The Frontier Formation of the Moxa Arch in southwestern Wyoming provides an excellent example of the interplay among sedimentation, diagenesis, and tectonics on reservoir quality and performance. During the Cretaceous, thrust uplift and crustal loading occurred in the Sevier Orogenic Belt to the west of the foreland basin. The thrust sheets provided ample sediment to the Moxa Arch area during Frontier time. These sediments accumulated in wave-dominated deltaic, strandplain, coastalplain, and incised valley-fill depositional environments. The tectonic activity in the Sevier Orogenic Belt caused recurrent differential movement of orthogonally-shaped basement blocks along the Moxa Arch. These Movements created fractured lineaments at block boundaries. In addition, the recurrent movements of basement blocks influenced paleostructuring, diagenetic fluid migration paths, and sediment dispersal patterns of the Frontier.
The depositional facies of Frontier sediments control the primary porosity and permeability trends of Frontier reservoirs along the Moxa Arch. Post-depositional fractures caused by recurrent differential movements along zones of weakness at basement block boundaries secondarily enhance permeability and performance characteristics of Frontier reservoirs. Both the depositional facies and post-depositional fracturing of the Frontier influence the diagenetic trends affecting secondary porosity and permeability characteristics of Frontier reservoirs along the Moxa Arch. It is this complicated interplay of depositional, tectonic, and diagenetic influences that control the characteristics of Frontier reservoirs along the Moxa Arch.
The Frontier Formation in the Green River Basin (Figures No. 1 and 2) is a stratigraphically complex exploration target. There are multiple stacked reservoirs that reflect the interplay among variations in sediment supply, fluctuations in eustatic sea level, tectonism (contemporaneous and post-depositional), and diagenesis. Since the l920's, more than 1.45 TCF of gas has been produced from reservoirs in the Frontier Formation of the Green River Basin. Estimated ultimate recovery from more than 1540 producing Frontier wells is almost 2 TCF of gas. Most of this gas occurs in stratigraphic traps.
This paper demonstrates a methodical approach in the implementation of current hydraulic fracturing technologies. Specific examples illustrating the evolution of a consistent reservoir/hydraulic fracturing interpretation are presented in a case history of three GRI-Industry Technology Transfer wells. Detailed modeling of these project wells provided an overall reservoir and hydraulic fracture description that was consistent with respect to all observations. Based on identification of the fracturing mechanisms occurring, the second and third project wells show the capabilities of real- time diagnostics in the implementation of hydraulic fracture treatments. By optimizing the pad volume and fluid integrity to avoid premature screenouts, significant cost savings and improved proppant placement were achieved. The production and pressure build-up response in the first project well verifies the overall interpretation of the reservoir/hydraulic fracture model and provides the basis for eliminating the use of moderate strength/higher cost proppant over sand in low permeability/higher closure stress environments.
The successful implementation of an applied hydraulic fracturing project requires a balanced mix of data acquisition combined with applied field implementation projects. What results from a project of this type is a reservoir and hydraulic fracturing interpretation that matches what actually occurs both during the fracture treatment and also during the actual production response. Once this type of interpretation is achieved, significant optimization can occur to provide the minimum cost for the best economic results.
Ideally, the data acquisition is mixed with the field implementation efforts in order to high grade the interpretation and quality of the data and minimize cost resulting from impractical applications, unnecessary data or an overkill of data acquisition efforts. Each step of a project should be pursued by methodically weighing the cost/benefit of each data acquisition and field implementation effort.
The basic framework for an applied hydraulic fracturing project consists of the following phases:
An expensive fracturing study has been completed, evaluating over 400 fracture treatments on more than 130 Mesaverde gas wells in the Piceance Basin. This study involved a comparison of many types of fracturing fluids and techniques. Linear gels, foams, energized fluids, titanate crosslinked fluids, zirconium crosslinked fluids and borate crosslinked fluids have all been utilized in stimulating the Mesaverde formation in the study area. The authors evaluated the success of previous treatments which ultimately resulted in stimulation design changes, operational modifications, and expanded quality control methods. Dead strings were utilized on a number of treatments to record static tubing measurements and incorporate into a three dimensional (3D) fracture stimulation model. The modeling work assisted in determining fracture geometry which led to job size changes and revised perforating schemes. This paper will illustrate a number of correlations relating the effect of fracturing fluid viscosity, job size and completion method. Based upon the correlations of the study and the modeling work conducted, Barrett Resources has been able to greatly optimize their fracture treatments from the standpoint of cost effectiveness and enhancement of natural gas production.
Barrett Resources Corporation has drilled and completed over 130 Mesaverde wells in the Piceance Basin since 1984. These wells are located in the Grand Valley, Parachute, and Rulison Fields and range in depth from 5900' to 8500' (Figure 1).
The Long Beach Unit (East Wilmington Field, California) consists of nearly a thousand active wells and currently produces about 47,500 BOPD. Values of the in-situ stresses are needed for designing hydraulic fracture stimulation treatments. Estimation of the stresses is complicated by the fact that the field is in an active strike-slip tectonic regime where the vertical stress magnitude is intermediate between the minimum and maximum horizontal stresses. Stress magnitudes from casing-leakoff tests and density log calculations are consistent with this structural interpretation.
Stresses were directly measured in several casing-leakoff tests, micro-fracs, and mini-fracs performed in various sands and shales in the field. The minimum horizontal stress in the shales is roughly linear with depth, while in the sandstones it increases more rapidly, and is more variable than in the shales.
This paper presents a semi-empirical model for calculation of in-situ stresses from logs. It assumes that the stresses are determined by a mechanical stability criterion, and that the ratio of the mean horizontal effective stress to the vertical effective stress is constant from one formation to another. This ratio can be determined from casing-leakoff tests. The new model matches the observed stress contrasts better than conventional elastic models.
Knowledge of the subsurface stresses is important for many reasons in the oil and gas industry. For instance, the horizontal in-situ stress magnitudes usually determine whether a hydraulic fracture will be contained within a given formation. The mechanical stability of a wellbore is partly determined by the relative magnitudes of the in-situ principal stresses. Formation compaction also depends on the stress magnitudes. However there are no generally applicable methods of estimating in-situ horizontal stresses short of direct fracturing measurements. Techniques that use acoustic logs to predict stress profiles have been developed for extensional tectonic environments, but are questionable where strike-slip tectonics prevail. This paper addresses the problem of stress estimation from logs in a strike-slip stress regime at Long Beach, California.
After a brief description of the geology at Long Beach, we will discuss conventional methods for predicting in- situ stress profiles from standard wireline logs. We then propose a new method specifically for strike-slip environments. Next we describe the state of stress at Long Beach as directly measured by microfracture, minifracture, and casing-leakoff tests. Finally, the stress-estimation models will be compared to the fracturing results, and conclusions will be drawn as to the applicability and limitations of the models.
Carbon dioxide injection, either by huff and puff or displacement operations, results in a crude oil viscosity reduction at pressures below the miscibility conditions. Carbon dioxide miscibility occurs in reservoirs at miscible temperature and pressure, but these conditions are not possible in shallow reservoirs. Improved oil recovery in a shallow reservoir depends on the degree of viscosity reduction at the reservoir temperature and pressure. A recovery project's success depends on the interaction between the carbon dioxide and the reservoir system.
Gas production from the Almond Formation in the Standard Draw trend can only be accounted for by draining numerous layers of tight gas sands via the permeable upper bar sand. Discovery of this field originally focused upon production from this bar sand. But continued development can not be explained simply by considering depletion of a 30 foot sand. Gas volumetrics verify the need to include lower sands in reservoir analysis. Core obtained from the Almond bar sand confirm petrophysical constants used in our models. Our results imply that economic levels of gas production should be possible wherever a similar horizontal conduit can be tied into gas saturated layers through massive hydraulic fracturing.
The Mesaverde Group sandstones of the Greater Green River Basin (GGRB) contain possibly one of the largest untapped natural gas resources in the United States. Total resource estimates are up to an astronomical 2000 Tcf (Law et al, 1986), yet less than 2 Tcf of gas has been produced to date.
Figure 1 shows an approximate contour map of gas production from the Mesaverde sandstones in the GGRB. Figure 1 covers an area encompassing the entire GGRB, about one-quarter of Wyoming, the southwestern portion. Details of this production map have been published earlier by Iverson (1992). The highs on this map indicate that most of the 2 Tcf produced from the Mesaverde group originates from high permeability sweet spots. Upon careful analysis of these "sweet spots" it is found that most are completions in the Almond Formation at the top of the Mesaverde Group. Notable exceptions are the Trail interval of the Ericson, and Adaville production at Big Piney. Figure 1 shows cumulative gas production from the Mesaverde Group sandstones If gas reserves are plotted, then a similar picture results, but the Standard Draw peak becomes much more dominating. When the distribution of this 2 Tcf gas production is compared to the 2000 Tcf of gas in place, it becomes apparent why the percent recovery is so low. Only small portions of the basin as a whole are actually productive from the Mesaverde. At those locations, only a small portion of the section (just the upper Almond) is being produced. Hopefully, an improved understanding of these isolated "sweet spots" will help convert a portion of the massive Mesaverde gas resource into reserves.
J.P. Spivey* and M.E. Semmelbeck*
A procedure is presented which allows forecasting long range performance of dewatered coal and fractured gas shale reservoirs having nonlinear adsorption isotherms, using constant pressure solutions to the flow equation for slightly compressible liquids. A correlation is presented to show the range of applicability of this procedure.
Production decline curves are routinely used by engineers to predict the future performance of oil and gas wells. Because the results of decline curve predictions are used for calculating asset value and estimating future revenue, they are one of the most important tools reservoir engineers use. There are numerous variations on the basic exponential or hyperbolic decline analysis method. Fetkovitch and others have extended the decline curve analysis method to handle gas wells properly and to be able to estimate reservoir properties from the analysis of these data. However, there has been considerable drilling activity in the last 10 years into unconventional reservoirs whose wells do not follow the traditional production decline characteristic shapes. Among these problem reservoirs are coalbed methane and fractured shale reservoirs.
Two factors complicate the prediction of future gas production rates in many coalbed methane and fractured shale reservoirs such as the Devonian Shale of the Appalachian Basin, the Antrim Shale of Michigan and the New Albany Shale in Indiana. The first factor common in Antrim and New Albany reservoirs is high initial water saturation and essentially zero gas flow rate at the beginning of production. The second factor common to all fractured gas shales is desorption of gas from organic material within the reservoir rock. Both of these factors can result in well behavior that is not properly predicted by conventional decline curve methods.
Because of the complex production behavior of coalbed methane and fractured gas shale wells, the best way to predict performance is to use a numerical reservoir simulator which accounts for all of the mechanisms occurring during production. However, use of reservoir simulation may not be practical for all situations, particularly when many wells must be analyzed rapidly or when reservoir simulation is not available. This paper presents a rapid analytical solution that can account for production from reservoirs undergoing desorption. One extension of this method over others presented in the literature is that it accounts for nonlinear Langmuir sorption isotherms.
A method for analyzing transient flow-after-flow (FAF) deliverability test data from both gas and oil wells is presented. The paper describes the derivation and application of this method to several field and synthetic cases. The method solves the transient absolute open flow potential (AOFP) and average reservoir pressure (p) by evaluating the parameters of the Forchheimer (a and b) and the empirical backpressure (C and n) equations together with p. Consequently, this new method allows one to describe the stabilized deliverability equation from transient test data, given reasonable estimates of the reservoir drainage area and shape. This approach is a significant improvement over currently available methods, which require an independent, a priori knowledge of p for establishing a well's AOFP, along with the need to conduct a stabilized flow segment of the test.
All the resulting formulations are flexible enough to handle the pressure, pressure-squared, and pseudopressure approach for gas wells. This methodology is not restricted to gas wells as oil wells also lend themselves to the proposed analysis procedures. For oil wells, we used the pressure approach for single-phase flow and the pressure-squared approach for two-phase flow.
In the proposed technique, a well's orientation is unimportant. For example, a horizontal well's deliverability may also be characterized using the new method. In addition, layered reservoirs may be analyzed with the proposed technique. We also show that this method can be extended to injection wells.
Synthetic data were initially used to verify the method. Field data from several gas and oil wells, including a horizontal well and gas injector, were then used to demonstrate the method's application. In all cases, the new method shows good agreement with the results obtained from conventional methods, both in terms of AOFP and p.
Gas well deliverability testing traces its origin to the work of Rawlins and Schellhardt in 1936. This work presented the well-known empirical backpressure equation for analyzing conventional flow-after-flow test data. Further work showed that this equation could also be used to analyze isochronal and modified-isochronal data.
The traditional concept of coalbed methane production is one where the coal natural fracture system is initially 100% saturated with water and that this water must be produced to initiate gas production. This paper summarizes an investigation designed to reconcile measured relative permeability data with well test analysis results obtained during single-phase and multi-phase tests, and with reservoir simulation projections of gas and water deliverability as a function of bottom-hole pressure. Improved well test and relative permeability measurement procedures are summarized so that projections of future fluid deliverability made during the dewatering and early multi-phase production stages are more accurate.
A variety of well tests were performed that included water slug tests, water injection tests, gas injection tests, and multi-phase production and shut-in tests. Estimates of absolute permeability obtained from these data were variable depending upon the test procedures. In addition, multi-phase and gas injection test analyses were strongly influenced by the relative permeability data used during the analysis. Based upon newly measured relative permeability data and by history matching multi-phase production data, it was possible to reconcile some of the differences in estimates of absolute permeability that were obtained from each of the test types. Finally, a new field procedure is proposed and demonstrated to measure permeability in wells producing both water and gas.
The understanding and modeling of methane production from Coal bed Methane (CBM) wells has proven to be a particularly challenging task for the gas industry. Unlike conventional gas wells where the production is typically modeled as single phase gas flow, CBM wells require that water be produced to reduce lower reservoir pressure below the desorption pressure so that gas can be produced. A thorough understanding of relative permeability is necessary in order to predict gas and water production rates. Typically, most well tests are initially conducted when the reservoir is water saturated so that single phase analysis techniques can be used to evaluate permeability. Unfortunately, there are far too many cases where the initial measured water permeability did not predict the eventual gas flow rates.
In order to study all aspects of CBM production, the Gas Research Institute, in conjunction with Taurus Exploration, has operated the Rock Creek Methane from Multiple Coal Seams pilot production site at Rock Creek in the Black Warrior Basin for the last 10 years. Here the required well tests have been conducted and field data, including production data, has been collected in a controlled manner so an understanding of CBM could be developed.
In 1989, a survey of mathematical models which describe methane production from coal seams was presented at the SPE Joint Rocky Mountain Regional/Low Permeability Reservoirs Symposium in Denver, Colorado. This paper was later published in SPE Formation Evaluation Journal March 1991 issue. In this survey, 37 distinct models for coalbed methane production were reviewed in mathematical detail. Since that time, 15 additional mathematical approaches to forecasting reservoir performance, each with its own assumptions and limitations, have been developed. The objective of this paper is to discuss each of these recent models and to compare the assumptions, features, and limitations of these models with each other and with the original 37 models. Recommendations are also presented on the use of each model.
The original survey1,2 included models on the core scale, single-well scale, full-field scale, and basin scale and discussed models for coal seam production, coal mine emissions, and Devonian Shale production. These models were included in the survey because the physics and mathematics used in these applications are identical to those used in coal seam gas production.
In the original survey, the various models were categorized by the treatment of methane desorption. The categories used in the original survey included:equilibrium (instantaneous) desorption and non-equilibrium (time-dependent)desorption. These classifications will also be used in this paper. In addition, one of the surveyed models uses a compositional approach and will be considered in a category of its own.
The convention of the original survey, which will be continued in this update, was to use the preprint dates for the papers. This is because the papers discussed in this survey were obtained from technical journals originating from many different engineering fields (petroleum engineering, mining engineering, environmental engineering, and safety engineering). Due to the different time periods required for technical review in these journals, we felt that the original preprint dates would be more indicative of the chronology of the development of coalbed methane models than would the actual publication dates.
In addition, the system of nomenclature and units used in this survey is the same as those used in the original survey. This is because we feel a standardized set of nomenclature and units will aid in the comparison of the various models. Consequently, the equations discussed in this survey may appear slightly different from those presented in the original papers. The processes described with these equations, however, are consistent with those in the original work.