, and with porosity/permeability and gas column height. Threshold entry pressures and corresponding heights above free water level are well correlated with permeability. Differences in Sw between facies increase with decreasing porosity and decreasing height above free water. For rocks with k > 0.1 md, relative permeabilities exhibit consistent exponents.
John H. Doveton
The wireline log database of the Chase Group in southwestern Kansas is one of the largest in the world and provides an almost unique opportunity to relate petrophysical measurements to geology and reservoir properties over a wide range of spatial scales. The nuclear logs of spectral gamma-ray, neutron porosity, density, and photoelectric factor are commonly available and show distinctive patterns that can be linked with lithological composition, as well as depositional and diagenetic signatures. The spectral gamma-ray uranium measurement distinguishes "hot" carbonates, where uranium has accumulated, primarily in diagenetic cements. Mathematical inversion of the nuclear logs transforms them to compositional profiles of pore volume, shale, dolomite, silica (quartz and/or chert), calcite, and anhydrite. These compositional transforms are validated by the "ground truth" from core, and the fine vertical resolution of the logs can be extended to detailed analyses of cyclic patterns and facies associations. Reservoir formation evaluation in the Chase Group is complicated by the extensive invasion that occurs in mud-drilled wells, with the result that saturations estimated from resistivity logs are commonly suspect. In addition, gas effects on porosity logs are highly variable but must be accommodated in the estimation of accurate porosity values. The drilling of foam-based wells and their use in comparative studies of invasion effects that have been published recently have been helpful in improving formation evaluation strategies of logs in the Chase Group.
Martin K. Dubois, Alan P. Byrnes, and Geoffrey C. Bohling
The Hugoton and Panoma Fields, North America's largest, produce from thirteen fourth-order marine-nonmarine sequences of the Wolfcampian Chase and Council Grove groups, respectively. The degree of heterogeneity, large volume to be modeled, and an immense data set made developing a geologic model by conventional methods impractical. Geostatistical methods (artificial intelligence and modern modeling software) and automation facilitated building a finely detailed 3D cellular geomodel using a four step workflow: 1) define lithofacies in cores and correlate lithofacies to electric log curves (training set), 2) train a neural network to predict lithofacies, 3) predict lithofacies at non-cored wells with trained neural network, and 4) predict lithofacies between wells using stochastic methods to populate a three dimensional cellular model. A fifth step is to populate the cellular model with lithofacies associated petrophysical properties and fluid saturations for volumetric analysis and numerical simulation.
The lithofacies spectrum was split into eight marine and two nonmarine lithofacies primarily based on texture and grain or pore size. Marine carbonates and sandstones are the principal reservoir facies in both the Chase (Hugoton) and Council Grove (Panoma). Two lithofacies unique to the Chase, dolomitized coarse-grained grainstones and fine-grained marginal marine sandstones, are the dominant storage and flow lithofacies in the Chase. Grainstones, packstones, wackstones and fine-crystalline dolomites are the dominant reservoir lithofacies in the Council Grove and contribute significantly in the Chase as well. Other marine lithofacies, siltstones and mudstones, and nonmarine lithofacies, coarse siltstones and shaly siltstones, provide some storage especially when high in the gas column.
Depositional model and distribution of marginal marine sands in the Chase Group, Hugoton Gas Field, Southwest Kansas and Oklahoma Panhandle
Nathan D.Winters, Martin K. Dubois, and Timothy R. Carr
Since the 1930’s, the Hugoton Gas Field of southwest Kansas and the Oklahoma panhandle has produced approximately 29 TCF from the Wolfcampian (Lower Permian) Chase Group. The rocks of the Lower Permian Chase Group were deposited on the broad shallow shelf of the Hugoton Embayment of the Anadarko Basin. Important reservoir lithofacies are dolomitized grainstone, carbonate packstone and grainstone, and marginal marine sandstone. In the Hugoton Field, marginal marine sandstone lithofacies comprise a significant portion of the reservoir volume, but are not well characterized. In many of these very fine grained sandstones, porosities range from 15-25% and have permeability in the 10-100 millidarcies range, making them excellent storage and flow units. The sandstones of the Herrington, Winfield, and Towanda Limestones and the Holmesville Shale are at the top and base of the marine portion of glacio-eustatic driven, marine-nonmarine cycles. Based on sedimentary structures, stratigraphic context (position within the marine-nonmarine cycles) and depositional geometry (broad, relatively thin sheets), these sandstones are interpreted as deposits of tidal flats to shallow subtidal environments of the upper shelf.
Saibal Bhattacharya, Martin K. Dubois, Alan P. Byrnes, John Doveton, and Geoff Bohling
Reservoir engineering studies, in the Hugoton Asset Management Project (HAMP), included analyses of available pressure and production data, material balance studies, and simulation of single and multiple well systems in select areas. The intent of these studies is to validate the underlying reservoir geomodel, developed by integrating inputs from geology, log analysis, petrophysics, and neural-logic based lithofacies prediction, by matching the production/pressure histories at both regional- and well-levels given the volumetrically estimated original-gas-in-place (OGIP).
Surface shut-in (SI) pressure data from Chase (parent and infill) and Council Grove wells in 2 different regions indicate that reservoir pressures declined along a common trend. However, questions remain about how representative 48-72 hr surface shut-in recordings are of average reservoir pressures. Available data also indicate that new wells at completion record 20-30 psi higher SI pressures than older neighbors but pressures soon fall inline with the regional decline trend. Material balance (MB) calculations indicate that the later Council Grove well added reserve volume beyond that drained by earlier Chase (parent) well in the same Section. Possible overlapping of drainage areas of newer wells with those of older ones limits the applicability of MB-calculated OGIP in charging the geomodel. DST permeability matches corresponding plug and whole core permeability values when corrected for sub-surface conditions. Reservoir simulation studies at Flower 1 area indicate that an upscaled geo-model of 25-layer can match differential pressure depletion, evident from layer-specific shut-in pressure data available from the Flower 1 well, while producing under historic production constraints.
Reservoir pressures suggest communication between Hugoton and Panoma Fields and provide insights on the nature of the connections
Martin K. Dubois1, Alan P. Byrnes1 and Richard Brownrigg2
1Kansas Geological Survey, University of Kansas; 2Department of Electrical Engineering and Computer Science, University of Kansas
Analysis of Hugoton and Panoma pressure through time and pressure versus cumulative production at well to field scales suggests the vertically stacked Hugoton and Panoma Fields are in communication and are one giant reservoir system. Insights on the nature of the communication are gained by comparing temporal and spatial relationships of reservoir pressure with spatial geologic variables. However, pressures by zone data support the concept that the Hugoton and Panoma are layered reservoirs with relatively low cross flow between differentially depleted zones, thus presenting a quandary.
Hugoton and Panoma Fields in Kansas have undergone three phases of development: pre-1950 Hugoton "parent" wells, Panoma wells in the 1960's and Hugoton "infill" wells in the late 1980's. Hugoton and Panoma biannual 72-hour wellhead shutin pressure (WHSIP) through time are nearly equal and paralleling and the correlation of change in slope of pressure versus cumulative production with development phases, all suggesting communication. Dynamic visualization of the WHSIP data volume through time and space provides a novel, multi-dimensional view of subtle anomalies in reservoir pressure. Linear to sub-linear anomalous pressure regions in 3D space coincide with lineaments in the first derivative of reservoir structure that are likely related to basement fractures and faults. Mile to tens of miles scale pressure anomalies may be a result of vertical communication enhancement by swarms of small-scale vertical fractures associated with these lineaments. Though avenues for large-scale communication between Hugoton and Panoma Fields the lineament spacing may be too distant to provide complete pressure equalization between zones.
Kansas Digital Petroleum Atlas: A Step Toward a Cyberinfrastructure for the Oil and Gas Reservoirs in the Hugoton Embayment
Timothy R. Carr, John R. Victorine, Jeremy D. Bartley, Melissa C. Moore, Asif Iqbal, Keith Hunsinger, Praveenkumar Ponnusamy, and Kurt K. Look
The Kansas Digital Petroleum Atlas (DPA) provides worldwide access to constantly increasing data and interpreted information from every oil and gas reservoir in Kansas. Data from over 6,000 fields and 300,000 wells in Kansas can be accessed. Programs, developed through the DPA, provide oil and gas operators the tools to make exploration and development decisions using production data, interpreted well logs, and real-time mapped petroleum information. The DPA provides online tools to query, interpret, map, and display the latest information and research results, which could be anywhere in the world. “Published” products are created on demand, customized to address specific questions, and access data continuously updated and enhanced. Systems will allow operators to submit data online. Through the DPA, the data will be available in real-time (e.g., well logs completion reports and production). The DPA has significantly altered the relationship between research results, data access, the transfer of technology, and our relationship with our clients.
“Pages” in the DPA are generated on demand using online clients. Previously completed products, such as field and basin studies, are automatically updated with the latest production and well data. Raster images such as completion reports are scanned and uploaded into relational databases and can be used for efficient construction of larger scale studies. The DPA Project continues to provide improved access to a “published” product and ongoing technology transfer activity.
The DPA is illustrated through examples of access, display and analysis of petroleum and geologic information generated by the Hugoton Asset Management Project (HAMP).