Coalbed methane production is initiated by the removal of formation water in quantities sufficient to accomplish a large reduction of fluid pressure in the cleat, system. Pressure reduction causes gas to be desorbed from the coal; the cleat system channels this gas to the wellbore. The rate and quantity of water removal is limited by permeability of the cleat system which is usually stimulated to facilitate dewatering and gas production.
Experiments were performed at IGT under in-situ conditions on coal from five U.S. basins. Results showed that water permeability of a given sample decreased by as much as two orders of magnitude over pressure ranges that, simulated normal dewatering and pressure ranges that, simulated normal dewatering and production cycles. production cycles. Knowledge of the behavior of permeability as a function of net stress allows a better understanding of the following realities of coalbed stimulation: 1) a horizontal drill hole of a given length is much more productive than a propped fracture of the same length; 2) in general, fractures induced in coalbeds are much shorter and wider than designed; and, 3) frac pressures in coalbeds are abnormally high pressures in coalbeds are abnormally high significantly above overburden in many cases.
Other consequences of a permeability-net stress functionality include: 1) skin factor determinations from injection tests are necessarily more optimistic than those based on drawdown data; 2) overbalanced drilling increases the invasion radius of drilling fluid filtrate, thereby creating the potential for more extensive formation damage than in conventional reservoirs; and, 3) leakoff during fracturing operations is significantly higher than for "conventional" sandstone or limestone beds having identical pre-injection permeabilities.
In 1983, IGT completed a laboratory study of drilling-fluid interactions with coal to characterize mechanisms of formation damage. To properly identify damaging agents, an understanding of "baseline" liquid permeabilities and functional changes with pressures permeabilities and functional changes with pressures which simulated normal phases of well development was required. Liquid permeability measurement were performed on samples from the San Juan, Piceance, and performed on samples from the San Juan, Piceance, and Appalachian Basins at simulated discovery, injection test and drawdown test pressures. Measurements were also conducted on samples from he Uinta and Warrior Basins, but only under discovery in-situ conditions Because the focus of this paper is the relationship between permeability and changes in net stress, results from these tests will not be included.
All liquid permeability measurements were performed with the apparatus depicted schematically in performed with the apparatus depicted schematically in Figure 1. Plug samples were placed between 1-1/2 inch (3.8 cm) endcaps of a standard Core Lab coreholder-Neoprene or Teflon sleeves were used to isolate plugs from the confining liquid (water) which generated equal triaxial pressures. Confining pressure was generated by a Ruska pump shown near the top of the diagram.
Fluid flow through the plug was maintained by a stepping motor, gear box, 9.0-CM3 positive-displacement, pump combination. The dynamic range of positive-displacement, pump combination. The dynamic range of the stepping motor was 0.2 to 22,000 steps per second, which corresponded to flow rates in the range of 2.2 x 10-7 to 2.4 x 10-2 cm3/sec. Three upstream and three downstream 1/8-inch (3.18 mm) diameter tubing loops permitted injection and withdrawal of different fluid permitted injection and withdrawal of different fluid types into the core with minimal cross contamination.
Two 0 to 15,000 psia (0 to 103 KPa) quartz crystal pressure transmitters with 0.01 psia (0.07 kPa) resolution were used to measure fluid pressures resolution were used to measured fluid pressures upstream and downstream of the core. Readings from these instruments were used to calculate the pressure drop across the sample during flow measurements.
Two 1-liter vessels on the downstream side of the core provided a constant pressure sink. About 5 00 cm3 of water was maintained in the bottom of these nitrogen-filled cylinders. The 9-cm3 pump volume added to these vessels raised the downstream pressure less than one percent.
A gas borehole flowmeter providing high resolution flow measurement is being developed to support natural gas production enhancement research within the United States Eastern Devonian Shale. The flowmeter utilizes ultrasonic technology which will provide excellent resolution and measurement accuracy. Early developmental test data indicate that the tool will be able to measure very low flow rates produced by borehole gas entry before fracture, i.e., .009 dm3/s (0.02 cfm). The use of time-of-flight dual transceivers will provide accurate gas velocity and flow measurements previously unavailable.
The Gas Research Institute (CRI) is sponsoring a program for the exploration and development of the United States Eastern Devonian Gas Shale via the application of advanced technologies. The gas shale development program focuses on the identification and characterization of fractures which control gas production. These efforts require the use of specialized instrument systems. One of the key instruments required is a borehole flowmeter designed to identify gas entry locations within a borehole and fracture zone. This instrument will also support test verification of stimulation technologies used in Devonian Shale.
The borehole flowmeter must locate and measure natural gas where it enters the side walls of the well at flow rates as low as .009 dm3/s (0.02 cfm) and must also define the total production characteristics of the well. The tool must production characteristics of the well. The tool must be easy to operate, reliable, and as durable as a standard wireline instrument. The downhole section of the tool will be designed to minimize flow restriction/interference, and must be small enough to pass through upper well casing and tubing.
The BDM Corporation is developing for GRI an ultrasonic borehole flowmeter for the specific purpose of analyzing and characterizing Devonian purpose of analyzing and characterizing Devonian Shale gas wells before and after fracturing. The borehole flowmeter will have commercial applications subsequent to stimulation technology research and development. The ultrasonic technology being applied by BDM to the borehole flowmeter promises to provide excellent resolution of gas flow. The ultrasonic measurements will also provide a direct indication of gas density and provide a direct indication of gas density and will determine the presence and location of turbulence, indicating pressurized gas entry in the borehole. With the use of bidirectional dual transceivers measuring differential time-of-flight, gas velocity can be determined independent of the local characteristics of the gas (temperature, pressure, and physical properties). Further, early test data indicate that velocity (or differential time) measurement capabilities are sufficient to allow logging at rates up to .23 m/s (40 fpm). The sonic flowmeter being developed by The BDM Corporation will be integrated into a well logging system using standard wireline equipment.
Sonic differential time-of-flight, static pressure, stagnation temperature, borehole pressure, stagnation temperature, borehole diameter, and tool location (depth) will be measured simultaneously by the borehole flowmeter. From these measured parameters, the gas velocity, density, volumetric flow, and the speed of sound within the gas media will be calculated. Gas dynamic analysis using these measured or calculated data may also provide additional information concerning the gas and borehole characteristics where desired. These data may be measured while logging from the top-down or bottom-up because of the bidirectional sonic measurement design.
Acoustic Time-of-Flight and Gas Velocity Measurements
The basic principle used to measure the gas velocity will be the time-of-flight of ultrasonic waves within the borehole. The measurement configuration is shown in figure 1. In effect, two ultrasonic transmitters and two receivers are oriented so that the signal flight paths are parallel to the borehole. parallel to the borehole. P. 195
The DOE Multi-Well Experiment (MWX) is a research oriented field laboratory whose objective is to improve the technology and provide additional insight into the processes required to develop provide additional insight into the processes required to develop low permeability lenticular gas sandstone reservoirs. The site of the MWX is located in the Rulison Field in the Piceance Basin of Colorado, as shown in Figure 1.
The MWX consists of three closely spaced wells drilled through the Marine Mesaverde Formation to depths of about 8,000 ft.
A detailed geological and geophysical assessment is a continuous effort being conducted throughout the Piceance Basin and other areas as a means to characterize the salient features of this site. Very extensive core and logging programs were carried out in all three MWX wells with emphasis on the zones of interest. Seismic surveys and sedimentological investigations were conducted to provide an assessment of morphology, areal extent of lenses and provide an assessment of morphology, areal extent of lenses and others. As a prelude to the intensive reservoir investigations that were scheduled for the lenticular sandstones, several well tests were performed in the blanket Cozzette sandstone. Some of the experiments conducted within the Cozzette included in situ stress testing, several short drawdown and buildup tests as well as a more sustained interference and pulse test that utilized two of the available MWX wells.
The marine blanket Cozzette sandstone characteristically exhibits gas permeabilities on the order of 10 microdarcies or less. However, the production and test data from the MWX site along with subsequent analytic methods require an average reservoir permeability on the order of 100's of microdarcies. Data obtained permeability on the order of 100's of microdarcies. Data obtained from the observation well during the interference test suggests that the pressure distribution from the production well was not radially symmetric and thus the reservoir was modeled as an anisotropic naturally fractured reservoir. Numerical modeling techniques including a dual porosity reservoir model were used to simulate this naturally fractured tight reservoir. Included in our analysis was a parametric sensitivity study for a number of salient well and reservoir parameters, such as wellbore damage a the effects of boundaries.
Several months of sustained winter pipeline production that followed well testing attests to the fact that the natural fracture system within the Cozzette extends over a large area.
This paper details some of the studies carried out in the Cozzette sandstone and includes core analysis, and the results of the MWX well testing.
INTRODUCTION AND BACKGROUND
At the MWX site the Mesaverde Formation is found to lie at a depth of between 4,000 and 8,300 ft. Outcrop studies indicate that most of the Upper Mesaverde Formation was deposited by meandering stream systems and includes a meandering fluvial system, that grades into more carbonaceous but gas-bearing paludal deposits. Three continuous marine blanket sandstones are paludal deposits. Three continuous marine blanket sandstones are at the base of the column, the Rollins which is water-saturated, Cozzette and Corcoran sandstone which are gas-bearing.
lsopachous contours in the Cozzette sandstone show that the interval thickens markedly eastward, and there is an east-northeast zone of thinning that must reflect an underlying anticline or sandstone buildup. The Cozzette lies between two thick deposits of marine Mancos shale and is separated into two zones by about 50 feet of mudstone and siltstone. The two zones were appropriately designated the Upper and Lower Cozzette. The well testing described herein was carried out in the more productive Upper Cozzette. The logs shown in Figure 2 detail the location of the Cozzette within a portion of the marine interval. The surface and sub-surface location of the wells, MWX-1 and MWX-2 are shown in Figure 3 along with the distance between wells at the depth of the Cozzette. (123 ft). During test periods, bottomhole pressure, bottomhole temperature, surface tubing pressure, casing pressure, fluid flow rates and other pertinent information were pressure, fluid flow rates and other pertinent information were acquired, manipulated and stored using a PDP-11/34 computer system. The field computer, interfacing hardware, accompanying calibration instrumentation and repair facility are all housed within the DOE/ Sandia Well Testing Facility located at the MWX site. Most of the subsequent analysis was performed on a Cyber 76 computer, DOE Nevada Operations and a VAX 730, CER Corp.
MWX COZZETTE CORE ANALYSIS
A total of 3,700 feet of 4 in. diameter core, including some oriented and pressurized were obtained from MWX-1 and MWX-2. The MWX cores were sealed at the site and shipped to several different laboratories for analysis. The following procedures were described by Randolph of IGT in performing their analysis.
When a continuous sand is bounded by zones of higher, but unequal, minimum in-situ stress, a vertically asymmetric hydraulic fracture results. The modeling is much more difficult than in the symmetric case mainly because the width equation is harder to formulate and solve. In this paper we present the principal components of the modeling, which includes principal components of the modeling, which includes non-Newtonian flow, leakoff with spurt loss, and "storage" of fluid due to volume expansion. The assumption is that the fracture is highly elongated, i.e., stress contrasts between pay and bounding zones are relatively large (>few hundred psi). Vertical gradients of minimum in-situ stress and fluid pressure can be included in the modeling. To illustrate. the results, we present design calculations for a 30,000 gallon fracture, which was the first stimulation in the Multi-Well Experiment. The 80 ft fracture interval in the Paludal zone has at its upper edge a 520 psi stress contrast, and at its lower edge a 1195 psi contrast. Computed fracture height growth above and below the perforated interval, bottomhole pressure, and width perforated interval, bottomhole pressure, and width profiles in vertical sections are displayed. profiles in vertical sections are displayed. Comparison is made with diagnostic measurements of fracture length, height, and bottomhole pressure.
At depths of a few thousand feet or more, induced hydraulic fractures will normally be vertical. Height growth containment is important so that the fracture will reach farther along the payzone, and so that the chance of vertical penetration into, for example, a water-bearing zone will be reduced. Although many factors influence height growth, the most important one appears to be the stress contrast between pay and bounding zones, where by stress we mean minimum in-situ stress. Here we study fracture height growth by developing a model for an expanding hydraulic fracture applicable when the fracture is highly elongated, with length along the payzone much greater than height. However, vertical variations in elastic parameters are not considered. The fracture shape in this paper is self-determined, in contrast to that in which an elliptical shape is chosens and the corresponding height or semi-minor axis determined. A variable-height fracture model has been intensively studied by Cleary and co-workers. The so-called "pseudo-3D" model treats the fluid flow as a dominant ID flow along the payzone, plus an auxiliary ID flow in the vertical direction. Although the models of Nolte and Palmer and Carroll take the vertical flow to be Palmer and Carroll take the vertical flow to be zero, thus simplifying the problem considerably, the general formulations are similar enough to Cleary's to be included under the rubric "pseudo-3D." In all these models, the fracture width is approximated by dividing the fracture into a number of vertical sections, and applying 2D elasticity theory to each vertical "line" crack. Thus the fracture is assumed to be highly elongated with length/height ratio >5. Finally, 3D modeling, with proper 2D fluid flow, is under development, but the problem is formidable and the computer run time enormous. In the interim we can learn much from pseudo-3D models. In general, the bounding layer stresses will not be equal, leading to a fracture which is vertically asymmetric, and furthermore both the minimum in-situ stress and the fluid pressure will vary with depth. This is the principal modification we make to the symmetric model, described previously. Other additions are: (i) spurt loss has been included in the leakoff, i) non-Newtonian flow is included. An extended model for the symmetric case, which has essentially the same components as herein, is described elsewhere. In that paper, a comparison is made between published results in three pseudo-3D models, some discrepancies are pointed out, and suggestions for reconciling the models are made. In the asymmetric model of this paper, calculation of fracture width is the most difficult task. We give most of the details here. Theoretical calculations of asymmetric fracture shapes have been reported by Settari and Cleary, but they appear to emphasize low stress contrasts (< couple hundred psi). Nolte gives one asymmetric width profile in a vertical section, but no method of calculation, nor any resultant fracture shapes, were given. Finally, to illustrate the results of the asymmetric model, we use the model to predict fracture height, pressure, and width for the first stimulation of the Multi-Well Experiment (MWX) carried out in December 1983. This prediction is compared with available fracture diagnostic measurements.
A model is presented for optimizing the number and locations of wells for various hydrologic and reservoir conditions. It can handle confined or unconfined aquifers, leaky or non-leaky aquifers, isotropic or anisotropic permeabilities, existing wells at fixed or model-determined optimal flow rates, complex boundaries, and specified regions to be excluded from possible well location sites for environmental or other reasons.
Examples demonstrate that optimum wellfield design differs significantly from the patterns widely used in the gas industry. optimum patterns depend heavily on the reservoir and hydrologic characteristics of the target, the shape of the project site, existing wells, the total number of project site, existing wells, the total number of wells, and water pressure drop in the coal seam required to commence gas flow.
Interest has grown recently in utilizing the vast resources of methane gas associated with deep coal seams. The majority of such methane is held in an adsorbed state on the surface of the coal pores by reservoir pressure; this pressure must be pores by reservoir pressure; this pressure must be reduced to allow desorption of methane from coal surfaces and subsequent methane production. The reservoir pressure is caused by an existing static pressure due to groundwater. Hence, unlike a pressure due to groundwater. Hence, unlike a conventional gas reservoir, gas production is obtained from coal seams by first depressurizing the coal seam, which can itself be classified as a low- permeability aquifer. To do so involves not only permeability aquifer. To do so involves not only reservoir engineering, but also hydrology. Both solve the same pressure or head diffusivity equation. However, the range of applications in each discipline is different. Reservoir engineers, for example, have concentrated more on single-well tests. The concept of skin is used routinely here but is virtually unknown in hydrology. Hydrologists have concentrated more on the use of observation wells, or interference tests. As a result, the range of available solutions is greater in the hydrologic literature for this purpose. The reason for the bias of each discipline is due primarily to the range in permeability and compressibility encountered. Permeabilities in oil and gas reservoir engineering tend to be lower, and compressibilities and well depths greater, forcing engineers to concentrate on the immediate vicinity of the wellbore; for economic reasons interference tests tend to take too long to be practical.
Hydrologists, on the other hand, find the range in conditions more suitable for observation well or interference testing. The production of gas from coal beds often places the engineer in the range of parameters between those which reservoir engineers and hydrologists are accustomed to. Effective exploitation of coalbed methane will involve a synthesis of both disciplines, and consequently the model we develop here will utilize solution techniques from both disciplines.
The parameters to be determined by the computer program are the number of wells, their locations, and individual pumping rates. In the past, hydrologists or engineers have usually attempted to optimize the number and locations of wells for depressurizing with a best guess based on experience, or sometimes using site-specific finite difference models.
This paper presents our progress to date on a computer program for automatically deter-mining the optimum number of wells and their best locations for dewatering/depressurizing for a wide variety of hydrologic conditions. This model can currently handle lease boundaries of complex shape, confined or unconfined aquifers, leaky or non-leaky aquifers, isotropic or anisotropic permeabilities. In the future the model will include vertical fractures and multi-phase flow. This versatility is achieved by using approximate analytical solutions and the method of superposition. The model includes routines to exclude specific regions as possible well location sites. The model can also possible well location sites. The model can also handle the effects of existing wells at a given fixed production rate or the optimum flow rate to be calculated by the model.
Laboratory measurements have shown that the permeability of coal samples varies significantly permeability of coal samples varies significantly for ranges of pressure encountered during the testing of coalbed methane wells.
A laboratory pressure-permeability relation has been used with a numerical, two-phase coalbed methane simulator to examine the effect of pressure-dependent permeability on well-tests and on production pressure-dependent permeability on well-tests and on production of coalbed methane.
Long-range transport of fluids through coal seams is generally thought to occur through the ubiquitous natural fracture network, which is comprised primarily of the butt and face cleat and bedding planes. The naturally fractures coal seam macrostructure can be characterized by an effective (cleat) porosity and permeability associated with the natural fracture network and by a low permeability coal matrix which has a large internal permeability coal matrix which has a large internal surface area containing physically adsorbed methane gas. The two-stage transport of fluids from the internal surfaces to a production well can be modeled by a variant of the now familiar dual-porosity models developed for naturally fractured reservoirs.
Laboratory measurements of coal samples have shown that the cleat permeability varies significantly with applied stress. In a recent study, analysis of laboratory data for samples from several coal seams provided quantitative evidence for a simple, general relation between normalized cleat permeability and normalized net confining pressure. pressure. In the present study, this general pressure-permeability relation based on the data or Rose pressure-permeability relation based on the data or Rose et al. is used with a numerical two-phase coalbed simulator to examine the effect of pressure-dependent permeability on single-phase well-tests pressure-dependent permeability on single-phase well-tests and on long-term production of water and gas.
The errors introduced by pressure-dependent permeability in the values of absolute cleat permeability in the values of absolute cleat permeability determined from conventional well-test permeability determined from conventional well-test analysis and the impact of these errors on predictions of long-term production are discussed in detail. A special test sequence and modified method of analysis of well-test data which eliminates these errors are presented.
COALBED METHANE NUMERICAL SIMULATOR
The numerical simulator used in this study is based on a finite-difference solution to the coupled two-phase Darcy flow equations, for water and gas through a porous, permeable medium. The porous medium is identified with the natural fracture network of the coal seam and assigned an effective absolute permeability, k o and effective porosity, methane is also considered to be porosity, methane is also considered to be present as a physically adsorbed phase on the internal present as a physically adsorbed phase on the internal surfaces of the coal matrix. The volume, V, of adsorbed gas is determined by a Langmuir adsorption isotherm, V = v L P/(P + P L ). Initially the volume of adsorbed gas is in equilibrium with the fluid pressure in the fracture porosity, P. The rate pressure in the fracture porosity, P. The rate of desorption of methane into the fracture porosity is assumes to be proportional to the difference between the actual and the equilibrium amounts of adsorbed gas and to a characteristic diffusion (or desorption) time, TL The model is mathematically, analogous to the so-called semi-steady model for naturally fracture, two-phase reservoirs.
The permeability and porosity were assumed to vary with fluid pressure according to
where phi o and k o are the cleat porosity and permeability at the initial coal seam fluid pressure, Po. Recent support for a relation of this form linking pressure and permeability has been reported by Rose et at, see Appendix A.
Most demethanation programs which use boreholes from the surface to access the coal, incorporate a stimulation program to enhance the well production. In most cases, the stimulation selected is a frac using nitrogen foam or other fluids. Novacorp has determined that fracing the coal may not be the best technique to apply to a well.
Novacorp, whose experience includes over eighteen (18) fracs on demethanation wells, has developed and patented an alternative stimulation method, called the patented an alternative stimulation method, called the Cavity Stress Relief (CSR) technique, which it believes is more effective than the frac, in most cases.
Through a review of the basic parameters which affect the desorbtion of gas from coal, a comparison is made between the frac and CSR stimulation techniques. A description of the conditions and results of two field tests of CSR are given. Also, the areas which need further development will be discussed. Although still not totally proven, the CSR stimulation technique promises to be a very effective method of stimulating most demethanation wells.
The hydraulic fracturing technique introduced in. the early 1950s was designed to enhance production of oil and gas wells by producing a large fracture in the reservoir which would act as a trunk pipework. This greatly increased the overall permeability of the reservoir by connecting the small fractures in the reservoir rock more directly to the well. Development of the technique since its introduction has been directed toward enhancing this effect and improving the control on the fracture size, length, direction, etc.
The production rate of oil and gas wells is limited only by permeability (i.e., if infinite permeability could be induced in the reservoir permeability could be induced in the reservoir containing the petroleum product, all the production could be obtained almost instantaneously). This not true of production from demethanation wells.
In the early work of the U.S. Bureau of Mines in demethanation, the need arose to improve well production. Hydraulic fracing was attempted because production. Hydraulic fracing was attempted because it was a convenient technique that was available. It solved the main problem that was thought to be the cause of low gas production, viz., the low permeability of the coal. The desired effect was permeability of the coal. The desired effect was obtained when the well was fraced, thus "proving" the validity of the theory.
It is our intent to show that the effect of hydraulic fracing is not that commonly thought to occur, and that once the actual mechanism that results in an improved gas production from a frac is understood, a better stimulation technique will be given as well as a review of the field experience gained by Novacorp Engineering Services Ltd.
NOVACORP'S EXPERIENCE WITH HYDRAULIC FRACS
Since 1977, Novacorp has carried out over eighteen hydraulic fracs with numerous variations in size, depth of application, injected fluid, presence or absence of propping agents, etc., to optimize its version of demethanation wells which it calls VCD for Virgin Coal Demethanation. The results of these fracs were analyzed and it was found that the effect of the fracs was generally greater than could be ascribed to an increase in permeability, even assuming the long fracture predicted by frac companies. Using hydro geologic techniques such as pump tests with observation wells, a significant amount of work was done to determine the permeability increase resulting from a frac.
Tailored-pulse loading is a gas well stimulation technique that is promising for particular situations. The key factor in determining the probability of success is the interaction of the loading probability of success is the interaction of the loading history with the material properties and in situ stresses of the geologic formation. On the basis of theoretical and experimental studies, tailored-pulse loading is expected to be successful if (a) the rock is fairly competent, (b) the yield stress is greater than about 100 MPa, (c) the overburden pressure is less than about 100 MPa, (d) the natural fracture or joint spacing is less than several meters.
The use of propellants or low-energy release rate explosives to stimulate oil or gas wells has been studied extensively during the past several years in programs funded by the Morgantown Energy Technology programs funded by the Morgantown Energy Technology Center (METC), other agencies of the Department of Energy, and the Gas Research Institute (GRI). The postulated advantages of the technique include the postulated advantages of the technique include the ability to create radial cracks at the borehole wall without causing rubblization and to drive the cracks by internal gas pressurization into the formation with their orientation unaffected by in situ stresses. During a recent GRI-funded program, Sandia National Laboratories Albuquerque (SNLA) demonstrated the practical feasibility of the tailored-pulse loading (TPL) technique in well is at Rowan County, Kentucky, and Meigs County, Ohio.
The purpose of this paper is to discuss the expected limitations of the TPL technique and to describe the laboratory tests required to both properly assess the potential of a given formation for TPL stimulation and to roughly predict the extent of permeability enhancement that is expected. To accomplish permeability enhancement that is expected. To accomplish this objective, we use the results of a three-year METC-funded program at SRI. During that program, SRI worked with SNLA, Science Applications, Inc., the University of Maryland, and Los Alamos National Laboratories (LANL) to perform laboratory experiments, to develop computer models of the fracture process, and to use the models to predict the process, and to use the models to predict the results of field experiments performed at the Nevada Test Site.
LIMITATIONS OF THE TPL TECHNIQUE
The scenario for a TPL stimulation is illustrated in Fig. 1. If the rise time of the pulse produced at the borehole wall is shorter than the time required for a sound wave to run around the borehole circumference, then multiple cracks can be produced by the circumferential tensions. However, several other requirements restrict the applicability of the technique. First, the circumferential tensions must be greater than the tensile strength of the rock. This requirement is easy to meet for moderate overburden pressures. However, two additional requirements are pressures. However, two additional requirements are that the peak radial compressive stress be lower than the compaction strength of the rock if it is porous and that the radial compressive stress also porous and that the radial compressive stress also be lower than that required to cause the rock to yield. If the above requirements are not met, the compaction and/or flow around the borehole will form a "stress cage" that tends to seal off the borehole from the fractured formation.
Although it is relatively easy to produce TPLs that fulfill one or two of the above requirements, it may in some formations prove difficult to fulfill all of them simultaneously. Fortunately, in Devonian shale the porosity is low, the tensile strength is not excessive, the yield stress is relatively high, and SNLA was able to successfully create radial cracks in the formations in the Rowan and Meigs County stimulations where the overburden pressures were not excessive (less than 40 MPa).
Another requirement for successful application of TPL is that the radial cracks be driven far enough by the gas pressurization to intersect the natural fractures in the formation. The importance of gas penetration into the cracks is illustrated in Figs. penetration into the cracks is illustrated in Figs. 2-4, which show the results of laboratory experiments in Plexiglass blocks containing scale model boreholes loaded by a low-power explosive. In one case the borehole was sheathed to prevent gas penetration, whereas in the other case the explosive penetration, whereas in the other case the explosive gases were allowed free access to the radial cracks. The gases drove the cracks 5 to 10 times farther than did the stress waves alone.
A new log interpretation system called TITEGAS has been developed to analyze the reservoir characteristics and producibility of tight gas sands. The system is based upon equations which define the response of conventional logging tools. A basic log suite of density, neutron, and resistivity are the only logs required by the system, however TITEGAS also allows for utilization of various other logs including photoelectric effect, spectral natural gamma, micro-resistivity, microwave travel time, dielectric constant anacoustic travel time.
The new system includes several innovative analytical procedures:
A technique to compute porosity in zones having variable gas saturation and clay content.
A new method to quantify gas saturation in the zone investigated by the density and neutron tools.
An improved method using the gamma ray to determine clay content in compacted sands composed predominately of illite-type clays.
An improved method to determine clay content from the sonic and density log when illite is not the dominantclay type.
A new non-geometric mathematical model using neutron and density logs to calculate clay content. This equation makes it possible to refine the clay log response constants.
A new technique for the interpretation of formation water resistivity. This procedure is useful in thick sand-shale sequences where water salinity is variable.
New methods to refine matrix density of the reservoir. Matrix is dealt with as a foot-by-foot variable rather than as a constant.
A new technique to analyze invasion profile and qualitative formation permeability.
New methods to detect natural fractures through the modeling of logs that are generally available.
The TITEGAS system is applied to the DOE/Sandia MWX wells. Comparisons are made between the TITEGAS system results and the high-quality core data and well test data from these wells. The comparisons indicate that the system is a powerful new procedure for the analysis of tight gas sands. procedure for the analysis of tight gas sands
Operators who try their hand at tight gas sand well completions generally find themselves involved in a "learning experience". Well performance is inconsistent with log analysis prediction. performance is inconsistent with log analysis prediction. During the past four years, a new computer log analysis system has been developed to solve the logging problems characteristic of the tight gas sand (TGS) resource. The "TITEGAS" system has been applied successfully to TGS sequences in Texas (Cotton Valley and Canyon Sands), New Mexico (San Juan Basin), Colorado (Piceance Basin), Wyoming (Green River Basin), Utah (Uinta Basin), and Ohio (Appalachian Basin).
The objective of this paper is to describe the TITEGAS log analysis system. Several major points are discussed:
* New analytical procedures;
* The computational process - the criteria for interval selection, logic for refinement of constants, and the mechanism for human interaction with a modular computer log analysis system;
* Typical output presentations using field examples;
* Applications of the system for improved reservoir characterization and for geological studies.
THE BASIS FOR THE MODEL
There are many physical reasons why conventional log analysis methods are not able to predict which tight gas sands are capable of gas production.
Porosity, gas saturation, and permeability interpretation using Porosity, gas saturation, and permeability interpretation using conventional techniques is not adequate, even if great care is taken.
Many operators do not routinely run prestimulation pressure buildup tests in prestimulation pressure buildup tests in low permeability gas reservoirs. Frequently, all that is available is a single-rate flow test. To obtain permeability estimates for FERC filings and to evaluate the reservoir, a new method has been developed for analyzing typical prefracture production data.
Permeability estimates obtained from this method have been used as a major source of data in numerous successful tight gas determinations, including the Cotton Valley, Wilcox Lobo, and Canyon formations in Texas.
The paper shows that this simple method can be quite reliable. This conclusion is based on a comparison of permeabilities estimated with this technique and with more rigorous determinations from pressure buildup tests.
Accurate reservoir description is necessary to develop tight gas reservoirs economically. A critical formation property
is in-situ permeability. An accurate estimate of formation permeability is needed for fracture design calculations and for predicting future production as a function predicting future production as a function of time. Permeability estimates are also needed to determine if a reservoir qualifies for FERC's "tight gas formation" classification.
In many cases, operators do not run pressure buildup tests on low permeability pressure buildup tests on low permeability gas wells prior to hydraulic fracture stimulation. Frequently a well is perforated, broken down and produced for only a perforated, broken down and produced for only a few days prior to performing a massive hydraulic fracture treatment. Chokes are changed frequently in this testing program and there are often shut-in periods of varying lengths between flow periods. This pre-fracture flow data cannot be analyzed pre-fracture flow data cannot be analyzed using conventional pressure transient analysis techniques; therefore, pre-stimulation permeability estimates are not pre-stimulation permeability estimates are not available. To compound the problem, post- fracture pressure buildup tests leading to unambiguous interpretations are often prohibitively time consuming. Thus, an prohibitively time consuming. Thus, an alternate method to estimate formation permeability, even if it is less accurate permeability, even if it is less accurate than the buildup tests, is needed to evaluate tight gas reservoirs using available pre-fracture flow data. pre-fracture flow data. This paper presents an approximate method (ONEPT) for determining formation permeability from single-point flow data. permeability from single-point flow data. This new method uses conventional transient flow equations in situations where only the initial formation pressure, the final flowing pressure, the cumulative gas production, and the final flow rate are known. production, and the final flow rate are known. The ONEPT method is appealing because, in contrast to pressure buildup data, which are often not available, single-point flow data are nearly always available on gas wells. This paper describes the ONEPT method and documents its validity in two field studies in which it is compared to the results of pressure buildup tests. In addition, the pressure buildup tests. In addition, the paper shows the effect on permeability paper shows the effect on permeability estimates of wellbore storage, variable rate history (including shut-in periods), and uncertainties in skin factor. These effects were determined using finite-difference reservoir simulation.