ANALYSES POINT TO UNTAPPED POTENTIAL IN TIGHT GAS SAND

April 6, 1992
Frank E. Syfan Jr., Bradley M. Robinson S.A. Holditch & Associates Inc. College Station, Tex. The Davis sandstone, located in the Fort Worth basin of north-central Texas, has been identified as a potential candidate for additional reserve development, as a result of a reservoir/production performance study. The fluvial deltaic sedimentary environment found in the Davis is representative of approximately 40% of the tight gas sands found in the U.S.
Frank E. Syfan Jr., Bradley M. Robinson
S.A. Holditch & Associates Inc.
College Station, Tex.

The Davis sandstone, located in the Fort Worth basin of north-central Texas, has been identified as a potential candidate for additional reserve development, as a result of a reservoir/production performance study.

The fluvial deltaic sedimentary environment found in the Davis is representative of approximately 40% of the tight gas sands found in the U.S.

The relatively small drainage area, obtained from detailed reservoir analyses, raises the possibility that infill drilling could recover significant additional amounts of gas from the tight Davis sandstone.

The average drainage area (less than 50 acres) associated with the Davis sandstone is small when compared to the 160 acre, or higher (up to 320 acres) drilling patterns.

It is very likely that additional tight gas formations located throughout the U.S. fall into this same category and are in need of additional reservoir analysis.

This study was done to determine if the Davis sandstone in the Fort Worth basin was a viable candidate for the Gas Research Institute's (GRI) planned Hydraulic Fracture Test Site.

(Plans for the test site are now focused on the Rocky Mountain area, according to GRI. -the editor)

To learn more about the Davis sandstone, an intense cooperative research study was initiated with an area operator. As such, several wells in northwestern Parker County and southeastern Jack County in north-central Texas were analyzed to determine values of formation permeability, the degree of stimulation (skin factor), drainage area, and in situ stress within the zone as well as the layers above and below.

Cooperative research was also performed on three additional wells to calculate in situ stresses of the Davis interval, using sonic logs to obtain a geological description of the Davis from whole core analysis and to determine natural and induced fracture azimuth.

GEOLOGY

The Davis sandstone, also known locally as the Pregnant shale, is a low-permeability, gas-bearing sandstone located within the Atoka group of the northern Fort Worth basin of north-central Texas (Fig. 1). The sandstone depth varies from 3,000 to 4,500 ft and is 50-200 ft thick. The unit has blanket-type sand deposition and is early to middle Pennsylvanian in age.

The Fort Worth basin, which contains the Davis sandstone, is a paleozoic foreland basin, about 20,000 sq miles in area. Uplifted areas surrounding the basin were the sources of the Pennsylvanian clastic sediments that filled the basin.

The Ouachita uplift, located east of the basin, was a predominant sediment source. Additional sediment was shed from the Muenster arch located north of the basin. A progressive westward shift of depocenters occurred in middle to late Pennsylvanian time."

The Davis sandstone has been interpreted to be a system of coalesced, wave-dominated deltas primarily composed of sand-rich coastal barrier facies. 1 6 These facies may consist of barrier islands, beach ridges, or sand ridges on a strand-plain that accumulated parallel to the shoreline to form a sand-rich facies having excellent strike continuity and moderately good dip continuity.

The Davis facies and the northern Fort Worth basin may include coastal barriers in western Parker and southern Wise and Jack counties that result from wave redistribution of substantial amounts of sands along the delta margins.

The accurate shape of the deltas suggests a period of low sediment input, which results in the dominance of marine over fluvial processes. 1 6

WELL A CORE

A portion of the cooperative research performed on the Davis sandstone consisted of analysis of whole core from Well A in northwestern Parker County, Tex. The core penetrated about 100 ft of Davis sandstone. Fig. 2 is a section of the porosity log covering the Davis sandstone interval in Well A.

The Davis interval in this core consists of several upward-coarsening sediment packages. At the base of the Davis interval, about 5 ft of sandstone interbedded with thin (millimeter to centimeter size) shale layers grades up into about 30 ft of clean, fine-grained to very fine grained sandstone.

This basal package is overlain by another sediment package that consists of about 25 ft of sandstone interbedded with thin shale layers that grade upward into about 5 ft of clean sandstone. A third coarsening-upward sediment package consists of 10 ft of shale, 5 ft of sandstone interlayered with thin shale layers and 5 ft of sandstone.

The uppermost part of the Davis consists of about 15 ft of sandstone interlayered with thin shale layers. The basal Davis contact with underlying shale is a sharp, apparently erosional contact that is characterized by a thin mud rip-up clast conglomerate with an erosive base.

The upper Davis contact with overlying shale was not cored in Well A.

Davis samples from Well A core are fine-grained to very fine grained sandstone, shale, and interbedded shale and siltstone, or sandstone interlaminated on a millimeter to centimeter scale. Quartz, plagioclase, and metamorphic rock fragments are the most abundant detrital grains, and the sandstones are classified as sub-litharenites. 7

Quartz is the most abundant cement, ranging from 8% to 15% of the rock volume in clean, well-sorted sandstones.

Authigenic chlorite (1.5-8%), illite (0.5-3%), kaolinite (0-2%), and ankerite (0-5%) also occur in the Davis sandstones. The authigenic minerals observed in these Davis sandstone samples are similar to those in the slightly older Atoka A-1 sandstone at Black Ranch field, Hood County, Tex. 8

Both natural and drilling-induced fractures occur in the Davis sandstone core from Well A. Most natural fractures occur in sandstone, but locally, fractures cross sandstone-shale contacts or occur entirely within shale.

Calcite is the main vein-fill mineral, although gypsum was observed in one fracture. Because infilling minerals are absent or difficult to detect in some natural fractures, and many induced fractures lack distinctive petal-centerline shapes, separating natural and drilling-inducted fractures can be difficult.

The majority of the natural fractures in Well A are characteristically short, vertically discontinuous, and widely separate from each other with heights of 10 cm or less.

PRODUCTION ANALYSIS

Nine wells located in the northwestern corner of Parker County and the southeastern comer of Jack County were analyzed using a production data analysis and history matching program, 10 The program, Promat, 9 10 uses constant pressure, analytical solutions to the flow equations generated for different reservoir conditions.

The variations in reservoir models include radial flow or hydraulic fracture conditions, single or dual porosity, and finite area or infinite acting drainage patterns. The flow equation used to analyze the data for the nine Davis wells was for radial flow in a single porosity, finite reservoir.

Table 1 presents the results of the Promat analyses. Permeability (k) ranged from 0.021 md in Well J to 0.396 md for Well B. Average permeability for the nine wells was 0.08 md.

Drainage area ranged from 10.3 to 130.1 acres. The average drainage area for the nine wells analyzed by Promat, plus Well L (analyzed using a reservoir simulator), was 48.9 acres. The relatively small drainage area was surprising, because many of the wells in this portion of the basin are drilled on 160 acre spacing or higher.

The relatively small drainage area is also substantiated by Well M drilled in November 1990. Well M was drilled approximately 2,600 ft south of Well L.

An analysis of pressure buildup data in the Davis sandstone of Well L indicated an average reservoir pressure of approximately 450 psi. However, a static bottom hole pressure of approximately 1,200 psi, which is close to original reservoir pressure for the Davis interval, was recorded in the newly drilled offset, Well M.

Well B, which has produced in excess of 1.0 bcf, calculated a minimum drainage area of 130 acres using Promat. Well B was the only well out of the nine wells zed that calculated a drainage area significantly greater than the average.

Al] nine wells were found to be slightly stimulated. Skin factors ranged from - 0.30 in Well D to - 3.75 in Well 1.

The relatively small negative skin factor values indicate very short effective hydraulic fracture half-lengths, if any. The slightly stimulated condition of these well bores could be the result of the acid balloff and small fracture stimulation work (maximum sand concentrations of 2.0 ppg) performed during completion operations for the Davis sandstone interval.

The presence of natural fractures could also be influencing the skin factors as much as the original stimulations. Even in an unstimulated well, a skin factor approaching -3.0 is possible due to the contribution of natural fractures.

A unique match of Well B production data was not obtained initially using Promat. Thus, the data were history matched using a single-phase, two-dimensional, finite-difference reservoir simulator, Fracsim.

Fracsim was built specifically to analyze single-phase, vertically fractured (oil or gas) wells. The fracture half-length, fracture width, and fracture conductivity are controlled by grid dimensions input into the model. Fracsim can also be used to investigate other single well problems such as a horizontally fractured well, or radial flow in a single or multiple layer reservoir.

The model is also capable of compensating for several reservoir heterogeneities such as damage around the well bore and hydraulic fracture, non-Darcy flow, fracture permeability reduction and closure with time, formation permeability reduction with time, and well bore storage.

Fracsim will usually yield results closer to the actual reservoir conditions when compared to a simpler model such as Promat.

A relatively good history match of Well B production was achieved using Fracsim. These results are presented in Fig. 3a. Table 2 compares the Fracsim history match with the Promat values.

The values for permeability and drainage area obtained from the Fracsim history match are similar to those obtained using Promat. However, the history match using the Fracsim model is in closer agreement with the actual production than that calculated by the simpler model, Promat.

We think the key to obtaining the better match was using a greater negative skin factor in the analysis. As such, it appears that the well is actually more stimulated than indicated by the Promat analysis and possibly more so than the other "typical" wells. This additional stimulation may be one of the contributing factors (along with the higher permeability) to the better drainage efficiency.

PRESSURE BUILDUP

Well L, located in Parker County, Tex., was shut in on Sept. 6, 1990, for a 21-day (504 hr) pressure buildup test. Well L produced without interruption for 25 days prior to shut in. The pressure buildup data were analyzed using Welltest, a conventional pressure transient analysis program (Table 3).

The pressure buildup test data have been analyzed using pseudopressures, 11 pseudotimes, 12 and effective pseudoshut-in times. 13

The data have been graphed using adjusted pressures and equivalent time functions which are variations of pseudopressure and pseudotime.

The finite conductivity, vertical fracture-type curve match of the pressure buildup data is presented graphically in Fig. 3b. Equation 1, using the formation parameters summarized in Table 3, indicated that the semilog straight line, or pseudoradial flow, would not begin until the shut-in time (t) reached 2,751 hr (115 days). Because the pressure buildup test for Well L lasted only 21 days, a Homer analysis would have been invalid.

See book for formula

where:

t = Time, hr

f = Porosity, fraction

m = Viscosity, cp

ct = Total compressibility, psi-1

Lf = Hydraulic fracture half-length, ft

tLfD = Dimensionless time referred to fracture half-length

k = Permeability, md.

The reservoir simulator, Fracsim, was also used to analyze the pressure buildup data of Well L. Table 3 also summarizes the reservoir properties obtained from history matching both the production and the pressure buildup data on Well L. Fig. 3c illustrates the history match of the pressure buildup data obtained using Fracsim.

Overall, the simulated pressure buildup data are in excellent agreement with the observed data, and the analysis results also agree closely with the values obtained from the conventional pressure buildup analysis.

It was noted during the early portion of the pressure buildup (1 x 10-4 to 3 x 10-3 days) that an additional pressure drop existed near the well bore. This pressure drop, calculated to be in the initial 5 ft of the fracture, yielded a fracture conductivity of 320 md-ft vs. 1,600 md-ft for the remaining fracture length.

The average fracture conductivity from the history match is higher than the type curve values, but the conductivity reduction near the well bore has the effect of reducing the overall "effective" conductivity that one might calculate from conventional analyses. This is why the type curve match indicated a lower value of fracture conductivity,

Without reservoir modeling, the near-well bore restriction could not have been quantified. This reduction in conductivity could be due to any number of factors, but the overall effect on long-term production performance was predicted to be negligible. However, past reservoir modeling studies have shown that fracture conductivity restrictions can and have affected recoveries of other wells. 14 15

STRESS DISTRIBUTION

Knowledge of the in situ stress distribution within the reservoir sandstones and the surrounding formations is one of the most important factors in the design and analysis of a hydraulic fracture treatment. Stress contrast between layers of rock controls the vertical fracture growth and, therefore, directly affects fracture length and width. Even though the importance of stress distribution is evident, these data are rarely measured due to the expense, interpretation problems, and mechanical risks.

Techniques exist which can be used to estimate in situ stresses from logs and cores. However, it is important that these estimates be calibrated with actual field measurements of in situ stress. Injection tests are by far the most reliable source of in situ stress data. If the values of in situ stress estimated from logs and cores can be correlated with field injection tests, then an operator can reduce the cost of future wells by using only logs and lithology to estimate the in situ stress profile.

Four zones in Well K, located in Jack County, Tex., were perforated using a 3-1/8-in. casing gun, over a 2-ft interval, with four shots/ft at 120 phasing. The zones were isolated with a retrievable packer and bridge plug. A seating nipple for a downhole shut-off tool was located directly above the packer. The zones tested were:

  1. 4,644-4,646 ft, shale

  2. 4,532-4,334 ft, shale

  3. 4,378-4,438 ft, Davis sandstone

  4. 4,310-4,312 ft, shale

  5. 4,266-4,268 ft, shale

Because the Davis sandstone had already been perforated over a 60-ft interval, a normal stress test could not be performed. Therefore, a mini-fracture treatment was pumped to measure the in situ stress in the Davis sandstone.

The GRI treatment analysis unit (TAU) was used to monitor and record data from the in situ stress tests. 16 Bottom hole pressures were measured using a quartz crystal pressure gauge with surface readout. The downhole shut-off tool minimized the well bore storage effects that tend to distort the shape of the pressure falloff curve.

During the in situ stress tests, 2% KCI water was pumped down 2-3/8-in. tubing at an injection rate of 5-20 gpm. After injection pressures stabilized, injection into the formation was stopped by closing the downhole shut-off tool. The pressure falloff data were recorded during the shut down. Injection volume per stress test was less than 400 gal. However, the volume of the Davis mini-fracture treatment was about 16,000 gal.

During an in situ stress test, the pressure decline is monitored until fracture closure is detected. At this point, the individual test stage ends. Several additional stages are pumped at the same injection rate until the results are reproduced. Fig. 3d presents an example of the multiple injection/falloff stages performed on the shale at 4,532-4,534 ft.

The testing began with the deepest zone of interest, 4,644-4,646 ft. After completing the tests on this zone, the straddle packer and tubing were moved and set across the next set of perforations. The entire test procedure was then repeated on the second zone. This procedure was continued until Zones 1, 2, 4, and 5 had been tested. The final zone tested was the Davis sandstone.

ANALYSIS TECHNIQUES

The minimum horizontal stress is also called the in situ stress or the fracture closure pressure. The value of in situ stress is sometimes estimated from the initial shut-in pressure (ISIP). However, this value may not be very accurate.

Immediately after the pumps are shut down, during the time before a fracture closes, the pressure at the well bore will be larger than the in situ stress. When small fluid volumes are injected, such as the in situ stress tests, the ISIP may be reasonably close to the value of in site stress. However the ISIP can be very difficult to determine accurately in many cases 17 and errors on the order of several hundred psi may occur when applying this method.

In our analysis of the in situ stress tests for the Davis interval, a reservoir engineering approach was used to analyze the falloff pressures. After fluid injection into the formation is stopped, the pressure decline resulting from fluid leak-off into the formation should represent linear flow while the fracture remains open. Therefore, a plot of the bottom hole pressure-vs.-square root of shut-in time should be a straight line.

When the fracture closes, the relationship between pressure and the square root of time ceases to be linear. The point at which the pressure curve deviates from the straight line is interpreted as an indication of fracture closure, and the pressure at that point is selected as the value of the in situ stress.

The assumption of true linear flow may not be exactly correct due to the constantly decreasing fracture volume (i.e., the effects of fracture compliance upon pressure while the fracture is open). But it is believed that by examining the pressure falloff data after several injection periods, the point at which the fracture closes can be determined and used to estimate the value of in situ stress.

The use of the downhole shut-off tool minimizes the well bore storage effects. However, well bore storage still affects the pressure falloff to some degree.

To determine the extent of any well bore storage effects, a log-log plot of the change in pressure (DELTAp) after shut in-vs.-shut-in time (DELTAt) is constructed. This log-log plot should exhibit a unit slope during well bore storage. This plot should also reveal a half-slope during linear flow or a quarter-slope during bi-linear flow.

Having both a square root of time plot and a log-log plot, the portions of the data that represent well bore storage and linear flow can be better identified.

RESULTS

Originally, only four zones were to be tested in Well K. These were the shales below the Davis, the shale above the Davis at 4,266-4,268 ft, and the Davis sandstone. However, because the test of the shale at 4,266 ft yielded an apparent stress value much less than the apparent stress values in the shales below the Davis, a fourth shale zone (4,310-4,312 ft) was added to the list of zones to be tested.

These tests were conducted Oct. 24-26, 1990. Additional tests of the Davis and the shale at 4,310 ft were conducted Nov. 8-9, 1990 (Table 4).

The tests of Zones I and 2 (shales below the Davis) resulted in high stress values. Our original analysis indicated closure pressures of 4,090 psi at 4,644 ft (0.88 psi/ft) and 4,440 psi at 4,532 ft (0.98 psi/ft).

The test of Zone 5 (4,266-4,268 ft) yielded a closure pressure of 2,690 psi (0.63 psi/ft) which was much lower than the results in the shales below the Davis. This led to the suspicion that there might be a problem with cement bond behind the casing.

Thus, another shale above the Davis, 4,310-4,312 ft (Zone 4), was tested. This zone was actually tested on two occasions with a total of eleven individual injection stages. Closure pressure for this zone was estimated to be 3,145 psi (0.73 psi/ft).

The Davis sandstone was also tested. Because the Davis had already been perforated (60-ft interval) and stimulated with a large acid treatment, a larger volume treatment (mini-frac) was necessary to determine the closure pressure of the Davis. From the step rate test which was part of the mini-fracture treatment, and the pressure falloff data, the closure pressure in the Davis sandstone was estimated to be 1,980 psi (0.45 psi/ft).

RESERVOIR PERFORMANCE

The results of the production data analysis listed in Table 1 indicate that the average drainage area for the Davis sandstone is approximately 50 acres or less, with the exception of Well B. The majority of the wells drilled in the northwestern portion of the Fort Worth basin have been drilled on 160 acre spacing, or higher.

Based on the Promat analyses, it appears that additional development drilling may be needed to recover untapped reserves in the Davis sandstone. Additional reservoir simulation should be used to determine the potential of future drilling activity.

Well C was used to forecast reservoir performance for a "typical" Davis well because of the quality of the production data history match (Fig. 4a) and also, because the reservoir properties determined for this well using Promat were fairly close to the average values calculated for the Davis sandstone.

These reservoir properties were input into the Fracsim model to predict future performance for a radial flow situation using drainage areas of 50 acres (actual Promat value), 80 acres, and 120 acres.

The 120-acre spacing was arbitrarily chosen as the largest drainage area based on Well B analysis, indicated in Table 1. In addition, future performance was predicted for several different effective fracture half-lengths varying from 150 to 600 ft.

All cases assumed an economic limit of 15 Mcfd, a constant final flowing bottom hole pressure of 140 psia, and a maximum production interval of 10,000 days (27.4 years).

The forecast of future performance for each case is presented in Table 5 and in Figs. 4b-4d.

All three drainage areas indicate substantial increases in initial productivity, cumulative recovery, and income acceleration as the effective hydraulic fracture lengths increase. As shown in the 50-acre drainage case, ultimate recovery for Well C increases by 32% by creating an hydraulic fracture with an effective propped half-length of only 300 ft.

As previously discussed for Well B, it is highly probable that the minimum drainage area could also be increased by achieving longer fractures (i.e., more stimulation). Well B was initially completed in the Davis sandstone in February 1974. The well was fracture treated with 15,000 gal of gelled 2% KCI water containing 20,000 lb of 20/40 sand. Maximum sand concentration during the fracture treatment was 2.0 ppg. This size treatment was fairly typical for the area.

Analysis of Well B stimulation treatment using a three-dimensional fracture propagation model, indicated very little effective fracture length was achieved, if any. Thus, we really cannot explain the reasons for the higher degree of stimulation in this well unless it is related to natural fractures.

However, if longer propped fractures were obtained, it is clear that the ultimate recovery for a well with reservoir properties similar to Well C could be expected to increase by 200-300% over the radial flow case. To achieve these results, larger volume treatments will most likely need to be pumped.

REFERENCES

  1. Finley, R.J., Geology and engineering characteristics of selected low-permeability gas sandstones: a national survey, the University of Texas at Austin, Bureau of Economic Geology Report of Investigations No. 138, 1984, p. 220.

  2. Herkommer, M.A., and Denke, G.W., "Stratigraphy and hydrocarbons, Parker County, Texas," Petroleum Geology of the Fort Worth Basin and Bend Arch area, Martin, C. A. editor., Dallas Geological Society, 1982, p. 97-127.

  3. Lahti, V.R., and Huber, W.F., "The Atoka Group (Pennsylvanian) of the Boonsville Field Area, North-Central Texas," Petroleum Geology of the Fort Worth Basin and Bend Arch area, Martin, C.A. editor, Dallas Geological Society, 1982, P. 377-99.

  4. Lovick, G.P., Mazzini, C.G., and Koitla, D.A., "Atokan clastics-depositional environments in a foreland basin," OGJ, Vol. 80, No. 5, 1982, pp. 181-99.

  5. Ng, D.T.W., "Subsurface study of Atoka (Lower Pennsylvanian) clastic rocks in parts of Jack, Palo Pinto, Parker and Wise Counties, North-Central Texas," American Association of Petroleum Geologists Bulletin, Vol. 63, 1979 pp. 50-66.

  6. Thompson, D. M., Atoka Group (Lower to Middle Pennsylvanian), Northern Fort Worth Basin, Texas: terrgenous depositional systems, digenesis, and reservoir distribution and quality, The University of Texas at Austin, Bureau of Economic Geology Report of Investigations No. 125, P. 62.

  7. Folk, R.L., Petrology of sedimentary rocks, Hemphill Publishing Co., Austin, 1974, p. 182.

  8. Crowder, W.T. Jr., and Helmers, M.E., "Alternate completion methods lead to success in tight upper Atoka sandstones, Black Ranch Area, Hood County, Texas: Southwest Section," American Association of Petroleum Geologists Transactions, 1987, p. 51-60.

  9. Murtha, J.A., et al., "Practical Analysis Methods for Well Test and Production Data," S.A. Holditch & Associates Inc., Topical Report to the Gas Research Institute, GRI Contract No. 5084-2130980, October 1987.

  10. Watson, A.T., and Lee, W.J., "A New Algorithm for Automatic History Patching Production Data," Paper No. SPE 15228, SPE Unconventional Gas Technology Symposium, Louisville, Ky., May 18-21, 1986.

  11. Al-Hussainy, R., Ramey, H.J. Jr., and Crawford, P.B., "The Flow of Real Gases Through Porous Media," JPT, May 1966.

  12. Agarwal, R.G., "Real Gas Pseudo-Time - A New Function for Pressure Buildup Analysis of MHF Gas Wells," Paper No. SPE 8270, SPE Technical Conference and Exhibition, Las Vegas, Nev., Sept. 23-26, 1979.

  13. Agarwal, R.G., "A New Method to Account for Producing Time Effects When Drawdown Type Curves Are Used to Analyze Pressure Buildup and Other Test Data," Paper No. SPE 9289, SPE Technical Conference and Exhibition, Dallas, Sept. 21-24, 1980.

  14. Robinson, B.M., Holditch, S.A., and Whitehead, W.S., "Minimizing Damage to a Propped Fracture by Controlled Flowback Procedure," Paper No. SPE 13250, Unconventional Gas Technology Symposium, Louisville, Ky., May 18-21, 1986.

  15. Berthelot, J.M., "Effects of Fluid Recovery Upon Well Performance and Ultimate Recovery of Hydraulically Fractured Gas Wells," MS thesis, Texas A&M University, Tex., May 1990.

  16. In-Focus-Tight Gas Sands, Vol 6, No. 1, Gas Research Institute, April 1989, pp. 11-17.

  17. McLennan, J.D., and Roegiers, J.C., "How Instantaneous are Instantaneous Shut-In Pressures?" Paper No. SPE 11064, SPE Annual Technical Conference and Exhibition, New Orleans, Sept. 2629, 1982.

  18. Sookprasong, P.A., "Plot procedure finds closure pressure," OGJ, Sept. 8, 1986, pp. 110-12.

Copyright 1992 Oil & Gas Journal. All Rights Reserved.