STIMULATING SHALES-1 UNDERSTANDING THE RESERVOIR IMPORTANT TO SUCCESSFUL STIMULATION

David D. Cramer
BJ Services Co.
Denver

In anisotropic Bakken shale reservoirs, fracture treatments serve to extend the well bore radius past a disturbed zone and vertically connect discrete intervals.

Natural fractures in the near-well bore area strongly control the well deliverability rate.

The Bakken is one of the few shale formations in the world with commercial oil production.

This first of a two-part series covers the Bakken reservoir properties that influence production and stimulation treatments. The concluding part will discuss the design and effectiveness of the treatments.

BAKKEN FORMATION

The Bakken formation produces oil over a widespread area in the Williston basin. Black shales within the zone contain a high concentration of kerogen (i.e., 7-13% of total rock volume) and are the source of this oil.1 2

Kerogen conversion to liquid hydrocarbon has occurred at sites where the subsurface temperature has reached a critical maturation level.1 As much as 30% of the original kerogen in the Bakken shale has converted to oil.2

The loss of "solids" volume as a result of oil generation caused the Bakken source rock to compact, forcing oil into the pores of the shales, and into the sands and carbonates surrounding the shales.3 The oil displaced interstitial water during this process.

At most locations, the low permeability and porosity of the rock beds in the gross Bakken interval inhibited fluid displacement. Similar to the action of an hydraulic press, pore pressure increased greatly in places where oil generation proceeded at a high rate (e.g., in areas having a high in situ temperature).4

IN SITU STRESS

Pressure gradients as high as 0.73 psi/ft have been gauged during pressure build-up tests of producing Bakken wells.5 6

Pore pressure influences the level of in situ stress. The general state of stress underground is that in which the three principal stresses are unequal.7

The least stress in tectonically relaxed areas is usually oriented in a horizontal direction.

The minimum horizontal stress is a function of an elastic property of the rock and the magnitudes of the overburden load and pore pressure, as expressed below.

[SEE FORMULA (1)]

where:

sHmin, = Minimum horizontal in situ stress (v/(1 - v)) (sv - a P) = Effective rock stress

v Poisson's ratio

sv Vertical or overburden stress

a = Poroelastic constant

(0 a

P = Pore pressure

T = Tectonic stress

Equation 1 shows that pore fluids "share" the transmitted overburden load with the reservoir rock matrix.8 As fluid pressure increased in the porous Bakken intervals during hydrocarbon generation, the effective rock matrix stress decreased.9

The reduced stress favored tensile rock failure and the formation of vertical fractures at sites where other stress relief mechanisms were active. Fracture orientation is normal to the least principal stress and thus parallel to the maximum horizontal stress.7

Such fracture systems may close partially when pore pressure decreases with fluid withdrawal (production). Beeks6 recognized this fracture volume compressibility and its affect on formation permeability and oil displacement in a material balance study of the Bakken/Sanish formation.

Localized in situ fracture systems control the productivity of most Bakken wel IS.5 69 Horizontal drilling increases the well bore radius in the Bakken formation, improving the chances of intercepting natural fractures.

PAY ZONE

Most horizontal Bakken wells are drilled in the upper Bakken shale with the presumption that the upper shale contains the entire reservoir or the shale is in effective communication with other permeable rocks. If these premises are incorrect, partial completion effects could impede production from horizontal wells.10

Various diagnostic methods have provided indications of the thickness and vertical continuity of Bakken reservoirs.6

Fig. la is a combined suite of wire line and drilling logs from a vertical well located in the Spring Lake field in Richland County, Mont.

The only significant hydrocarbon shows were observed when a porous dolostone in the middle Bakken interval was penetrated at a depth of 9,770 ft. Drill stem tests (DST) indicated that the initial pore pressure gradient was 0.47 psi/ft in this well. This well was stimulated with a gelled oil frac treatment and completed in September 1981. It has produced more oil (110,000 bbl) than any other Bakken well in Richland County.11

Within a mile of the above well, a nonproductive horizontal well was drilled in the upper shale. A comparison of these two wells suggests that reservoir potential exists in the middle Bakken "silt" zone but not in the upper shale in the Spring Lake area.

TEMPERATURE LOGS

Temperature logging of producing wells has provided information on the contribution of deeper intervals. The lowest depth of fluid entry can be estimated by identifying the point at which traces of static and flowing temperature surveys start to overlay.

Fig. 1b shows this diagnostic method. Multiple temperature surveys were run several weeks after stimulating a vertical well in the Buckhorn field in Billings County. It appears that oil flowed from the upper 10 ft of the Three Forks formation. This analysis agrees with the interpretations of similar testing of wells in the Devils Pass and Elkhorn Ranch fields.

HYDROCARBON SHOWS

Exceptional hydrocarbon shows often occur when penetrating the organic-rich claystone in the lower Lodgepole formation, informally named the False Bakken interval. This happened during the drilling of a vertical well in the Elkhorn Ranch field in Billings County.

Oil and gas influx persisted because of the high pore pressure gradient (i.e., 0.68 psi/ft) that existed in this developing Bakken reservoir (Fig. 2a). Sustained hydrocarbon inflow prevented evaluation of the Bakken and Three Forks intervals.

Hydrocarbon mud logs have provided clues regarding the relative quality of Bakken subintervals when the drilling-fluid hydrostatic gradient was close to or exceeded the pore pressure gradient.

This method is exemplified by analyzing a well log obtained from a recently completed vertical well in the Elkhorn Ranch field (Fig. 2b). Production from other Bakken wells in the field reduced the reservoir pore pressure close to the density of the mud used to drill the well (i.e., the 0.50-0.55 psi/ft range).

Gas influx started when drilling into the False Bakken interval. Gas levels increased when the upper Bakken shale was penetrated, then diminished during the drilling of a hard limestone below the shale. Hydrocarbon entry peaked in magnitude when the silty dolostones of the middle Bakken and upper Three Forks were penetrated.

This example hints that discrete zones exist within the Bakken. A low permeability interval seems to restrict communication between the upper shale and lower dolomitic intervals.

DRILL STEM TESTS

Drill stem tests assessed the relative quality of the various Bakken zones in a vertical well in the Antelope field in McKenzie County, N.D.13 The first test evaluated 100 ft of interval from the False Bakken to the middle of the lower Bakken shale. This test recovered gas-cut drilling mud with negligible pressure buildup during shut-in periods (i.e., maximum pressure 970 psi).

The second test exposed 25 ft of interval in the lower Bakken shale and upper Three Forks formation.

This second test recovered high oil and gas cut drilling mud and reached a maximum pressure of 6,962 psi during the initial 90 min shutin period.

A comparison of the tests suggests that the sandstone and dolostones of the upper Three Forks formation, and possibly the lowest portion of the lower Bakken shale, have the best reservoir potential in this well. Yet, selective cased-hole production testing has indicated that upper Bakken zones in the Antelope field also contribute to oil production.

For example, the Rose Hopkins Hand No. 1 was initially an open-hole completion in the upper Three Forks formation.

The addition of new perforations in the middle Bakken silt interval increased total well production from 87 to 165 bo/d.

ROCK PROPERTIES

X-ray diffraction (XRD) analysis determines the mineral content of rock samples. This helps identify rock composition and potential problems with stimulation fluids and techniques in the Bakken formation. Table 1 presents the results of XRD analyses of core from wells in the Bicentennial and Whitetail fields of McKenzie and Billings Counties, respectively.

Quartz framework grains and clays constitute a high percentage of the minerals in the False Bakken and upper Bakken shale intervals. Clay content varies in the Bakken shale, ranging from 5 to 35% by volume.

The Bakken shale has a small amount of acid-soluble minerals. Pyrite (FeS2), present in the False Bakken and Bakken shale, is very slowly soluble in acid. Pyrite does not present a risk of iron hydroxide precipitation after acid spending because the iron has a valence of + 2 (ferrous ion).14

The basal Lodgepole interval is a carbonate rock that contains a small amount of dispersed quartz silt grains. Clay content is low.

The middle Bakken silt interval varies in mineral composition, both vertically within the gross interval and laterally throughout the basin in the same relative vertical position.'

Quartz concentration in this zone increases northward in the basin. The middle Bakken is a sandstone reservoir in the Rocanville pool of southeastern Saskatchewan.15 In the Bicentennial field, the middle Bakken and upper Three Forks intervals are silty dolostones with low clay content.

ACID SOLUBILITY

Table 2 summarizes the results of special hydrochloric (HCI) acid solubility tests of the same Bicentennial field core samples. This testing uses XRD analysis to quantify the amount of acid-soluble minerals in each rock sample.

A rock sample is incubated in a particular strength of HCI acid at the simulated bottom hole temperature. The chosen sample contains a stoichiometric excess of acid-soluble minerals, giving the acid a chance to react completely with the rock.

After several hours of acid/rock contact, analysts measure rock dissolution, residual acid strength, and fines production.

Acid dissolving efficiency (i.e., actual dissolution divided by potential dissolution) is very high for the basal Lodgepole interval. Due to the dispersed quartz sand that is present in this zone, the HCI acid generated a large amount of fines. Fines generation was proportional to acid dissolving power.

The Bakken shale is almost inert in HCI acid. The HCI acid dissolved very little of the acid-soluble matter in the rock.

Acid dissolving efficiency is high in the middle Bakken silt interval. The HCI acid generated a small amount of fines, suggesting that the quartz silt is more cohesive and less dispersed than in the Lodgepole sample.

Bicentennial field core samples do not appear to be sensitive to HCI. Additional tests show that all zones are more reactive with hydrochloric/hydroflouric (HCI/HF) acid blends, and rock stability problems could exist with this acid type. These results replicate test findings in Bakken core from other areas of the basin.

OTHER CORE TESTS

Wettability tests indicate that the Bakken shales are oil-wet. The False Bakken and sandstone and carbonate intervals in the Lodgepole, Bakken, and Three Forks are water-wet.

"Drain" tests, using columns of pulverized rock, show that fresh water and kerosine flow equally well through samples from each Bakken interval. This suggests that clay minerals in the Bakken zones are not easily dispersed by water. The low matrix permeability of the Bakken intervals limits the use of core flow testing with stimulation fluids.

Pore geometry and fluid-content analysis of core plugs provides information on the porosity and oil saturation of the various Bakken intervals.

Figs. 3a and b graph analytical results for core obtained from a vertical well in Richland County. High oil saturations in the basal Lodgepole and Bakken silt intervals show that fluid entered these zones from the Bakken shale.

Fluid storage capacity is greatest in the upper 15 ft of the Bakken silt member. As is common in the basin, porosity in the basal Lodgepole and Bakken shale is very low.

Matrix permeability is low in all zones and significant only in rock exceeding 6% porosity. Formation flow capacity (permeability times net zone height) was calculated from DST analysis and agreed well with the matrix permeability/height product of the middle silt core samples. This correlation suggests that natural fracturing has not occurred to any appreciable extent near the evaluated well.

The apparent lack of natural fracture development in Richland County Bakken producers may be partly related to the lower pore pressure in this part of the basin. The above well had an initial pore pressure gradient of 0.54 psi/ft.

Fig. 3c shows the analytical results for a core obtained from a well in the Elkhorn Ranch field. Both the basal Lodgepole and Bakken silt intervals have high oil saturations.

This implies that fluid displacement occurred from the Bakken shale, and high pore pressure, measured during testing of the initial Bakken producing wells, also existed in these zones. Porosity and matrix permeability are very low in all intervals and do not compare with the significantly higher permeability values calculated from pressure transient tests.6 16

The large discrepancy between permeability measurements derived from well testing and core analysis implies that natural fractures are responsible for the flow capacity of Bakken wells in the Billings Nose area. Pore fluid charging in rock intervals adjacent to the Bakken shale reduced the effective rock stress in these intervals, establishing an environment conducive to natural fracture development.

High-pressure mercury injection testing has provided additional understanding of the pore geometry of Bakken matrix rock. Fig. 3d shows the test results for core samples obtained from a well located in the Roosevelt field.

In all intervals, very small pore throats connect most of the voids, which explains the low permeability measured during core flow testing.

Two distinct pore types are present in the Bakken shale and Three Forks intervals. Large pore throats connect roughly 1 0% of the pores (or 0.4% of total rock volume) in these zones. These larger pores could be microfractures.

The very small pores in the rock do not contribute in a significant way to reservoir flow capacity but do store large volumes of oil. Bakken microporosity could recharge depleted natural fractures. Fracture recharging occurs in other overpressured, low matrix permeability reservoirs.17

Fig. 3e shows the test results of core samples obtained from a well in the Antelope field. As in the Billings Nose area, the Bakken shales have charged the middle Bakken and upper Three Forks intervals with oil.

The permeability of Antelope field Bakken core is very low and comparable with Richland County and Billings Nose core samples. As in the Billings Nose area, fracture systems contribute greatly to the productivity of many Antelope field wells.5

MECHANICAL ROCK PROPERTIES

The mechanical properties of rock influence the natural fracture development and hydraulic fracture propagation. Rock mechanical constants can be inferred by acoustic logging techniques. Accurate measurement of the compressional and shear-wave velocities through the rock with a long-spaced sonic log enables the calculation of an important elastic property, Poisson's ratio.18

[SEE FORMULA (2)]

where:

v = Poisson's ratio

Vc = Compressional-wave velocity

vs = Shear-wave velocity

Poisson's ratio is the ratio of lateral expansion to longitudinal contraction for a rock under a uniaxial stress condition. Poisson's ratio relates the fraction of the overburden stress that has transferred horizontally to rock layers.

Poisson's ratio is proportional to the minimum principal horizontal stress (Equation 1). The Poisson's ratio log identifies possible contrast of in situ stress among the Bakken intervals. This information enables the prediction of hydraulic fracture height growth.

Fig. 4 show plots of log-derived Poisson's ratios for Bakken intervals in various parts of the basin. All basin areas show similar tendencies. The massive Lodgepole interval exhibits a relatively high Poisson's ratio, which may inhibit fracture growth.

A reduction in Poisson's ratio occurs in the False Bakken interval, and the highest Poisson's ratio is in the basal 10 ft of the Lodgepole interval. The Bakken shales exhibit relatively low Poisson's ratio levels. In the Bakken silt and Three Forks intervals, the Poisson's ratio is intermediate in value to the shales and Lodgepole zones.

Several pitfalls need to be considered when using Poisson's ratio logs for predicting fracture height growth. Small errors in measuring shearwave velocity can lead to significant variations in the calculated elastic properties.

Pore pressure levels vary within the gross Bakken interval, complicating the computation of in situ stress. Pore pressure is normally lower in the massive Lodgepole interval above the False Bakken. If the difference in pore pressure between these zones is as high a 400 psi, unconfined fracture height growth could occur.

Portions of the Bakken shales have a very high clay content and could behave plastically, negating the implicit assumption of perfectly elastic rock behavior that is required in Equation 1.

Increased fracture toughness in bounding rock layers could blunt vertical fracture growth. If treatment materials (e.g., sand, gel) bridge or dehydrate at a point of reduced fracture width (i.e., pinchpoint), fluid pressure at the vertical extremities of the fracture will be reduced and fracture height growth will be arrested.

PERMEABILITY ANISOTROPY

In situ stress fields are normally anisotropic. Wherever this is the case, a preferred fracture azimuth exists that is perpendicular to the minimum, principal in situ stress. Fractures prefer to take the path of least resistance and open against the smallest stress.18

The minimum principal stress is usually horizontal. As indicated in Fig. 5a, tensional or extensional-type fractures, both natural and hydraulic, will tend to be vertical and extend in the direction of maximum horizontal stress. Extensional fractures may be responsible for the productivity of most Bakken wells.9

Directional, anisotropic permeability is an expected feature of this type of reservoir.

In cases where multiple sets of natural fractures exist in the subsurface, the most permeable fractures should also be oriented parallel to the maximum principal stress due to the smaller closure pressure acting on this fracture set.

Information from various sources has contributed to an understanding of permeability characteristics in anisotropic reservoirs.

A study of oriented core of Mission Canyon dolostones in the Little Knife field in Billings County found that open natural fractures are unidirectional, were formed in recent geologic time, and are parallel to the azimuth of maximum principal horizontal stress.

The fracture azimuth trends northeast-southwest to east-west and is roughly perpendicular to the axial trace of the Little Knife anticline.19 Open fractures in the Bakken may have also formed in recent geologic time, concurrent with the generation of oil.4

A study of pressure transient tests in Bakken horizontal wells supports the idea of a northeast-southwest to east-west trending anisotropic fracture system.16

Analyses of oriented Bakken cores in the Elkhorn Ranch and Roosevelt fields suggest that the dominant maximum horizontal stress azimuth is N45E 10 in the Billings Nose area.

Fig. 5b shows the direction of maximum horizontal stress as determined by analysis of petal fractures observed in cores of the Lodgepole, Bakken, and Three Forks intervals of the Roosevelt field.

These results are comparable with fracture studies of Mission Canyon cores in the Billings Nose area, where both open natural fractures and the maximum principal horizontal stress trend N50E.20

Multiple-well "interference" testing in the Billings Nose area confirmed that restricted hydraulic communication exists between off-trend Bakken wells.

An overpressured, highly anisotropic gas reservoir at the multiwell test site in Garfield County, Colo., behaved similarly.17 21

At the multiwell site, permeability in a direction transverse to the unidirectional fracture trend occurs by the low-angle intersection of subparallel natural fractures (Fig. 5c). This tortuous flowpath is responsible for the low interwell conductivity (permeability) estimated from well interference tests.21 A similar network could exist in Bakken reservoirs.

Permeability anisotropy affects stimulation treatment results because induced hydraulic fractures will tend to parallel natural fracture systems, rather than cross them. This reduces the efficiency of hydraulically created fractures and explains why the fracture length calculated from post-treatment well testing in the Bakken is usually very short or undetectable.17 19

PERMEABILITY CALCULATION

The permeability calculated from single well tests in an anisotropic reservoir is an average of the minimum and maximum horizontal permeability.22

[SEE FORMULA (3)]

where:

kt = Test permeability

kx = Maximum horizontal permeability

ky = Minimum horizontal permeability

The properties of the low-permeability matrix rock and secondary natural fractures control inflow to the propped hydraulic fracture (i.e., fracture face is exposed to the minimum horizontal permeability). This limits production, and the well will "act" as if it were stimulated by a shorter fracture.

Fig. 6a shows the effect of permeability anisotropy on calculated fracture length for the expected case where the hydraulic fracture parallels the maximum permeability trend. Calculated dimensionless fracture conductivity (Cr) will actually be greater than expected because hydraulic fracture conductivity is contrasted to the minimum horizontal permeability of the reservoir. Fig. 6b shows this effect.

A reservoir simulator matched the production of a vertical Bakken well in the Antelope field using a horizontal permeability anisotropy ratio of 100:1.13.

This calibrated reservoir model evaluated the effect of simulated propped fracture length on well productivity.

Increasing the fracture length affected long-term well productivity only marginally (Fig. 6c). This shows the potential advantage that properly oriented horizontal wells have over hydraulically fractured vertical wells in areas where natural fractures control reservoir performance in the Bakken.

EXCEPTIONS

There are exceptions to the general rule of hydraulic fracture inefficiency in the Bakken.

As noted previously, Richland County Bakken production probably occurs from the matrix porosity network in the Bakken silt.

Natural fractures do not seem to influence reservoir flow capacity. Postfrac well testing of a Richland County Bakken well indicated the presence of an effective fracture half-length of 430 ft. This value is comparable to the propped fracture length calculated by a fracture simulator program.

Naturally fractured reservoirs sometimes exhibit vertical permeability anisotropy. The Little Knife core study indicated that the average height of natural fractures was only 1 ft.19

VERTICAL ANISOTROPY

Natural fractures often terminate at vertical discontinuities at the multiwell site (Fig. 5c).21 Vertical fractures abruptly terminate in core specimens of the lower Bakken shale in the Antelope field.

Other indications of vertical permeability anisotropy in the Bakken were mentioned previously.

Vertical permeability anisotropy limits the success of horizontal well completions and may require hydraulic fracture stimulation to overcome its affect.10

REFERENCES

1. Webster, R., "Petroleum Source Rocks and Stratigraphy of the Bakken Formation in North Dakota," Rocky Mountain Association of Geologists Symposium, 1987, pp. 259-85.

2. Schmoker, J., and Hester, T., "Organic Carbon in Bakken Formation, United States Portion of Williston Basin," AAPG Bulletin Vol.67, No.12, December 1983, pp. 2165-74.

3. du Rouchet, J., "Stress Fields, A Key to Oil Migration," AAPG Bulletin, Vol. 65, No. 1, January 1981, pp. 74-85.

4. Gosnold, W., and Huang, Y., "Factors Determining the Thermal History of a Continental Basin," 5th International Williston Basin Symposium, pp. 17-21.

5. Murray, G., "Quantitative Fracture Study-Sanish Pool, McKenzie County, North Dakota," AAPG Bulletin. Vol. 52, No. 1, January 1968, pp. 57-65.

6. Cramer, D., "Reservoir Characteristics and Stimulation Techniques in the Bakken Formation and Adjacent Beds, Billings Nose Area, Williston Basin," paper No. SPE 15166, Rocky Mountain Regional Meeting, Billings, Mont., May 1921, 1986.

7. Hubbert, K., and Willis, D., "Mechanics of Hydraulic Fracturing," paper No. 686-G, 31st Annual SPE Fall Meeting, Los Angeles, Oct. 14-17, 1956.

8. Roegiers, J.C., "Elements of Rock Mechanics," Reservoir Stimulation, Schlumberger Educational Services, 1987, Chapter 2, pp. 1-21.

9. Meissner, F., "Petroleum Geology of the Bakken Formation, Williston Basin, North Dakota and Montana," MGS Williston Basin Symposium, 1978, pp. 207-27

10. Cramer, D., "Guides exist for fracture treatment in horizontal wells," OGJ, Mar. 27, 1989, pp. 41-52.

11. Hansen, W. ad., Study Notes No. 1, Bakken Formation Short Course, Montana Geological Society, March 1990.

12. Ferti, W., Abnormal Formation Pressures-Implications to Exploration Drilling and Production of Oil and Gas Resources, Elsevier Scientific Publishing Co., 1976, p. 325.

13. Simtech Consulting Services Inc., "Antelope Field: Preliminary Horizontal Drilling Evaluation-Bakken Formation," Aug. 10, 1989.

14. Smith, C., et al.,"Secondary Deposition of Iron Compounds Following Acidizing Treatments," JPT, September 1969, pp. 1121-29.

15. Osinski, W., Geology and Production History of the Bakken Formation in the Rocanville Area, Southeast Saskatchewan, Dept. of Mineral Resources, 1970, p. 16.

16. Williams, E., and Kikani, J., "Pres sure Transient Analysis of Horizontal Wells in a Fractured Reservoir," paper No. SPE 20612, 65th Annual SPE Technical Conference, New Orleans, Sept. 23-26, 1990.

17. Branagan, P., et al., "Designing and Evaluating Hydraulic Fracture Treatments in Naturally Fractured Reservoirs," paper No. SPE/DOE 16434, SPE/DOE Low Permeability Symposium, Denver, May 18-19, 1987.

18. Warpinski, N., and Smith, M., "Rock Mechanics and Fracture Geometry," SPE Monograph, Vol. 12, Recent Advances in Hydraulic Fracturing, 1989, pp. 57-80.

19. Narr, W., and Burruss, R., "Origin of Reservoir Fractures in Little Knife Field, North Dakota," AAPG Bulletin, Vol. 68, No. 9, Sept. 1984, pp. 1,087-1,100.

20. Breig, J., "Mission Canyon Reservoirs of the Billings Nose," Occurrence and Petrophysical Properties of Carbonate Reservoirs in the Rocky Mountain Region, Rocky Mountain Association of Geologists, Denver, 1988.

21. Lorenz, J., Warpinski, N., Branagan, P., and Sattler, A., "Fracture Characteristics and Reservoir Behavior of Stress Sensitive Fracture Systems in Flay-Lying Lenticular Formations," JPT, June 1989, pp. 615-22.

22. Ben Naceur, K., and Economides, M., "Production From Naturally Fissured Reservoirs Intercepted by a Vertical Hydraulic Fracture," SPEFE, December 1989, pp. 550-58.

23. Dobkins, T.,"Improved Methods to Determine Hydraulic Fracture Height," JPT, April 1981, pp. 719-26.

Copyright 1991 Oil & Gas Journal. All Rights Reserved.