Professional Documents
Culture Documents
Modeling of fluid flow through rough fractures has gained where bo represents the distribution peaks, and the mean
importance over the years. This can be attributed to the aperture is 2 bo . The same distribution was assumed when
extremely low ultimate recoveries obtained from naturally
fractured reservoirs, in spite of their huge reserves. Attempts considering the channeling of flow through fractured media9.
are being made to develop efficient models to better formulate Tsang and Tsang9 assumed the channel width to be a constant
depletion plans. The first comprehensive work on flow of the same order as the correlation length λ , where
through open fractures was done by Lomize1, in which he used correlation length is the spatial length within which the
parallel glass plates and demonstrated the validity of cubic law apertures have similar values. The reduction in channel
for laminar flow. He modeled fluid flow with different apertures affected the tracer breakthrough curves when normal
fracture shapes and investigated the effects of changing the stress across a fracture was increased.
fracture walls from smooth to rough.
2 SPE 89941
Although gamma distribution was considered, some requirements were chosen typical of a sandstone sample. An
authors7,10 observed a lognormal distribution of fracture X-ray CT scanner identifies density differences between
apertures from well logs and experiments conducted on cores. various objects16. As the density of a fluid is less compared to
Later Moreno et al.11 followed the lognormal approach when the density of rock, the X-ray scans can be used to clearly
conducting flow and tracer transport model in a single distinguish a fluid filled fracture from the matrix. Although a
fracture. Using the approach, the flow patterns showed strong clear visualization of fracture and matrix is possible, the CT
resemblance to field observations reported by Bourke12. Tsang numbers (dimensionless numbers that are functions of
and Tsang13 used lognormal approach to distribute fracture densities) do not represent any physical characteristic. In order
apertures for obtaining a relationship between flow and to obtain the fracture aperture measurement, a calibration
transport measurements of variable apertures. The frequency technique has to be used to correlate the CT numbers with the
of the lognormal distribution is written as follows14: actual aperture size. Keller17 made use of non-intrusive
imaging to obtain aperture calibration curves for granite and
1 − (lnx − α) 2 sandstone. Using the curve, fracture apertures as small as
f(x) = exp[ ] ............................ (2)
35µm could be measured. A similar technique was followed
x 2πβ 2 2β 2 by He18 in 1998. Both techniques use the integrated CT signal
where, to determine fracture aperture. The integration was done by
summing up the differences between a minimum rock CT
σ2
β 2 = ln(1 + ) ........................................................ (3) number and the CT numbers pertaining to the fracture. Our
δ2 calibration curve was developed based on this existing
and technique.
β2 The experimental setup consists of two halves of the rock
α = lnδ − .............................................................. (4) sample (Berea Core), feeler gauges and core holding
2
The mean aperture was determined from fluid flow equipment. The two halves were polished thoroughly to
measurements and this value was used to distribute the reduce surface roughness as much as possible. Smooth
apertures lognormally. The aperture variance was determined surfaces ensure that there is a proper match between the halves
from tracer breakthrough. Alfred et al.14 also assumed and hence smaller fractures can be measured. Feeler gauges
lognormal distribution for fracture apertures to account for were then placed between the two flat surfaces to create
fracture roughness. Keller15 imaged fracture apertures using an different fracture widths (Fig. 1). The following sizes of feeler
X-ray CT scanner. Based on the study using three different gauges were used in the calibration experiment: 38 µm, 51
cores, he concluded that the fractures follow a lognormal µm, 64 µm, 76 µm, 102 µm, 127 µm, 152 µm, 178 µm, 508
distribution. µm, and 813 µm.
The core sample with fracture was inserted into a sleeve
Fracture properties are strongly affected by changes in to avoid artefact effects and held in the core holder. Multiple
stress conditions and therefore the effect of stress acting on the scans were then taken along the length of the core,
reservoirs should be considered in obtaining fracture aperture perpendicular to the longitudinal axis. Fig. 2A shows a typical
distributions. Even though extensive research has been done set of scans for the Berea sample with 38 mm fracture size.
on fracture aperture distributions, the effect of varying stress The color variations indicate variable densities in the sample,
conditions has not yet been investigated. Stress changes with the relatively high density steel feeler gauge set to bright
aperture size and might have an effect on the distribution of pink color, the less dense rock material set to a bright orange
fracture apertures. Thus, the aim of this study is to measure color and the lower densities set to green, blue and black in
fracture apertures under different stress conditions using an X- decreasing order of densities. First scan in Fig. 2A is taken
ray CT scanner. Several feeler gauges were used to create through the steel feeler gauge placed on one end of core
fractures of known size and obtain a calibration curve. Once sample; hence the color appears brighter in the middle of the
the calibration curve was obtained, fracture aperture core. Both ends of core sample have a feeler gauge and scans
distributions from a fractured core were measured under were taken starting from one feeler gauge to the start of the
different overburden pressures. Probability distributions of next feeler gauge, with intervals of 5 mm. The next three scans
fracture apertures were examined to obtain the mean aperture show a fracture aperture of 38 µm along the length of the core.
and variance values. Using those values, the probability The dip in CT number curve in the middle corresponds to the
distributions of density function were plotted to determine the location of the fracture region and this clearly distinguishes
distribution of fracture apertures at each overburden pressure. fracture from the matrix. Since this dip (Fig. 3) corresponds to
Waterflooding was then performed using a fractured core a fracture size of 38 µm, a correlation can be obtained between
to validate the use of lognormal approach for fracture the area enclosed by the curve and the fracture size. The dip is
apertures and to prove that the parallel plate model is not valid not abrupt, but rather a smooth transition from matrix to
for modeling fluid flow through rough surface. fracture. This effect is due to dispersion of CT numbers, which
is caused due to a finite beam width and oversampling. Fig.
2B shows the fracture aperture size of 102 µm, along the
Calibration Technique length of the core. As the aperture size increases, the CT
A Picker 4th generation CT scanner was used for imaging number decreases and the fracture aperture can be seen clearly
the core samples. The power, voltage and algorithm in the middle of the core, with a darker color. Fig. 4 shows a
SPE 89941 3
comparison of the average CT number curves obtained for The apertures are distributed in the range of 0 to 2000
various fracture sizes. microns as shown in Fig. 8. The tail of the aperture
distribution is long and the aperture distribution is right
Calibration Curve skewed which suggests that the distribution might follow
either gamma or lognormal distribution as given in the earlier
A plot of CT number versus the pixel number was studies8,9,15. The tail of the apertures controls the fluid flow as
made based on the CT numbers obtained from each scan. An fluid will tend to flow through preferred channels of least
average CT number plot was obtained to account for minor resistance14. Since most apertures fall in the range of 100 to
variations in fracture sizes caused by surface roughness. An 500 microns, the permeability measurements will largely be
example of such plot is shown in Fig. 5. From this plot, a controlled by small apertures. The presence of some large
threshold CT number was identified, which essentially is the apertures in the range of 600 to 2000 microns may provide the
minimum rock CT number. All the CT numbers in the area main path for the fluid flow if they are interconnected. The
below the minimum rock CT number are subtracted from the calculation of flow rate based on these measured
rock CT number and the differences are used to obtain the permeabilities will be different from the calculation using a
integrated CT signal for the fracture size. The integrated CT single mean value of permeability as assumed by parallel plate
signals are calculated for each pixel number and are given an approach.
index starting from 1 to n as shown in Fig. 5. The area of
Using the mean and variance of the aperture values
integrated CT region is calculated from the following formula:
obtained from the probability distribution plot for each
For index i equal to 1 to c, overburden pressure, the distributions of probability density
1 function can be obtained by applying Eq. 2. The distribution
Area(i) = IntCT(i − 1) + [IntCT(i) − IntCT(i − 1)] .. (5) plots are presented in Fig. 9. Comparison of actual dataset to
2 calculated distribution reveals that the distribution of pdf
For index i equal to c+1 to n,
follows lognormal distribution at all stresses.
1
Area(i) = IntCT(i) + [IntCT(i − 1) − IntCT(i)] ........ (6) Without applying an overburden pressure, the probability
2 density function that follows lognormal distribution has a
mean aperture value of 370.53 microns and a variance of
After measuring nine different feeler gauges, we obtained a 44847 (Fig. 9). This lognormal distribution confirms the
calibration curve for fracture aperture (Fig. 6). The plot shows previous findings4,11,13,14,15. As the overburden pressure
a linear relationship between the integrated CT signal and increases, the mean and variance of the apertures decrease
fracture aperture. The linear relationship holds even at small suggesting that the permeability decreases due to increase in
fracture apertures. The equation obtained from this calibration overburden pressures. The mean of the apertures decreases
curve is, drastically from 370.53 microns to 197.997 microns when an
y = 8.7616x + 460.7 ..................................................... (7) initial overburden pressure of 500 psi is applied.
By rearranging the equation we can calculate fracture The fracture surfaces are rough and have many contact points
aperture, using the integrated CT signal as follows: along the fracture. The contact points tend to breakdown under
overburden stress. The decrease in aperture sizes is not
x = 0.1141y − 52.582 .................................................. (8) significant with further increase in overburden pressures.
Given a fracture sample, the fracture apertures can be Since each reservoir experiences different overburden
measured using the calibration curve. pressures, this research will be important to identify some of
the uncertainties in predicting fluid flow through fractured
Fracture Aperture Distribution reservoirs.
saturation was reached. CT numbers of the water saturated seen from the figure, the water saturation profiles predicted by
core are seen to be in the range of 1650 to 1700. Due to high the distributed aperture model are in good agreement with the
permeability of the fracture, water is seen to preferentially CT profiles. Fig. 17 and Fig. 18 show that a good match is
flow along the fracture plane. Fig. 11 shows the 3D water obtained between the observed and simulated oil recovery and
saturations along the length of the core. water breakthrough. Fig. 18 also shows an improvement in the
water cut match compared to the previous case. Fig. 19 shows
the match obtained between observed and predicted
Parallel Plate Simulation Model cumulative water production.
4. Analysis of sensitivity to injection rates shows that, 7. Bourke, P. J., Dunance, E. M., Heath, M. J., and
as the injection rate increases, the difference in the oil Hodgkinson, D. D.: “Fracture Hydrology Relevant to
recoveries predicted by the two models also Radionuclide Transport, AERE Rep. 11414, Atomic
increases. Energy Res. Estab., Harwell, United Kingdom, 1985.
8. Gentier, S.: “Morphologie et comportement
5. At higher injection rates, higher viscous forces cause
hydromécanique d’une fracture naturelle dans un
early breakthrough from the fracture and hence the
granite sous contrainte normale,” doctoral thesis,
oil recovery obtained is lesser.
Univ. d’Orléans, France, 1986.
6. For large aperture sizes, the performance of the 9. Tsang, Y. W. and Tsang, C. F.: “Channel Model of
parallel plate model becomes closer to that of the Flow Through Fractured Media,” Water Resources
distributed aperture model. The sensitivity of oil Research, 23(3), 467-479, 1987.
recovery to variance of aperture sizes makes the 10. Bianchi, L. and Snow, D.: “Permeability Crystalline
parallel plate model invalid. Rock Interpretated from Measured Orientations and
Apertures of Fractures,” Annu. Arid Zone, 8(2), 231-
245, 1968.
Nomenclature 11. Moreno, L., Tsang, Y. W., Tsang, C. F., Hale, F. V.,
and Neretnieks, I.: “Flow and Tracer Transport in a
Area(i) = Area of section i Single Fracture: A Stochastic Model and Its Relation
bo = Distribution peak to Some Field Observations,” Water Resources
b = Aperture size Research, 24(12), 2033-2048, 1988.
12. Bourke, P. J.: “Channeling of Flow Through
c = Index of lower CT number
Fractures in Rock,” in Proceedings of GEOVAL-87
f(x) = Frequency of apertures International Symposium, Swed. Nucl. Power
i = Index Inspectorate, Stockholm, Sweden, 1987.
IntCT(i) = Integrated CT area for section i 13. Tsang, Y. W. and Tsang, C. F.: “Hydrological
n = Index of higher CT number Characterization of Variable-Aperture Fractures,”
n(b) = Probability density distribution Rock Joints, Barton & Stephansson (eds), Balkema,
x = Fracture aperture Rotterdam, 1990.
14. Alfred, D., Putra, E., and Schechter, D.S.: “Modeling
y = Total integrated CT signal
Fluid Flow through Single Fractures Using
Experimental, Stochastic and Simulation Approaches,”
References paper SPE/DOE 89442 presented at 2004 Improved
Oil Recovery Symposium, OK, Tulsa, 17–21April.
1. Lomize, G. M.: “Seepage in Fissured Rocks,” State 15. Keller, A. A.: “Single and multiphase flow and
Press, Moskow-Leningrad, 1951. transport in fractured porous media,” PhD dissertation
2. Witherspoon, P. A., Wang, J. S. Y., Iwai, K., and submitted to Stanford University (1996).
Gale, J. E.: “Validity of Cubic Law for Fluid Flow in 16. Wellington, S. L. and Vinegar, H. J.: “CT Studies of
a Deformable Rock Fracture,” Water Resources Surfactant- Induced CO2 Mobility Control,” paper
Research, 16(6), 1016-1024, 1980. SPE 14393 presented at the 1985 SPE Annual
3. Alfred, D.: “Modeling Fluid Flow through a Single Technical Conference and Exhibition, Las Vegas,
Fracture Using Experimental, Stochastic and September 22-25.
Simulation Approaches,” master’s thesis, Texas 17. Keller, A.A.: “High Resolution CAT Imaging of
A&M University, 2003. Fractures in Consolidated Materials,” Int. J. Rock
4. Tsang, Y. W.: “The Effect of Tortuosity on Fluid Mech. Min. Sci. (1997), 34 (3/4), 358-375.
Flow through a Single Fracture,” Water Resources 18. He, M.: “Application of X-ray Tomography to
Research, 20(9), 1209-1215, 1984. Measurement of Fractures in Rocks,” M.S. Thesis,
5. Tsang, Y. W. and Witherspoon, P. A.: “The Stanford University (1998).
Dependence of Fracture Mechanical and Fluid 19. Putra, E., Muralidharan, V., and Schechter, D.S.:
Properties on Fracture Roughness and Sample Size,” “Overburden Pressure Affects Fracture Aperture and
J. Geophys. Res., 88(B3), 2359-2366, 1983. Fracture Permeability in a Fracture Reservoir,” SA
6. Pyrak, L. R., Myer, L. R., and Cook, N. G. W.: Journal of Technology (Fall 2003), pp 57-63.
“Determination of Fracture Void Geometry and
Contact Area at Different Effective Stress,” Eos
Trans. AGU (abstract), 66(46), 903, 1985.
6 SPE 89941
A. 38 µm feeler gauge
1300 38 microns
1200
Smoothed 1100
surface 102 microns
1000
900
0 20 40 60 80
1420
1400 0 n
n-1
1380 1
1360
Minimum Rock
1340 CT Number
CT Number
1320
1300
1280
c-1
1260
c+1
A. No overburden pressure
1240
c
1220
30 32 34 36 38 40 42 44 46 48
Pixel Number
8000
Integrated CT signal
6000
4000
0
0 100 200 300 400 500 600 700 800 900
FIG. 6 – Calibration curve between integrated CT signal FIG. 7 – Sample scans taken along the length of the core
and fracture aperture. with different overburden pressures.
8 SPE 89941
3500
3000
Frequency of Occuring
2500
2000
1500
1000
500
0 0
0 0
0
0 0
0 0
0 0
0 0
0 0
0 00
-2 0
0
10 10
30 30
00 00
20 20
40 40
50 50
60 60
70 70
80 80
00 10
00 20
00 30
00 40
00 50
00 0
00 70
00 80
00 90
00
90 - 9
6
10 - 1
11 - 1
12 - 1
13 - 1
14 - 1
15 - 1
16 - 1
17 - 1
18 - 1
19 - 1
-
-
-
-
0
0.9
0.8
0.7
0.5
0.4
0.3
0.2
FIG. 11 – 3D images of water saturations during 0.1
waterflooding at various timesteps.
0
0 30 60 90 120 150 180 210 240 270
Time (min)
0 min
3 min
9 min
14 min
FIG. 14 – Comparison between Observed and simulated
water cut using parallel plate model.
20 min
36 min
275 min
80
70
60
Oil Recovery (% IOIP)
50
40
30
20
10
0
0 50 100 150 200 250 300 FIG. 15 – 3D map of aperture distribution.
Time (min)
0 min
FIG. 13 – Comparison between Observed and simulated
oil recovery using parallel plate model. 3 min
9 min
14 min
20 min
36 min
275 min
80 60
70
50
40
50
40 30
30
20
20
10
10
0 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300
1 70
69
Oil Recovery (% IOIP)
0.9
68
0.8
67
0.7
66
Water Cut (%)
0.6 65
0.5 64
63
0.4
62
0.3
61
0.2 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9