Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 13

RELIABILITY ANALYSIS

OF HYDRAULIC
FRACTUREING
Contents
1. Introduction ........................................................................................................................................... 2
2. First order reliability method ................................................................................................................ 2
3. Hydraulic fracture models ..................................................................................................................... 3
4. Hydraulic fracturing reliability analysis ............................................................................................... 4
5. Conclusion .......................................................................................................................................... 10
6. References ........................................................................................................................................... 11
1. Introduction
Hydraulic fracturing is the process of extracting hydrocarbons from nonconventional reservoirs.
Conventional reservoirs have a high concentration of hydrocarbons are in high porosity zone. The
hydrocarbons from these reservoirs are extracted by the confining pressure and pumping. In
nonconventional reservoir has a low concentration of hydrocarbon in a rock formation with low porosity
such as shale rocks. The hydraulic fracturing process involves initiation and propagation of fractures by
hydraulic loading of fracture fluid. Hydraulic fracturing has also been applied in other geotechnical fields,
including for the disposal of waste drill cuttings underground (Moschovidis et al., 2000), heat production
from geothermal reservoirs (Harlow & Pracht, 1972), CO2 sequestration (Boschi et al., 2009), coal bed
methane recovery (Heo et al., 2015), gas control in coal mines (Li, Quangui et al., 2015), and for the in-situ
characterization of stress (Desroches, 1995).
Various simplified analytical hydraulic fracturing models have been developed with the assumption of
constant fracture height. The most widely used analytical models include Khristianovic-Geertsma-de Klerk
(KGD) model (Zheltov, 1955; Geertsma & De Klerk, 1969) and the PerkinsKern-Nordgren (PKN) model
(Perkins Jr & Kern, 1961; Nordgren, 1972). These models work under the assumptions about crack
geometry, the criteria governing crack origination, and the dynamics of crack propagation, internal fluid
flow, formation, and fluid properties. The available analytical and numerical models do not take into
account uncertainties in variables such as parameters of rock mechanics, in situ stress, and injection rate.
Lukyanov and Chugunov, 2015 applied global sensitivity analysis to quantify and rank uncertainties
associated with hydraulic fracture process. The work has shown that it is essential to take into account
uncertainties in hydraulic fracturing because these exert considerable influence on the shape and dimensions
of propagating fractures. Reliability analysis is one method that can be used to appropriately evaluate
uncertainty. This analytical approach was developed and has been successfully applied in rock mechanics
and engineering (Zhao et al., 2014; Oreste, 2005; Mollon et al., 2009; Lü & Low, 2011; Li, Hang-Zhou &
Low, 2010; Hoek, 1998).
This study employs the work of Zhao et al., 2018 in order to characterize the uncertainties involved with
hydraulic fracturing. First order reliability method (FORM) with constrained optimization routine is utilized
to analyze the reliability index associated with various aspect of hydraulic fracturing. Further investigations
are conducted to analyze the effect of model parameters and their variation on the reliability of the
performance of hydraulic fracturing.

2. First order reliability method


The reliability index (𝛽) for correlated normal variables is given by Hasofer, 1974 as follows:

𝛽 = 𝑚𝑖𝑛𝑋∈𝐹 √(𝑋 − 𝜇)𝑇 𝐶 −1 (𝑋 − 𝜇) (1)

Where 𝑋 is the vector of normal random variables 𝑥𝑖 , 𝜇 is the vector of mean values, and C is the covariance
matrix. 𝐹 is the failure domain defined by limit state functions. Low, B. K., 1997; Low, B. K. and Tang,
1997 presented a different interpretation of reliability index based on the concept of expanding ellipsoid in
original space as follows:

𝑥𝑖 − 𝜇𝑖 𝑇 𝑥𝑖 − 𝜇𝑖
𝛽 = 𝑚𝑖𝑛𝑋∈𝐹 √( ) [𝑅]−1 ( ) (2)
𝜎𝑖 𝜎𝑖
In this expression, [𝑅] is the correlation matrix while 𝜎𝑖 is the standard deviation associated with 𝑥𝑖 . Based
on the reliability index, the probability of failure is approximated as follows:

𝑝𝑓 ≈ 1 − 𝜑(𝛽) (3)

Where 𝜑() is the cumulative distribution function of the standard normal variables. Building on this, Low,
B. K. and Tang, 1997; Low, Bak K. and Tang, 2004 developed a constrained optimization framework using
a varying non-dimensional number 𝑛𝑖 associated with 𝑥𝑖 . Eq. (2) is rewritten as:

𝛽 = 𝑚𝑖𝑛𝑋∈𝐹 √[𝑛]𝑇 [𝑅]−1 [𝑛] (4)

Here [𝑛] is a vector of 𝑛𝑖 , varying in constrained optimization. The corresponding values of 𝑥𝑖 can be
calculated as follows:

𝑥𝑖 = 𝐹 −1 [𝜑(𝑛𝑖 )] (5)

Where 𝐹(𝑥) is the original CDF evaluated at 𝑥.

3. Hydraulic fracture models


Various models have been utilized for the hydraulic fracturing process. Under overburden pressure at great
depth, the horizontal plane has a minimum in situ stress. Due to this, vertical fractures are generated
perpendicular to this plane. This growth of vertical fracture depends on two factors: Variability in material
properties and variability of in situ stress. The initial models such as KGD (Geertsma & De Klerk, 1969)and
PKN (Nordgren, 1972) models operate under the assumption of constant fracture height. In this work, KGD
model has been used to conduct reliability analysis of the fracturing process.
In KGD model, the fracture is assumed to be a channel with opening width 𝑤. Assuming there is no fluid
leak-off, the model provides the following expressions. Fracture length is given by:
1/6
8𝐺𝑄 3
𝐿 = 0.48 [ ] 𝑡 2/3 (6)
(1 − 𝜈)𝜇

Maximum fracture opening width is defined as:


1/6
8(1 − 𝜈)𝑄 3 𝜇
𝑤0 = 1.32 [ ] 𝑡1/3 (7)
𝐺

Wellbore pressure can be defined as:


1/4
2𝐺 3 𝑄𝜇
𝑝𝑤 = 𝜎𝐻𝑚𝑖𝑛 + 0.96 [ ] (8)
(1 − 𝜈)3 𝐿2

The breakdown pressure 𝑝𝑏 is the minimum pressure to initiate the fracture process and the expression for
which is given by Hubbert and willis as :
𝑝𝑏 = 3𝜎𝐻𝑚𝑖𝑛 − 𝜎𝐻𝑚𝑎𝑥 + 𝑇 (9)

In these expressions, 𝐺 and 𝜈 are shear modulus and Poisson’s ratio of the rock, respectively. 𝑄 is the fluid
injection rate, 𝜇 is the fluid viscosity, and 𝑡 is the time of the operation. 𝜎𝐻𝑚𝑖𝑛 and 𝜎𝐻𝑚𝑎𝑥 are the minimum
and maximum in situ horizontal stress, while 𝑇 is the tensile strength of the rock.

4. Hydraulic fracturing reliability analysis


The Hydraulic fracture model provides an expression for wellbore pressure and fracture dimensions, thus
can be used to evaluate the performance of hydraulic fracturing. In the context of wellbore pressure,
hydraulic fracturing failure is caused when wellbore pressure is less than the breakdown pressure. The limit
state functions for fracture dimension is constructed by defining limiting values of fracture length and
width. The limit state functions are expressed as follows:
𝑔1 (𝑥) = 𝑝𝑤 − 𝑝𝑏 (10)

𝑔2 (𝑥) = 𝐿 − 𝐿𝑙 (11)

𝑔3 (𝑥) = 𝑤 − 𝑤𝑙 (12)

In these expressions, 𝐿𝑙 and 𝑤𝑙 denotes limiting values of fracture length and width. When the limit state
function is negative, the failure occurs in the fracturing process.
In order to perform reliability analysis, the elastic parameters of the rock (𝐺 and 𝜈) and in situ stresses
(𝜎𝐻𝑚𝑖𝑛 and 𝜎𝐻𝑚𝑎𝑥 ) are modelled as random variables. Statistical description of these parameters is given
in Table 1. The other parameters of hydraulic fracturing are considered to be deterministic, and are listed
in Table 2.
Table 1. Statistical description of random variables

Random variables Distribution Mean Std


Bulk Modulus, G (Pa) Normal 1.00E+10 1.00E+09
Poisson's Ratio, ν Normal 3.00E-01 3.00E-02
σHmax (Pa) Normal 3.00E+07 3.00E+06
σHmin (Pa) Normal 1.00E+07 1.00E+06

Table 2. Deterministic model parameters.

Parameters Value
Tensile Strength, 𝑇 (MPa) 2
Injection Rate, 𝑄 (m3/s) 0.0005
Fluid Viscosity, 𝜇 (Pa.s) 0.0001
𝐿𝑙 (m) 10
𝑤𝑙 (m) 0.0003

Utilizing the constrained optimization procedure developed by Low and Tang, the reliability analysis is
conducted on limit state functions in Eq. (10-12). A built-in constrained optimization tool ‘Solver’ in
Microsoft Excel is utilized. Table 3 shows the excel spreadsheet associated with the limit state function in
Eq. (10). The optimization procedure is used to minimize reliably index in cell A13 by varying cells F7 to
F10, subject to the constraint of limit state function in cell C13. As shown in the Table 3, the reliability
index is 2.993 for injection time 5s and design point listed in cells E7 to E10. This results in 0.137%
probability of failure. Similar optimization routine is conducted for limit state function in Eq. (11) and Eq.
(12), and the results of the optimization are presented in Table 4 and Table 5, respectively.
In order to understand the effect of the tensile strength of the rocks, the reliability index is calculated for
the varying value of tensile strength T in Eq. (9). The Fig 1 shows the variation of reliability index and the
probability of failure for different values of tensile strength. The results show that the difficulties in the
hydraulic fracturing process increase with the increase of tensile strength of rock.
An investigation is also conducted for the effect of the injection time. The limit state function in Eq. (10) is
analyzed with six different values of injection𝑡, 1s, 2s, 5s, 10s, 20s and 50s. The Fig 2 show the reliability
index and probability of failure for these six cases. The results show that the hydraulic fracturing process
improves with the increase of injection time. The limit state function associated with fracture length and
fracture width are also analyzed for varying injection rate and fluid viscosity. Fig 3 shows the reliability
index for varying injection rate from limit state function in Eq. (11) and Eq. (12). The results show that
both the reliability of fracture length and width increases with the injection rate. The Fig 4 shows the
reliability index for varying fluid viscosity. The reliability index of fracture length from Eq. (11) decreases
with increasing fluid viscosity, while the reliability index of fracture width from Eq. (12) increases with
increasing fluid viscosity. These results show that the performance of the hydraulic fracturing process can
be improved by adjusting the model parameters such as inject rate and fluid viscosity. Hence, the reliability
analysis provides additional information and data for reliability based design of hydraulic fracture. This
method provides a mean to account for model uncertainties and more rational approach than deterministic
design.
Table 3. Optimization procedure for limit state function associated with wellbore pressure in Eq. (10)

A B C D E F
1 Tensile Strength, Injection Rate, Q Fluid Viscosity, Injection Limiting L,
T (MPa) (m3/s) μ (Pa.s) Time, t (s) Ll
2 2.00E+06 5.00E-04 1.00E-04 1.00E+01 1.00E+01
3 Breakdown Wellbore Crack Length, Crack Width, Limiting w,
Pressure, pb (Pa) Pressure, pw (Pa) L (m) w (m) wl
4 1.44E+07 1.44E+07 1.60E+01 3.95E-04 3.00E-04
5
6 Variables Distribution Mean Std xi ni
7 Bulk Modulus, G Normal 1.00E+10 1.00E+09 9.80E+09 -1.96E-01
(Pa)
8 Poisson's Ratio, ν Normal 3.00E-01 3.00E-02 2.98E-01 -8.19E-02
9 σHmax (Pa) Normal 3.00E+07 3.00E+06 2.25E+07 -
2.49E+00
10 σHmin (Pa) Normal 1.00E+07 1.00E+06 1.17E+07 1.66E+00
11
12 β Pf g1(x) g2(x) g3(x)
13 3.00E+00 1.37E-03 0.00E+00 6.05E+00 9.49E-05
14
Table 4. Optimization procedure for limit state function associated with fracture length in Eq. (11)

A B C D E F
1 Tensile Injection Fluid Injection Limiting L, Ll
Strength, T Rate, Q (m3/s) Ciscosity, μ Time, t (s)
(MPa) (Pa.s)
2 2.00E+06 5.00E-04 1.00E-04 5.00E+00 1.00E+01
3 Breakdown Wellbore Crack Length, Crack Width, Limiting w, wl
Pressure, pb Pressure, pw L (m) w (m)
(Pa) (Pa)
4 2.00E+06 1.21E+07 1.00E+01 3.17E-04 3.00E-04
5
6 Variables Distribution Mean Std xi ni
7 Bulk Modulus, Normal 1.00E+10 1.00E+09 9.27E+09 -7.30E-01
G (Pa)
8 Poisson's Normal 3.00E-01 3.00E-02 2.91E-01 -2.87E-01
Ratio, ν
9 σHmax (Pa) Normal 3.00E+07 3.00E+06 3.00E+07 -3.93E-07
10 σHmin (Pa) Normal 1.00E+07 1.00E+06 1.00E+07 -6.02E-07
11
12 β Pf g1(x) g2(x) g3(x)
13 7.84E-01 2.16E-01 1.01E+07 2.38E-08 1.68E-05
14

Table 5. Optimization procedure for limit state function associated with fracture width in Eq. (12)

A B C D E F
1 Tensile Injection Fluid Injection Limiting L, Ll
Strength, T Rate, Q (m3/s) Ciscosity, μ Time, t (s)
(MPa) (Pa.s)
2 2.00E+06 5.00E-04 1.00E-04 5.00E+00 1.00E+01
3 Breakdown Wellbore Crack Length, Crack Width, Limiting w, wl
Pressure, pb Pressure, pw L (m) w (m)
(Pa) (Pa)
4 2.00E+06 1.28E+07 1.06E+01 2.99E-04 3.00E-04
5
6 Variables Distribution Mean Std xi ni
7 Bulk Modulus, Normal 1.00E+10 1.00E+09 1.22E+10 2.19E+00
G (Pa)
8 Poisson's Normal 3.00E-01 3.00E-02 3.36E-01 1.21E+00
Ratio, ν
9 σHmax (Pa) Normal 3.00E+07 3.00E+06 3.00E+07 -4.86E-07
10 σHmin (Pa) Normal 1.00E+07 1.00E+06 1.00E+07 -5.13E-07
11
12 β Pf g1(x) g2(x) g3(x)
13 2.50E+00 6.22E-03 1.08E+07 5.81E-01 -6.06E-07
14
Figure 1. Reliability index and probability of failure for varying tensile strength
Figure 2. Reliability index and probability of failure for varying injection time
Figure 3. Reliability index for varying injection rate

Figure 2. Reliability index for varying fluid viscosity


5. Conclusion
A reliability-based analysis of hydraulic fracturing is conducted to incorporate uncertainties. A constrained
optimization process is utilized to obtain the reliability index in FORM. Three limit state function associated
with wellbore pressure, fracture length and width are considered. The KGD model is used to simulate the
hydraulic fracture process. Optimization routine in Microsoft Excel is used to calculate the reliability index.
Reliability analysis method provides an efficient and effective scientific approach for the uncertainty
analysis of hydraulic fracturing. This study may help to have a good understanding of the uncertainty and
reliability-based design for hydraulic fracturing. The major finding of this study consist:

 The study quantifies the uncertainties associated with hydraulic fracturing models and enables the
identification of the most influential parameters. The sensitivity of parameter can be used to reduce
the uncertainties in the model performance.
 The reliability analysis calculates the dimension of hydraulic fracture and the probability associated
with the fracture dimension. This information regarding the fracture can be used to determine
hydraulic fracture design.
 The study also investigates the effect of model parameters such as tensile strength of rock, injection
time, injection rate and fluid viscosity on the performance of the fracture model.
6. References

1. Boschi, C., A. Dini, L. Dallai, G. Ruggieri and G. Gianelli. 2009. Enhanced CO2-mineral
sequestration by cyclic hydraulic fracturing and Si-rich fluid infiltration into serpentinites at
Malentrata (Tuscany, Italy). Chemical Geology 265: 209-226.
2. Desroches, J. 1995. Stress testing with the micro-hydraulic fracturing technique—focus on fracture
reopening. Proceedings of the 35th US symposium on rock mechanics. Rotterdam: Balkema, pp.
217-223.
3. Geertsma, J. and F. De Klerk. 1969. A Rapid Method of Predicting Width and Extent of
Hydraulically Induced Fractures. SPE-2458-PA 21: 1571-1581.
4. Harlow, F. H. and W. E. Pracht. 1972. A theoretical study of geothermal energy extraction. Journal
of Geophysical Research 77: 7038-7048.
5. Hasofer, A. M. 1974. Exact and Invariant Second-Moment Code Format. Journal of Engineering
Mechanics 100: 111-121.
6. Heo, W., W. Lee and D. S. Lee. 2015. Hydraulic fracturing design for coalbed methane in Barito
Basin, Indonesia. Geosystem Engineering 18: 151-162.
7. Hoek, E. 1998. Reliability of Hoek-Brown estimates of rock mass properties and their impact on
design. International Journal of Rock Mechanics and Mining Sciences 35: 63-68.
8. Li, H.-Z. and B. K. Low. 2010. Reliability analysis of circular tunnel under hydrostatic stress field.
Computers and Geotechnics 37: 50-58.
9. Li, Q., B. Lin and C. Zhai. 2015. A new technique for preventing and controlling coal and gas
outburst hazard with pulse hydraulic fracturing: a case study in Yuwu coal mine, China. Natural
Hazards 75: 2931-2946.
10. Low, B. K. 1997. Reliability analysis of reinforced embankment on soft ground. Canadian
Geotechnical Journal 34: 672-685.
11. Low, B. K. and W. H. Tang. 1997. Efficient Reliability Evaluation Using Spreadsheet. Journal of
Engineering Mechanics 123: 749-752.
12. Low, B. K. and W. H. Tang. 2004. Reliability analysis using object-oriented constrained
optimization. Structural Safety 26: 69-89.
13. Lü, Q. and B. K. Low. 2011. Probabilistic analysis of underground rock excavations using response
surface method and SORM. Computers and Geotechnics 38: 1008-1021.
14. Lukyanov, A. and N. Chugunov. 2015. Uncertainty analysis for hydraulic fracturing modeling in
porous media containing oil and gas. AIP Conference Proceedings, p. 630002. AIP Publishing.
15. Mollon, G., D. Dias and A.-H. Soubra. 2009. Probabilistic analysis of circular tunnels in
homogeneous soil using response surface methodology. Journal of Geotechnical and
Geoenvironmental Engineering 135: 1314-1325.
16. Moschovidis, Z., R. Steiger, R. Peterson, N. Warpinski, C. Wright, E. Chesney, J. Hagan, A. Abou-
Sayed, R. Keck and M. Frankl. 2000. The Mounds Drill-Cuttings Injection Field Experiment: Final
Results and Conclusions. IADC/SPE Drilling Conference. Society of Petroleum Engineers.
17. Nordgren, R. P. 1972. Propagation of a Vertical Hydraulic Fracture. SPE-3009-PA 12: 306-314.
18. Oreste, P. 2005. A probabilistic design approach for tunnel supports. Computers and Geotechnics
32: 520-534.
19. Perkins Jr, T. and L. Kern. 1961. Widths of Hydraulic Fractures. JPT: 937-49. Trans.
20. Zhao, H., Z. Li, C. Zhu and Z. Ru. 2018. Reliability analysis models for hydraulic fracturing.
Journal of Petroleum Science and Engineering 162: 150-157.
21. Zhao, H., Z. Ru, X. Chang, S. Yin and S. Li. 2014. Reliability analysis of tunnel using least square
support vector machine. Tunnelling and Underground Space Technology 41: 14-23.
22. Zheltov, A. K. 1955. 3. Formation of vertical fractures by means of highly viscous liquid. 4th world
petroleum congress. World Petroleum Congress.

You might also like