WATER FRACS OUTPERFORM GEL FRACS IN COALBED PILOT

Ian D. Palmer
Amoco Production Co.
Tulsa

R. T. Fryar, Kelly A. Tumino
Amoco Production Co.
Houston

Rajen Puri
Amoco Production Co.
Denver

Initial results from over 12 months work indicate that water fracture treatments outperform gel fracture treatments in gas production from coalbeds.

The water fracs are also less expensive-only about half the cost.

The coalbed methane pilot project is being conducted by Amoco Production Co. in the Black Warrior basin of Alabama.

The well completion method consists of installing casing, with perforations into the individual coal seams. The Black Creek series is stimulated first, followed by separate stimulation of the Mary Lee/Blue Creek seams.

The borate cross-linked-gel fracture treatments are designed to carry high concentrations of proppant (10 ppg). In contrast, water fracture treatments carry proppant concentrations to only a few pounds per gallon.

In both cases 12/20 mesh sand is used. About 20 wells, half of them stimulated with water and half with gel, are interspersed in the Oak Grove field to nullify geological variability.

OAK GROVE FIELD

Amoco has been developing the coalbed methane resource in the Oak Grove field of the Black Warrior basin for over 2 years, and several hundred wells have been drilled, completed, and stimulated.

Fig. 1 shows a sketch of the Mary Lee/Blue Creek zone and the Black Creek coal zone. Depths are fairly shallow, with the Black Creek zone mostly less than 2,000 ft.

Amoco's standard completion has been through casing and perforations in the individual coals. The Black Creek zone is perforated and fracture-stimulated before a bridge plug is inserted to isolate this zone from the Mary Lee/Blue Creek, which is then perforated and fracture-stimulated.

Permeability of the coals appears to lie in the 5-20 md range. Until recently, the fracture stimulations have been of two kinds:

1. Gel fractures, which use 30 lb/1,000 gal borate cross-linked gelled fluids carrying 12/20 sand up to concentrations of 10 ppg

2. Water fractures, which use plain water as fracturing fluid, with sand concentrations up to about 5 ppg.

A separate study of gel fractures, to be released later, shows that generally the Black Creek fractures are vertical, with substantial height growth, and are characterized by low fracture propagation pressures.

The coalbeds are generally below 1,200 ft. In the Mary Lee/Blue Creek coalbeds (800-1,200 ft), approximately half of the gel-fracture treatments are just like the Black Creek fractures, and they are interpreted similarly.

The other half are different and exhibit high fracture propagation pressures. They are probably T-shaped fractures, with a T-fracture confined to a coal seam.

HYDRAULIC FRACTURING

The history of hydraulic fracturing has been a search for viscous fracturing fluids which would carry higher concentrations of proppant and place them in the formation so that more of the productive pay height is propped.

However, the use of such viscous fluids, especially crosslinked gels, always damages the formation permeability to some extent (due to invasion of fracture fluids into the formation, or to filter cakes forming on the walls of the fracture).

In general, production impairment by this damage is compensated for by the increase in production due to the enlargement of the effective well bore radius. However, a recent article has argued that the damage to a coal formation by gelled fracturing fluids may be a lot worse than in competent formations.1

Damage to coal permeability appears to be caused by sorption-induced swelling of the coal matrix and possibly some plugging of the cleat system by residual gel.

Assuming that gelled fluids result in better proppant placement over the coal seams intercepted by the fracture, when compared with water fracturing, this project is actually a test of whether the loss in production due to permeability damage is compensated for by the gain in production due to better propping of coal seams (more seams propped, and better fracture conductivity).

To understand better which seams will be propped by the different fracture treatments, and what the resultant fracture conductivity will be, some fracture modeling is discussed here.

PILOT PROJECT

To better compare the production from water fractures vs. that from gel fractures, a 23-well pilot was designed in the western portion of the Oak Grove field (Fig. 2). Thirteen wells were gel fractured, and 10 were water fractured.

The wells were interspersed to rule out geological variability, which otherwise could skew a comparison such as this. The Mary Lee/Blue Creek and Black Creek coalbeds typically lie in the 1,600-2,200 ft interval. In each case, two treatments were performed per well.

The gel-fractured wells used 30 lb/1,000 gal crosslinked borate gel with around 100,000 lb/zone of 12/20 sand (up to 10 ppg) pumped at 40 bbl/min.

The water fracture treatments consisted of pumping around 70,000 lb of 12/20 sand per zone (up to 4 ppg) at 50-60 bbl/min (Table 1).

Variations were made to the water fracture design. Sand concentrations as high as 6 ppg and as much as 100,000 lb of sand were pumped. Figs. 3a and 3b show the results.

The start time for each well's production is made time zero in Fig. 3. Thus, in Fig. 3a, nine wells have been on-line for 240 days, eight wells for 290 days, etc.

The gas production plotted is simply the average of the number of wells on-line. Most of the pilot wells have been on production for almost 1 year. Furthermore, after about 150 days the wells are pumped-off, i.e., the water level lies below the bottom perforated coal.

Under this condition, which allows proper comparison of gas production, it can be seen that water fractures outperformed gel fractures by roughly 115 to 80 Mcfd, or by a factor of 1.4 (using the period 150-300 days).

The main difference lies in the cost.

To fracture both zones of Fig. 1, the water fractures cost $28,000/well, while the gel fractures cost $50,000/well (Table 2).

Thus, the water fractures are only about one half as expensive as the gel fractures, but their gas production appears to be better. Notice that the gel fractures produce more water (about 100 b/d) than the water fractures (about 25 b/d), although the gel fracture result is biased because two wells are close to an undersaturated area where water rates are very high.

Because gas production follows dewatering of these coalbeds (initially water saturated), one possible explanation is that with water fractures a smaller region of the reservoir is stimulated, and dewatered more quickly (e.g., because fracture length is shorter).

A major question is whether the gas production from the water fracture treatments will decline faster with time than that for the gel fracture treatments.

Some independent evidence from other portions of the Oak Grove field indicates that, over a longer time period, water-fractured wells do not decline any faster than gel-fractured wells.

MODELING GEL FRACS

Various fracture designs have been tried by Amoco in the Oak Grove field of the Black Warrior basin. To predict the propped length and conductivity for these fracture designs, we need to know two pieces of information in particular:

1. Appropriate fracture model

2. Leakoff coefficient.

APPROPRIATE FRACTURE MODEL

A downhole television camera was used to measure propped fracture heights and widths in the Black Creek coal zone following an openhole completion, and crosslinked gel fracture .2

The results indicated substantial height growth in agreement with projections based on a measured in situ stress profiles This same profile implies downwards height growth when the Mary Lee/Blue Creek coals are fractured.

Although such downward height growth has been modeled by a 3D fracture simulator, independent fracture diagnostics seem to infer upwards fracture growth above the Mary Lee/Blue Creek seams (about 200 ft) following a gel fracture treatment .4 The downhole television experiment also suggested that for fracture height much greater than coal thickness, all fracture widths (including coal) are controlled by properties of the bounding zones (shale, siltstone).

All of this means that a radial fracture model in the vertical section, which uses bounding zone moduli (not coal), is appropriate as a first approximation. Such a model would eventually approach a Perkins-Kern geometry if the fracture encountered sufficient stress barriers.

LEAKOFF COEFFICIENT

Several crosslinked gel minifracs have been pumped (typically 6,000 gal), mostly in the Oak Grove field.

Fracture fluid efficiencies for these minifracs have been found to be 0.47, 0.77, 0.71, and 0.63 for the Mary Lee/Blue Creek seams. For the Black Creek seams the coefficients are 0.55 and 0.68. Thus, it appears that fracture fluid efficiencies lie in the range of 50-80%.

A fracture simulator has been used to match the case with efficiency of 0.63, found in the Mary Lee/Blue Creek seams in a well in the White Oak field. The White Oak field lies across the Warrior River from the Oak Grove field. The simulation assumes a radial fracture model in the vertical plane.

The pressure decline (including closure) was matched by a leakoff coefficient, C = 0.023 ft/min 0.5, if it is assumed that all leakoff occurs through 14.5 ft of net coal pay (the interseam strata are tight,,5 and the coalbeds are the aquifers).

Note that the initial shut in pressure (ISIP) was 0.95 psi/ft, in support of the assumption of a vertical fracture.

An economic study was then carried out to find an optimal fracture half-length 6 in the Mary Lee/Blue Creek seams, assuming the above leakoff coefficient and a limiting fracture height of 200 ft (which was arbitrarily chosen).

The results of the study are shown in Fig. 4, which gives optimal half-length as a function of absolute reservoir permeability, between 5 and 20 md.

For example, for a permeability of 10 md, the optimal fracture half-length is about 175 ft. If fracture height were hot confined to 200 ft, but unconfined, the curve of Fig. 4 would be shifted a little to shorter half-lengths.

Fig. 4 can also be used as a guide for optimizing fracture length for Black Creek stimulations in Oak Grove field, because:

• We have not found a systematic difference yet in minifrac fluid efficiencies calculated for Mary Lee/Blue Creek zones, and for Black Creek zones.

• A systematic difference in permeabilities is also difficult to establish, in part because the measurement depends on the type of test (e.g., slug test versus fall-off test).

The fracture design on the left in Fig. 5 has been computed using the fracture stimulator STIMPLAN 7 for a desired fracture half-length of 170 ft (corresponding to coal permeability of 10 md).

Predicted average proppant concentration is about 1.6 psf, corresponding to a fracture conductivity of 6,600 md-ft and Fcd = 15.5.

A factor of 0.4 has been used to account for damage to the proppant pack. Fcd comes from:

Fcd = kfw

----

keff xp

where:

kfw = Fracture conductivity

xp = Propped fracture half-length

keff = Effective reservoir permeability to gas

A rule of thumb that is a useful approximation for describing relative permeability effects as gas desorbs from coal during dewatering is:

keff = k/4

where: k = Absolute reservoir permeability

Thus, keff = 2.5 md for a 10 md coalbed permeability, and we use this in all Fcd computations in this article.

The design on the left side of Fig. 5 is an aggressive design, with small pad volume (38%). Only a few of these were pumped in the pilot project. In general, too many of these designs screened out and terminated prematurely.

An alternative design has been used in the Mary Lee/Blue Creek zone in the Oak Grove field. The design on the right of Fig. 5 used an excessive pad (49%), and all the treatments were successfully implemented with virtually no screenouts.

This design is similar to the majority of those used in the pilot project of Fig. 2 (about two thirds of Mary Lee/Blue Creek, and half of Black Creek).

Fracture length and conductivity can be predicted for this design using STIMPLAN, with a fixed fracture height of 200 ft, and leakoff coefficient C - 0.023 ft/min 0.5 for 14.5 ft of net coal. The prediction is propped fracture half-length xf 330 ft, average proppant concentration is about 0.9 psf (corresponding to 3,610 mdft), and Fcd = 4.4 (again assuming effective coalbed permeability of 2.5 md).

By comparing the predicted fracture lengths and conductivities with those of the water fracture treatments in the pilot project, it appears that the optimal fracture design for the Mary Lee/Blue Creek seams (or Black Creek seams) lies somewhere between the two extremes of Fig. 5. (Unfortunately such designs were not tried.)

The fact that the design on the left of Fig. 5 was not successful all the time implies that standard minifrac pressure decline analysis needs to be adjusted in some way.' Possible adjustments are:

• Bridging of 12/20 proppant in width constrictions in the fracture (e.g., offsets/obstructions, or multistrands)9

• Pressure-dependent leak off8,

• Leakoff into a naturally fractured reservoir which has a time-dependence different from t 0.5 10

An important point is that we expect essentially all of the coal seams to be propped, whether they lie in the Mary Lee/Blue Creek zone or in the Black Creek zone (due to much better proppant/carrying capacity of these 30 lb/1,000 gal crosslinked gels).

MODELING WATER FRACS

Several water fracture treatments from the pilot project in Oak Grove field are listed in Table 3. Two are from Mary Lee/Blue Creek and two from Black Creek.

Total 12/20 mesh sand loads range from 53,000 to 76,000 lb/zone, and maximum sand concentrations range from 3 to 5 ppg.

Note that the maximum pressure in the pad portion of the treatments is always less than 1 psi/ft, and this implies a vertical fracture, as is characteristic of fractures in the Oak Grove field at these depths (1,600-2,200 ft).

The fracture modeling in this section emphasizes the interpretation of the rapidly rising pressures during proppant injection by a screenout condition in a vertical fracture. (Modeling of T-shaped fractures is not presented.)

Screenout modeling is a rarely used tool to estimate leakoff coefficients, and to predict propped fracture lengths and conductivities." 12 13

The pseudo-3D fracture simulator STIMPLAN has the sophistication to model fracture tip screenouts (and even pressure decline following a screenout up to the point of fracture closure). In essence, the increasing pressure is matched by adjusting the leakoff coefficient. The assumptions and approximations are as follows:

• In the case of several individual perforated intervals, the modeling assumes that one big vertical fracture links them all.

• A vertical profile of minimum in situ stress23 is not included; in fact, uniform vertical stress is assumed, except that artificial barriers are imposed to confine fracture height to 300 ft. (Thus, fracture growth is symmetrical upwards and downwards.)

• The vertical fracture is assumed to be controlled by the bounding zones (as discussed for the gel fracture treatments above); a Young's modules of 2.9 x 106 psi, obtained from an average of core measurements throughout the region between the Mary Lee/Blue Creek and the bottom of the Black Creek coals, is assumed.

• Pressure matching during the pad portion of the main fracture treatments is not done-it is difficult to deduce with accuracy the pad pressure from surface pressure measurements.

• Other possible contributions to the pressure rise, such as sand bridging in perforations or width constrictions in the fracture itself, are excluded. In fact, with the high leakoff coefficients found by this matching technique, the fracture simulator usually predicts some proppant bridging even before tip screenout. This warning is ignored in the execution of the simulator.

In our judgment, any more sophisticated modeling would not alter very much the conclusions of this screenout modeling. In the Mary Lee/Blue Creek zone, an example of screenout modeling is shown in Fig. 6. The pressure match in Fig. 6a is inadequate after shut-in, for reasons we do not understand.

However, the match until shut-in is reasonable, for a leakoff coefficient of Cc = 0.074 ft/min0.5, corresponding to an assumed net coal height of 19 ft (bounding zones are assumed completely tight').

The sand bank that forms at the bottom of the fracture, due to rapid 12/20 sand fall out through the water, is shown in Fig. 6b. Because symmetrical upwards and downwards height growth is assumed from the center of the perforated interval, the Mary Lee/Blue Creek seam is only partially propped, while the Newcastle seam is not propped at all.

The high leakoff coefficients (0.05-0.07 ft/min 0.5) in the Mary Lee/Blue Creek (Fig. 6, Table 4) reflect a low viscosity fracture fluid (water), and a relatively high coal permeability (5 md).

In fact, the permeability may be increased during fracture treatments by a strong dependence of permeability on effective stress.

In the Black Creek zone, an example of screenout modeling is shown in Fig. 7. Again, there are discrepancies in the pressure matching, but we have sought to find an approximate overall match. (Perhaps Cc = 0.04 ft/min O.5 is the best.)

Once again, the problem of 12/20 sand settling out of the fracturing fluid is shown in Fig. 7b. Here only about one third of the Black Creek coal seams are propped.

If downwards fracture height growth is assumed to be zero, as has been inferred previously,2 3 4 this would improve the situation slightly in Fig. 7b, but still several of the coal seams would not be propped.

In the other Black Creek case, where Cc< 0.02 ft/min0.5 (Table 4), the pressure profile was relatively flat until late times, when a very rapid pressure increase ensued, which evidently reflected a blockage near the well bore (modeling suggests less than 17 ft).

This is clearly not a tip screenout. Thus, in two separate Black Creek cases, there is not consensus on the leakoff coefficient. Nevertheless, propped fracture length and conductivity are similar for the two cases.

The results of the screenout modeling are summarized in Table 4. With the exception of the one Black Creek case mentioned above, leakoff coefficients in the coal are about 0.05 ft/min (1.5 . These would correspond to fracture fluid efficiencies of < 20% in typical minifrac tests (about 6,000 gal). This is considerably lower than the fluid efficiencies 50-80% found in crosslinked gel minifracs, as reported previously. This is expected because rate of leakoff with plain water should exceed that of crosslinked gels.

From Table 4, the screenout modeling predicts:

• Truncated propped lengths (about 100 ft)

• Large propped widths and high fracture conductivities (about 5,000 md-ft)

• Fcd values in the 13-45 range, compared with 2-10 for optimal stimulations. (Again an effective coalbed permeability of 2.5 md is assumed.)

Propped lengths are much less than the 330 ft predicted for the majority of those gel fractures used in the pilot. Also, the average water fracture conductivity is greater, at 5,600 md-ft, than 3,600 md-ft for the gel fracture majority.

The major difference, however, lies in the Fcd values, which are substantially higher in the water fractures (13-45) than in the majority of gel fractures (4.4).

Note that fracture conductivities in Table 4 are averaged over total fracture height.

In the Black Creek zone, fracture conductivities would be about three times higher, if averaged over sand bank height, meaning extremely high conductivity; but this, of course, is mitigated by the fact that several seams are not propped.

For Black Creek stimulations, the modeling predicts that only the bottom few seams will be propped (out of a total of 7 or 8).

For Mary Lee/Blue Creek stimulations, if the fracture is assumed to grow symmetrically up and down, the principal gas-bearing seams are not predicted to be propped effectively. (Of course, this is just the reason water fracture treatments were replaced by gel fracture treatments years ago.)

However, there are two possible situations which could improve the proppant placement:

1. The fracture grows upwards but not downwards in the Mary Lee/Blue Creek stimulations. (It appears to do this in the Black Creek zone. 4 5 6) Some support of this has been provided by mineback observations following water fracture treatments."

2. Proppant bridging/accumulation due to tortuosity/offsets in the vertical fracture in the coal.9 We have not included proppant bridging due to smaller fracture width in the bounding zone below a coal seam because such were not seen by the downhole television camera in the Black Creek coals .2 (We cannot rule out completely such an effect in the Mary Lee/Blue Creek coals.)

GEL DAMAGE

Some formation damage is a side effect of all fracturing operations with gelled fluids (filter cake formation and filtrate invasion).

For conventional reservoirs, the permeability damage is compensated for by the increase in effective well bore radius achieved by the fracture treatment. However, in naturally fractured formations gel damage to the natural fractures may be more serious, and steps may need to be taken to limit the access of gels to the natural fractures.15 16

Coal is, of course, a naturally fractured medium. Whole core samples of coal from San Juan and Black Warrior basins have been severely damaged after being exposed to gelled fracture fluids that have been broken and filtered.1 Permeabilities were reduced by factors mostly in the 13-23 range.

Because of its chemical composition, coal has the ability to sorb a variety of liquids and gases, resulting in the swelling of its matrix.' Even slight swelling can reduce cleat porosity and permeability substantially, because porosity is typically only 1-2%.

In the case of liquids (such as gelled fracturing fluids), sorption by coal is highly irreversible.

Although the lab results appear to implicate sorption as the cause of the permeability damage, some mechanical plugging by gel residue cannot be excluded. in fact, during coalbed fracturing in the field, it is anticipated that a spectrum of natural fractures exists, and some may be wide enough to transmit the gelled polymers.

Finally, neither reverse water flushing nor injection of acid significantly removed the permeability damage in the lab experiments. This may support the idea that the primary cause of the damage is sorption, not gel plugging. An effort has been made to find out how much gelled fracture fluid was left in the formation after fracture stimulation of a well in Oak Grove field using 30 lb/1,000 gal borate cross-linked HPG (hydroxypropyl guar) gel.

Ten samples of fluid were recovered over a 19-day period after the well was put online. These samples were analyzed to determine recovered gel concentration and to estimate total gel recovered.

Fig. 8 illustrates a linear decrease of gel concentration with cumulative volume of water recovered. During the 19-day period, 61% of the gel was recovered. The total gel recovery is predicted to be 68%.

If the data are fitted by an exponential decrease, the total gel recovery is predicted to be 82%. Altogether, possibly 20-30% of the injected gel remained in the formation.

If there is significant damage to coal by fracturing fluids, it may be exacerbated, at least in the Black Warrior basin, by an unusual set of circumstances which apply in the case of a vertical fracture:

• Most of the fracture fluid leaks off into the coal seams, and not the bounding zones of shale and sandstone.'

• The coal seams are relatively thin, totaling less than about 20 ft in a fracture of total height probably 200 ft or more.

• The coal porosity (cleats and larger natural fractures) is very low (1-2%).

Four implications are as follows:

1. The coal permeability is relatively high (about 100 md) during early fracture pressure decline. This estimate comes from standard minifrac pressure decline analysis assuming reservoir-controlled leakoff. 17 Such high permeabilities may not be inconsistent with a strong stress-dependent permeability in coal," whereby fracture fluid leaks off into the coal, pressurizes it, reduces effective stress, and causes the cleats to open up. (The reverse effect would happen during pressure drawdown.)

2. There is likely to be incomplete filter cake buildup on the face of the coal.

3. There is relatively deep invasion of the fracturing fluids into the coal formation (of order 50 ft). The explanation is given in the box.

4. Under high fracturing pressures, natural fractures may open up and allow gel to flow through, but then trap the gel as they close after the fracturing pressure decreases.15 (Check-valve effect-this would also happen in a T-fracture.)

ACKNOWLEDGMENT

Thanks are due to Amoco operations personnel and management for support and to Amoco Production Co. for permission to publish this article.

REFERENCES

1. Puri, R., King, G.E., and Palmer, I.D., "Damage to Coal Permeability During Hydraulic Fracturing," Joint Rocky Mountain Regulation and Low Permeability Symposium, Denver, April 1991.

2. Palmer, I.D., and Sparks, D.P., "Measurement of Induced Fractures by Downhole TV Camera in Coalbeds of the Black Warrior Basin," SPE 20660, 65th Annual Technical Conference of SPE. New Orleans, Sept. 23, 1990.

3. Lambert, S.W., Graves, S.L., and Jones, A.H., "Warrior basin drilling, stimulation," OGJ, Nov. 13, 1989, p. 87.

4. Saulsberry, J.L., Schraufnagel, R.A., and Jones, A.H., "Fracture Height Growth and Production from Multiple Reservoirs," SPE 20659, 65th Annual Technology Conference of SPE, New Orleans, Sept. 23, 1990.

5. Quarterly Review of Methane from Coal Seams Technology, Gas Research Institute, Vol. 3, No. 1, June 1985, p. 38.

6. Meyer, B., "Three Dimensional Hydraulic Fracturing Simulation on Personal Computers: Theory and Comparison Studies."SPE Eastern Regional Meeting, Morgantown, W.Va., Oct. 24, 1989.

7. NSI Technologies, Tulsa.

8. McDaniel, B.W., "Benefits and Problems of Minifrac Applications in Coalbed Methane Wells," Paper CIM/SPE 90-103, International Technology Meeting of CIM and SPE. Calgary, June 10, 1990.

9. Palmer, I.D., Davids, M.W., and Jeu, S.J., "Analysis of Unconventional Behavior Observed During Coalbed Fracturing Treatments," 1989 Coalbed Methane Symposium, Tuscaloosa, Ala., April 1989.

10. Soliman, M.Y., Poulsen, D.K., and Kuhlman, R. D., "General Minifrac Analysis for Heterogeneous Formation Including Spurt Loss," SPE 20705, 65th Annual Technology Conference of SPE, New Orleans, Sept. 23, 1990.

11. Nolte, K.G., "Application of Fracture Design Based on Pressure Analysis," SPE PE, February 1988, p. 31.

12. Smith, M.B., Miller, W.K., and Haga, J., "Tip Screen-Out Fracturing: A Technique for Soft, Unstable Formations," SPE 13273, 59th Annual Technology Conference of SPE, Houston, September 1984.

13. Martins, J.P., Leung, K.H., and Jackson, M.R., "Tip Screen-Out Fracturing Applied to the Ravenspurn South Gas Field Development," SPE 19766, 64th Annual Technical Conference of SPE, San Antonio, October 1989.

14. Steidl, P.F, personal communication 1990.

15. Warpinski, N.R., Branagan, P.T., Sattler, A.R., Cipolla, C.L., Lorenz, J.C., and Thorne, B.J., "A Case Study of a Stimulation Experiment in a Fluvial, Tight, Sandstone Gas Reservoir," SPE 18258, 63rd Annual Technology Conference, Houston, October 1988.

16. Gall, B.L., Sattler, A.R., Maloney, D.R., and Raible, C.J., "Permeability Damage to Natural Fractures Caused by Fracturing Fluid Polymers," SPE 17542, Rocky Mountain Regional Meeting, Casper, Wyo, May 1988.

17. Economides, M.J., and Nolte, K.G., Reservoir Stimulation (2nd ed.), Prentice Hall, Englewood Cliffs, N.J., 1989.

18. McKee, C.R., Bumb, A.C., and Koenig, R.A., "Stress-Dependent Permeability and Porosity of Coal," Coalbed Methane Symposium, Tuscaloosa, Ala. November 1987.

Copyright 1991 Oil & Gas Journal. All Rights Reserved.