Study quantifies Powder River coal seam properties

Matt Mavor
Tesseract Corp.
Park City, Utah

Tim Pratt
Ticora Geosciences Inc.
Denver

Roland DeBruyn
Redstone Resources Inc.
Denver

Gas Research Institute (GRI) is funding an ongoing study to understand why operators have been so successful at producing gas from Powder River basin, Wyoming, coal gas reservoirs.

This article summarizes results from the study aimed at quantifying coal seam properties.

In the early 1990s, conventional wisdom concluded that large-scale, commercial natural gas production from these coals was not possible. The coals were recognized as high-permeability aquifers that could not be sufficiently "dewatered" to produce sorbed natural gas.

Conventional wisdom was wrong, as has so often been the case when dealing with coal seams.

Gas can be released from the coal matrix in commercial quantities even when gas content is low and permeability is high. This fact has been proven by numerous Powder River basin operators over the past 6 years who are developing the basin's margins.

Powder River coal gas

The operators have concentrated development along the basin's margins in thick coals typically at a 250-1,000 ft depth. In these areas, the subbituminous Powder River coal possesses characteristics that are dramatically different from the commercially developed areas in the San Juan, Warrior, and other basins.

In particular, the Powder River coal has:

• High water content
• Low gas content
• Low ash content
• High absolute permeability, ranging from tens of millidarcies to more than a Darcy.

Because coal-gas reservoir data are scarce, GRI's purpose is to obtain quantitative estimates of the properties of specific coal seams. Redstone Resources Inc. allowed GRI access to Well Triton Federal No. 21C-2623 in Section 26, Township 52n, Range 73w, Campbell County, Wyo. This well was cored and logged and is being tested during the first part of 1999.

Redstone has also provided data from Well Twenty Mile No. 21C-3523, in Section 21, roughly 1 mile south of the Triton well. Both wells are on the eastern margin of the basin (Fig. 1) [223,427 bytes]. This is just west of large strip mines located near the coal outcrop.

The specific target was the Canyon coal seam that belongs to the Tongue River member (late Paleocene) of the Fort Union formation. The Canyon coal seam is shallow in both wells: 557-621 ft in the Triton well, and 530-594 ft in the Twenty Mile well. Tests indicated a gas productivity of 30 Mscfd in the Triton well and about 300 Mscfd in the Twenty Mile well.

Activity levels

The Powder River basin has recently been a hotbed of coal gas activity. Operators are excited about the potential gas rates that can average 100 Mscfd or more from shallow wells that cost less than $50,000 to drill and complete.

Although water production rates can be high, Wyoming allows the very fresh produced water to be discharged into ponds and drainages, greatly improving production economics.

During the 1980s, several coal-gas pilot projects were undertaken in thick, deep coals near the center of the basin. These projects produced large quantities of water and not much methane. The projects did indicate that the coals had high permeability.

In late 1989, coalbed gas production began in the Rawhide field, near several large surface coal mines. Betop Inc. and Martens & Peck Production Co. were the principal producing companies. This effort had spotty success but served as a valuable testing ground for drilling and completion methods.

Torch Operating Co. and Lance Oil & Gas Co. Inc. now operate these wells. Lance is an affiliate of Western Gas Resources Inc., a major gas-pipeline company in the area.

Martens & Peck's Marquiss project, south of Gillette, was the first large-scale coalbed methane development. The company drilled about 60 wells, with first production starting in 1993.

Flow line and pipeline constraints restricted early production rates, but these have since been resolved. Torch now operates these wells.

In 1994 and 1995, Martens & Peck expanded development into the Lighthouse and Maysdorf areas, south and southeast of Marquiss. Lance, Barrett Resources Corp., MTG Operating Co., and other companies have extended the play further south, and recently downdip to the west. Pennaco Energy Inc. and others are expanding development north from Marquiss to Gillette.

Redstone Resources Inc. began drilling in its Rough Draw project in 1995, north of the Rawhide mine. First production started in early 1997.

In this general area, Redstone, Devon Energy Corp., Kennedy Oil, Lance, and others are continuing to drill. Further north, near the Recluse and Spotted Horse fields, drilling by Pennaco, Lance, and Wiepking-Fullerton Energy LLC is under way with more activity planned.

On the western and northwestern sides of the basin, north and south of Sheridan, drilling is under way on large leaseholds controlled by Sonat Exploration Co., J.M. Huber Corp., and two Redstone affiliates, Redstone Gas Partners and Preston, Reynolds & Co.

In the north-central basin, Pennaco and CMS Energy Inc. are planning for a very large-scale development along the Wyoming-Montana state line.

Accurate production statistics are difficult to acquire partly because of the intense activity levels. Much of the data have yet to reach the public domain.

A commercial data base as of Jan. 31, 1999, indicates that 626 wells are producing gas from the Fort Union coal intervals in Campbell County. Total production was about 87 MMscfd of gas and 205,000 bw/d. These volumes correspond to an average 140 Mscfd of gas and 330 bw/d per well.

Discussions with operators suggest that the data base production estimates are low. Recent gas production probably exceeds 100 MMscfd.

Well completions

To date, operators have focused on single coal zone completions because drilling twin wells costs less than a single well bore completed open hole in multiple seams.

Air, air-mist, and water drilling with open-hole completion methods are used because the coal is very susceptible to damage by conventional drilling fluids. Casing cemented over the coal seam and perforated usually results in unsuccessful completions.

The wells are drilled with Gardner Denver 1500 style, truck-mounted water well rigs. A typical well has a 97/8-in. hole, drilled with water, to the top of the target coal interval. A string of 7-in. casing is run and cemented in the 97/8-in. hole.

The target coal interval is drilled with a 61/4-in. bit using air, air-mist, or water. The coal interval is often underreamed to 10 or 12 in., cleaned out, and filled with water. The rig is moved off location after a 7-in. wellhead is installed. These procedures normally take 2 days to complete.

Historically, well site facilities have consisted only of an insulated enclosure for the wellhead and flow line connections plus a control box for a downhole pump.

A service rig runs tubing and a submersible pump. The production tubing is either plastic, fiber glass, or steel, depending on the pump setting depth, pump size, or horsepower required.

Wells shallower than 1,000 ft are equipped with electric submersible pumps that lift water to surface and then onto surface discharge points. Water well pumps are economical and have proven suitable for this service. The pumps, however, are sensitive to gas locking and formation fines. They work best if speeds are regulated from the surface in response to fluid level.

After wellhead valves, surface controls, and piping are installed, the well is placed on production for 1-2 months to clean out the near well bore region, prior to hydraulic fracturing.

Gas is produced up the casing-tubing annulus and gathered at a central battery where it is metered, compressed, and sent to the main terminal station. Large-diameter flow lines minimize pressure drop.

Rotary screw compressors have proven to be efficient. Compressor suction pressures are typically 5 psig or less.

At the terminal, gas is dehydrated, treated to remove CO₂, and compressed into sales lines. Produced water is routed through collection lines separate from the gas lines and then discharged at various locations throughout the area.

Many ranchers welcome this water supply because the region has an arid climate.

Stimulation improves the productivity of the open hole completions. Early experience with complex fracturing fluids was usually disappointing and the operators have turned to moderately high-rate water injection without proppant.

A typical job consists of pumping 500 bbl of city water down the casing at a 30-40 bbl/min rate with a 130-psig surface injection pressure. Dramatic pressure breaks are rarely observed during the 10-12 min injection.

After stimulation, production tubing with the submersible pump is rerun, and the well is placed back on production.

Formation evaluation

Formation evaluation quantifies estimates of key reservoir properties such as:

• Gas content
• Gas storage capacity
• Gross and net reservoir thickness
• Gas-in-place volume
• Natural fracture system permeability
• Reservoir pressure.

The data required for these estimates are obtained from core measurements, open hole logs, and well tests,1-4 and are used to forecast future production rates and gas reserves.

GRI's data collection effort has two major goals. The first is to evaluate gas storage characteristics so that gas-in-place volume can be estimated from a combination of core and log data.

The second is to measure well-test pressure and production data from which the average reservoir pressure and effective permeability to gas and water can be evaluated.

One goal of well testing is to estimate average reservoir pressure as a function of cumulative production. Interpretation of these data allows one to estimate the drainage volume contributing to production for a specific well.

The well testing has only recently started and results are not yet available.

Coal reservoirs

Coal-gas reservoirs commonly contain two porosity systems. One system has pores of less than or equal to 50 nanometers (nm) in diameter and the other is composed of pores greater than 50 nm in diameter, as well as natural fractures.

The primary porosity system, which consists of microporosity (less than 2 nm) and macroporosity (between 2 nm and 50 nm) contains most of the gas. Gas is stored predominantly by sorption.

Sorption is caused by a net attraction between the large surface area of the small pores and the gas molecules. The attraction pulls the molecules close together, resulting in a gas density greater than would be expected from the reservoir pressure.

Mass transfer through the primary porosity is by diffusion and driven by concentration gradients.

Canister desorption measurements determined the sorbed-gas content of the primary porosity. Sorption isotherm data quantified methane and carbon dioxide storage capacities.

At the data collection locations, in situ gas content and storage capacity were very similar. This indicated that the primary porosity system was nearly saturated; that is, it contained almost all the gas that it could hold. This condition may not be true elsewhere, however.

The secondary porosity system consists of pores greater than 50 nm in diameter and natural fractures. Most flow towards wells occurs through this system. Compression dominates the gas storage. Mass transfer is primarily driven by pressure gradients and described by Darcy's Law. If coal seams are isolated from surrounding aquifers, water production originates from within secondary porosity.

Measurement modifications

The greatest problems in obtaining core measurements were to ensure that the coal did not react with oxygen and that the moisture content of the coal was preserved.

For this study, these problems were solved by storing samples in produced formation water that had been deoxygenated.

Some cores were cut into 1-ft long samples and inserted into tightly fitting plastic sleeves. The sleeved samples were then placed into desorption canisters.

For bituminous coal samples, canisters are sealed with a significant void volume (referred to as headspace) remaining inside the canister. This could not be done for subbituminous samples because the air remaining in the headspace would have reacted with the coal and the water content would have evaporated. Measurement errors would have been large because the desorbed-gas volume could have been less than the headspace volume.

To solve this problem, the canisters were filled with produced coal water that had been purged with argon and preheated to the 60-65° F. reservoir temperature. The water completely covered the core samples, preserving the moisture content and reducing the headspace to the air volume within the canister cap. This procedure also minimized air con tamination in gas samples taken for sorbed-gas composition measurements.

Sorption isotherm measurements can suffer from problems caused by coal oxidation and moisture loss. The sorbed-gas volumes are small, necessitating precise calibration of the isotherm apparatus and the need to minimize dead space volumes. Some isotherm measurement problems have been solved because gas content estimates agree with measured storage capacity data.

Once the apparatus was properly calibrated, the greatest error introduced into the sorption isotherm measurements would have been caused by coal oxidation during sample preparation.

Two preparation approaches were used. In the more rigorous procedure, all samples were crushed and handled inside a glove box filled with argon. This worked well but was time consuming.

The second approach was rapidly to crush samples in air and transfer them quickly into the measurement apparatus. This approach saved time and did not result in significant oxidation.

Cores, logs

At the Triton well, Baker Hughes Inteq cut two 31/2-in. diameter, 30-ft cores with a 67/32-in., RC 315, face-discharge bit. Produced water was the drilling fluid.

The core was collected inside conventional core barrels that included a PVC liner. Core recovery was excellent, between 97 and 100%.

Five 1-ft samples from the first core and 20 from the second core were placed in gas desorption canisters maintained at 65° F. The remaining core was preserved under water.

Desorption measurements were continued until desorbed volumes were minimal, at which time the samples were crushed to determine the residual gas content. Lost gas contents were estimated with a variation of the Direct Method.^1-4 The lost, measured, and residual gas contents were added to obtain the total sorbed-gas content.

Gas samples were collected at various stages of desorption to determine the sorbed-gas composition.

Proximate analyses on all desorbed samples determined ash, moisture, and sulfur content. Helium pycnometry measurements determined crushed density.

Sorption isotherm data were measured on one composite sample. Three composite samples from the upper, middle, and lower intervals were prepared for full proximate analysis, ultimate analysis, calorific value, maceral composition, and vitrinite reflectance measurements.

Table 1 [41,889 bytes] summarizes the measurements for the two wells. Some measurements are still in progress.

The thermal maturity data indicate that the coal rank is subbituminous C. Organic composition is predominantly vitrinite (72-78%) with a large inertinite component (19-24%).

Coal moisture content is high (27.5-27.9%) as expected for this rank. Ash content is low (less than 4%) and quite uniform in the center of the seams. In situ coal density, including the primary porosity system, is 1.31-1.33 g/cc.

Both reservoirs had similar in situ sorbed-gas content (19.2-21.2 scf/ton). Sorbed-gas composition was dominated by methane (89.5-89.9%) with CO₂ (8.2-8.5%), some nitrogen (1.5-2.3%), and traces of other hydrocarbons. Produced gas should be roughly 99% methane and 1% CO₂, based on isotherm data.

Calculations indicate that gas storage capacity was within 2.2-3.8 scf/ton of the gas content estimates. The differences are likely within the measurement error of gas content and storage capacity.

Both wells produced gas quickly during initial tests, indicating that the coal is nearly saturated. This is in agreement with the gas content and isotherm data.

Open hole log readings obtained included high-resolution array induction resistivity, neutron porosity, density, gamma ray, and caliper data. Fig. 2 [117,160 bytes] illustrates caliper, gamma ray, density, neutron, and microresistivity readings.

Most available permeability estimates have been obtained from slug tests. These tests involve rapidly filling the well bore with water and measuring the water level vs. time.

From the decline rate, one can estimate the effective permeability to water, which for these wells ranged from 100 md to more than 1 Darcy.

Production performance

Significant production history exists for some multiwell projects developed with the drilling and completion techniques described previously.

For instance, in the heart of the Marquiss and Lighthouse areas in Townships 47 and 48n, Range 72w, 136 wells have been producing for more than 1 year. The average gas recovered by March 1999 from these wells is 231 MMscf/well. This is greater than the sorbed gas-in-place estimates for 80-acre drainage areas (considered the optimum spacing by major operators) based on the Triton well gas content and isotherm data.

Almost all wells in the 136-well group continue to produce, some at increasing rates over time. One of the best Marquiss wells, Well Lynde Trust 4-23 operated by Torch, has produced almost 903 MMscf, as of Dec. 1, 1998. It currently produces 500 Mscfd.

Such performance cannot easily be explained by sorbed-gas content alone, unless drainage areas are hundreds of acres, which exceeds development well spacing.

Redstone's production performance evaluation indicates that total gas content must be several times greater than the sorbed-gas content.

Other authors also have included greater gas content in assessing production potential, without explaining their justification.^5-7 These analyses are based on production performance but suffer from a lack of specific data concerning reservoir properties and storage mechanisms.

Potential sources of additional gas are free gas within the coal matrix and natural fractures, dissolved gas, gas migration through the coal reservoirs and sandstones, gas produced from bounding shales, or other factors not yet identified.

The ongoing GRI efforts are hoped to improve the understanding and data availability concerning the gas storage and production characteristics to improve operators' future development decisions.

One possible explanation of the discrepancy between gas content estimates and the production behavior is that the gas desorption measurements are failing to measure all gas contained within the coal seam. The total gas volume includes the sorbed-gas in the primary porosity, the free gas in the secondary porosity, and the gas dissolved in water in the secondary porosity.

Coal desorption measurements are believed to exclude gas contained within the high-permeability secondary porosity system because this gas is most likely flushed by water while coring.

At the Triton well, the sorbed-gas content is 21.2 scf/ton. Free gas may be contained in secondary porosity. Based on density log and coal density data, the secondary porosity may approach 6%. The gas saturation within the secondary system may be in the range of 35%.

At a gas saturation of 35%, a porosity of 6%, and a reservoir pressure of 150 psia, the contribution of free gas to the total gas content is 5.2 scf/ton. The contribution due to solution gas is 0.6 scf/ton. The sum of these may be the total gas content, 27.0 scf/ton. Free and solution gas displacement out of the core may have caused the total gas content to be underestimated by 21% or 5.8 scf/ton.

Plans have been made to measure secondary system porosity in cores to improve the density-log interpretation. The core secondary porosity possibly could be greater than 6%, causing errors exceeding 21% in the total gas content estimates.

Gas content performance discrepancies may be caused by migration. Reservoir simulation models were built to evaluate gas-migration potential.

These models are considered conceptual because very limited data were available near the production wells included in the models. The models were constructed from maps of the Canyon coal seam and included depth and thickness variation.

The models predicted that the observed gas production rates could be achieved at the sorbed-gas content level (21.2 scf/ton) if two essential items were included in the models.

First, free gas had to be present in the secondary porosity when the wells were placed on production to match actual gas and water rates.

Second, the modeled wells had to drain a much larger area than the well spacing; otherwise, the simulated rates would have been too low. The average drainage area in the model corresponded to 134 acres/well, rather than the existing drilled spacing.

One explanation for the larger drainage area is that as production reduced pressure in updip areas, downdip pressure was also reduced, thus allowing downdip gas to be released and migrate updip to the producing wells.

The data required to resolve the production volume-gas content issue can be obtained from well tests, if the tests are properly conducted. The tests require modern pressure transducers and accurate gas and water production-rate measurements.

Analysis of the data gives estimates of average drainage region pressure, effective permeability to gas and water, absolute permeability, and degree of near-well permeability alteration. The gas volume contributing to production can be determined by material balance analysis of the decline in average reservoir pressure vs. cumulative gas production.

Progress has been made in obtaining quantified estimates of reservoir properties and in beginning to understand the production behavior.

If operators implement careful data collection plans, the "mysteries" of subbituminous coal-gas production will be solved in a straightforward manner.

The data must be carefully collected because the pressure and gas content are low. Small errors in reservoir property estimates can greatly change data interpretations and predictions of future performance.

Acknowledgments

GRI funding under the supervision of Charles Nelson was responsible for the data collection and interpretation efforts. John Robinson of Tesseract performed the simulation work.

References

1. Mavor, M.J., and Nelson, C.R., "Coalbed Reservoir Gas-In-Place Analysis," Gas Research Institute Report No. GRI-97/0263, Chicago, October 1997.
2. Saulsberry, J.L., Schafer, P.S., and Schraufnagel, R.A. (editors), A Guide to Coalbed Methane Reservoir Engineering, Gas Research Institute Report GRI-94/0397, Chicago, March 1996.
3. Mavor, M.J., Close, J.C., and McBane, R.A., "Formation Evaluation of Exploration Coalbed Methane Wells," SPE Formation Evaluation, December 1994, pp. 285-94.
4. McLennan, J.D., Schafer, P.S., and Pratt, T.J., A Guide to Determining Coalbed Gas Content, GRI Reference No. GRI-94/0396, Gas Research Institute, Chicago, 1995.
5. Hobbs, G.W, IV, "Coalbed Development in the Powder," Western Oil World, August 1990, pp. 17-20.
6. Holland, J.R., and Kimmons, J.W., "Characteristics and Economics of Coalbed Methane Production from the Fort Union Formation, Powder River Basin," Paper No. 9549, Intergas '95, The University of Alabama, Tuscaloosa, Ala., May 1995, pp. 115-23
7. Schwochow, S.D., "Coalbed Methane-State of the Industry, Powder River Basin, Wyoming and Montana," Quarterly Review of Methane from Coal Seams Technology, Vol. 11, No. 1. Gas Research Institute, Chicago, August 1993, pp. 28-32.

The Authors