Professional Documents
Culture Documents
Rock Deformation Models and Fluid Leak-Off in Hydraulic Fracturing
Rock Deformation Models and Fluid Leak-Off in Hydraulic Fracturing
Accepted 2013 May 13. Received 2013 May 10; in original form 2012 July 31
GJI Marine geosciences and applied geophysics
SUMMARY
1514
C The Authors 2013. Published by Oxford University Press on behalf of The Royal Astronomical Society.
Rock deformation and hydraulic fracturing 1515
coefficients with exponential (Mukherjee et al. 1991; Barree 1998) Compaction is a fundamental process in which the pore space of
or power-law dependence on the effective pressure rise (Castillo sedimentary rocks decreases in response to an applied load. It may
1987; Meyer & Jacot 2000). Warpinski (1991) considered fissured result from the closure of microcracks, grain rearrangement and
rocks with pressure-dependent permeability and derived a leak-off pore collapse, or in some cases from elastic reversible deformation
coefficient that is a power-law function of the logarithm of stress. of the rock. Inelastic irreversible deformation contributes to com-
Pressure-dependent leak-off also results from the model of Fan & paction processes in sedimentary rocks, especially in weak and frac-
Economides (1995), where the dependence is given implicitly as tured media such as shale, soft limestone, sandstone, coalbeds and
a solution to the coupled fluid flow equations in the invaded and unconsolidated sands (Byerlee 1978; Barton et al. 1985; Franquet &
reservoir zones. These authors considered non-Newtonian fractur- Economides 1999; Connolly & Podladchikov 2000; Gil et al. 2005;
ing fluids and assumed a classical Kozeny–Carman type relationship Yao et al. 2010). Inelastic rate-independent (plastic) deformation
between permeability and porosity. All these models assume, how- leads to the dependence of the elastic moduli on effective pressure
ever, that porosity and total compressibility of the fractured rock do and loading history. These phenomena are observed even at mod-
not change with increasing pressure. The effects of poroelasticity erate effective pressures and are enhanced by the presence of shear
on fracture growth were examined in the works of Detournay et al. stresses. Experimental results of Franquet & Economides (1999)
M AT H E M AT I C A L M O D E L
Indices ‘i’ and ‘r’ refer to the invaded and reservoir zones; p
A N A LY T I C A L R E S U LT S is the fluid pressure rise on the fracture
√ face. The interface between
the two zones moves as s(t) = 2η k0 t/ϕ0 , where η is given as a
The solution to eqs (1) and (2), which describe the coupled processes
solution to an algebraic equation:
of fluid flow and reservoir compaction, can be simplified for linear
c̃i μr
−1
and non-linear elastic rheologies if a similarity transform is applied
(see Appendix C). Use of a similarity solution implies that we βη π 1 − erf μr c̃r η =
are concerned with steady-state fluid infiltration, and we neglect the μi c̃r
initial stages of leak-off. The numerical solution to the full equations
c̃i √
shows that this approximation does not introduce significant error (exp (β p) − 1) − βη πexp c̃i μi η2 erf μi c̃i η . (14)
(Fig. 3). Corrections accounting for the initial transient stages of μi
fluid leak-off can be accounted for by spurt loss (Economides &
Nolte 2000). Analytical solutions to eqs (1) and (2), accounting for According to this analytical solution, most of the pressure and
a moving boundary between the invaded and reservoir zones can be porosity drop occurs in the invaded region (Fig. 4). However, the
obtained for the special case when the permeability exponent n = presence of the reservoir region accounts for the speed of propa-
1, and the initial reservoir porosity ϕ0 is small (see Appendix C for gation of the fluid invasion front. Without the reservoir region, the
details). Defining normalized porosity f = ϕ/ϕ0 , then the invaded pressure diffusion front is smooth and moves much faster (dashed
zone has line in Fig. 4). Leak-off of the fracturing fluid changes the poros-
√ ity of the formation, and these changes, which might erroneously
erf ci∗ ξi be considered insignificant, lead to a pressure-dependent leak-off
f i = exp (β p) + ( f b − exp (β p)) √ , (11)
erf μi c̃i η coefficient and also vary the rock permeability.
1518 V. M. Yarushina, D. Bercovici and M. L. Oristaglio
C O M PA R I S O N O F T H E M O D E L T O
Figure 10. Total leak-off coefficient per unit area in linear and non-linear F I E L D D ATA
poroelastic rocks as a function of a pressure drop during hydraulic fracturing. Leak-off due to rock deformation is readily observed when the nor-
Conventional total leak-off coefficient is plotted for reference as a dotted malized leak-off coefficient is plotted versus pressure on a semi-log
line.
plot. All three types of rheology considered herein give their own
distinctive features. Both linear elastic and elastoplastic deforma-
tion give leak-off versus pressure relations with negative curvature
so that d2 ln C/d p2 ≤ 0. However, the relation with elastoplas-
porosity-dependent permeability in the following manner:
ticity flattens at higher pressures leading to d2 ln C/d p2 = 0 just
k0 ϕ0 p (1 + exp (nβ p)) (1 + exp (β p)) before full pore collapse leading to rock failure (Fig. 12). Non-
Ci = . (20) linear elastic deformation, as described by (6) and associated with
8μi
microcrack opening, yields d2 ln C/d p2 ≤ 0 at small pressures
In gas reservoirs most of the porosity increase occurs in the and d2 ln C/d p2 ≥ 0 at moderate to high pressures. The initial
invaded zone, thus the correction to Cr is not needed. For n > 10, stages characterized by negative curvature and associated with rapid
growth of ln(C/C0 ) (Fig. 13) are due to a jump from almost un-
k0 ϕ0 p (1 + exp (β p)) (1 + exp (nβ p)) detectable leak-off at small pressures (C = 9.25 × 10−9 ms−0.5 at
Ci =
16μi 2 p = 0.1 MPa for n = 30) to one that is orders of magnitude
larger at later stages of deformation (C = 3.83 × 10−6 ms−0.5 at
⎞
p = 15 MPa). At intermediate pressures d2 ln C/d p2 ≈ 0, in
(1 + exp (β p))n+1
+ ⎠. (21) which case the leak-off curves can be approximated as inclined
2n straight lines.
Rock deformation and hydraulic fracturing 1521
Figure 12. Leak-off ratio plot for linear elastic and elastoplastic rocks for
n = 5 and Y = 3.46 MPa (other parameters used in these calculations are C O N C LU S I O N S
summarized in Table 2). Both linear elastic and elastoplastic deformation
lead to concave down curves but the curve for elastoplastic material flattens
When fluid loss into reservoir rock during hydraulic fracturing is
at high stress just before full pore collapse. modelled via pressure diffusion coupled with porosity evolution,
the pressure drop occurs in the narrow invaded zone with dynamic
characteristics such as time-varying thickness, permeability and
compressibility. Inclusion of pore compressibility and porosity-
dependent permeability into the flow model leads to pressure-
dependent leak-off that is more pronounced than in the classical
model. The inferred plasticity (irrecoverable deformation) also al-
lows an estimation of the permeability increase caused by the frac-
turing process. The ability of the rock to dilate in response to the
fluid invasion increases the leak-off rate.
Of the three different types of rock rheologies considered—
namely, linear poroelastic, non-linear poroelastic and poroelasto-
plastic—power-law or frictional elastoplasticity give a better
description of naturally fractured and soft rocks. Poroelastic be-
haviour, on the other hand, may be expected to give the best de-
scription of conventional reservoir. Irrecoverable time-independent
(plastic) deformation gives the highest values of leak-off. Whereas
Carter’s classical model can accurately predict leak-off in poroelas-
tic rocks with low porosity and near-constant permeability (depend-
ing weakly on porosity). New analytical and numerical solutions
Figure 13. Leak-off ratio plot for non-linear elastic rocks. At small pres- show that leak-off in rocks with more strongly porosity-dependent
sures leak-off is orders of magnitude smaller than at higher pressures and permeability or more complicated rheological behaviour will dif-
can be neglected. Note the concave upward segments at higher pressures. fer significantly from the classical theory. The inclusion of rock
Parameters used in these calculations are summarized in Table 2. compaction and decompaction may be important for describing
naturally fractured and tight gas reservoirs as well as CO2 injection
sites and enhanced geothermal systems; it may also be important to
Field examples (Mukherjee et al. 1991; Barree 1998; John- model soft and poorly consolidated formations whose mechanical
son et al. 1998) demonstrate that all three types of deforma- behaviour departs from the simple elastic laws and/or have strongly
tional mechanisms can occur (Fig. 14). Here, we consider sur- porosity-dependent permeability.
face pressure data from pre-fracturing injection tests in the Upper
Lewis sand, Mesa Verde group, Wyoming (Fig. 14a), which can
be characterized as a tight reservoir. These data exhibit pressure-
independent leak-off at low surface pressure (up to 11 MPa) in
which the deformational mechanisms give negligible leak-off. As AC K N OW L E D G E M E N T S
pressure increases, non-linear elastic deformation due to opening Authors are grateful to Nina Simon for helpful discussions. This
of pre-existing cracks (solid red line in Fig. 14a) gives good ap- work was supported in part by a grant (DE-FE0004375) from
proximation to a nearly linear segment on the leak-off curve with the National Energy Technology Laboratory of US Department of
d2 ln C/d p2 ≈ 0. At yet higher pressures, the leak-off versus pres- Energy.
1522 V. M. Yarushina, D. Bercovici and M. L. Oristaglio
1. Nomenclature.
Symbol Meaning Units
c f , cs , ct Compressibilities of fluid and solid, total compressibility MPa−1
Ci , Cr , Ct Leak-off coefficients for the invaded and reservoir zones, total leak-off coefficient ms−1/2
Ci0 , Cr0 , Ct0 Carter’s leak-off coefficients ms−1/2
cϕ Pore compressibility MPa−1
g Gravity ms−2
Gs Solid shear modulus MPa
k0 , k Formation permeability before and during hydraulic fracturing m2
n Porosity exponent for k
p f , pt Fluid pressure, total (lithostatic) pressure MPa
pe Effective pressure, pe = pt − p f MPa
p0 Initial formation effective pressure MPa
p Pressure drop during hydraulic fracturing MPa
q Porosity exponent for effective pressure
t Time s
v f , vs Fluid and solid velocity ms−1
x Distance from the fracture face m
Y Cohesion MPa
β Coefficient of pore compressibility MPa−1
μ Fluid viscosity Pas
ξ Similarity variable, eq. (C2)
ρ f , ρs Fluid and solid density kgm−3
ϕ0 , ϕ Formation porosity before and during hydraulic fracturing
φ Friction angle
ψ Dilation angle
2. Simulation input.
REFERENCES
Parameter Value used
Athy, L.F., 1930. Density, porosity and compaction of sedimentary rocks,
Initial formation porosity 0.1 Bull. Am. Assoc. Pet. Geol., 14, 1–24.
Initial formation permeability, m2 1 × 10−19 Barree, R.D., 1998. Applications of pre-frac injection/falloff tests in fis-
Elastic pore compressibility β, 1/MPa 5 × 10−3 sured reservoirs—field examples, paper SPE 39932, in SPE Rocky Moun-
Fracturing fluid viscosity, Pa s 1 × 10−3 tain Regional/Low-Permeability Reservoirs Symposium and Exhibition,
Reservoir fluid viscosity, Pa s 1 × 10−6 Denver, CO, USA, pp. 277–288.
Fracturing fluid compressibility, 1/MPa 0.45 × 10−3 Barton, N., Bandis, S. & Bakhtar, K., 1985. Strength, deformation and
Reservoir fluid compressibility, 1/MPa 10 conductivity coupling of rock joints, Int. J. Rock Mech. Min., 22, 121–
Compaction power law exponent q 0.032 140.
Friction angle φ, rad π/6 Bercovici, D. & Ricard, Y., 2003. Energetics of a two-phase model of litho-
Dilation angle ψ, rad 0 spheric damage, shear localization and plate-boundary formation, Geo-
Cohesion, MPa 7.5 phys. J. Int., 152, 581–596.
Rock deformation and hydraulic fracturing 1523
Biot, M.A., Masse, L. & Medlin, W.L., 1986. A two-dimensional theory of Jaeger, J.C., Cook, N.G.W. & Zimmerman, R.W., 2007. Fundamentals of
fracture propagation, SPE Prod. Eng., 1, 17–30. Rock Mechanics, 4th edn, Blackwell Pub., Malden, MA.
Byerlee, J., 1978. Friction of rocks, Pure appl. Geophys., 116, 615–626. Ji, L.J., Settari, A. & Sullivan, R.B., 2009. A novel hydraulic fracturing
Carroll, M.M., 1980. Compaction of dry or fluid-filled porous materials, J. model fully coupled with geomechanics and reservoir simulation, SPE J.,
Eng. Mech. Div.-ASCE, 106, 969–990. 14, 423–430.
Carter, J.P., Booker, J.R. & Yeung, S.K., 1986. Cavity expansion in cohesive Johnson, R.L., Dunn, K.P., Bastian, P.A., Hopkins, C.W. & Conway, M.W.,
frictional soils, Geotechnique, 36, 349–358. 1998. Qualifying hydraulic fracturing effectiveness in tight, naturally frac-
Castillo, J.L., 1987. Modified fracture pressure decline analysis including tured reservoirs by combining three-dimensional fracturing and reservoir
pressure-dependent leakoff, paper SPE 16417, in SPE/DOE Low Perme- simulators, paper SPE 49048, in SPE Annual Technical Conference and
ability Symposium, Denver, CO, USA, pp. 273–281. Exhibition, New Orleans, LA, USA, pp. 1–23.
Cleary, M.P., 1979. Rate and structure sensitivity in hydraulic fracturing Kelemen, P.B., Matter, J., Streit, E.E., Rudge, J.F., Curry, W.B. & Blusztajn,
of fluid-saturated porous formations, paper SPE 79–0127, in 20th U.S. J., 2011. Rates and mechanisms of mineral carbonation in peridotite: nat-
Symposium on Rock Mechanics, American Rock Mechanics Association, ural processes and recipes for enhanced, in situ CO2 capture and storage,
Austin, TX, pp. 127–142. Annu. Rev. Earth planet. Sci., 39, 545–576.
Connolly, J.A.D. & Podladchikov, Y.Y., 2000. Temperature-dependent vis- King, G., 2012. Hydraulic fracturing 101, paper SPE 152596, in SPE Hy-
coelastic compaction and compartmentalization in sedimentary basins, draulic Fracturing Technology Conference, Society of Petroleum Engi-
Vajdova, V., Baud, P. & Wong, T.F., 2004. Compaction, dilatancy, and where
failure in porous carbonate rocks, J. geophys. Res., 109, B05204, d ∂ ∂
doi:10.1029/2003JB002508. = + vs .
Valko, P. & Economides, M.J., 1994. Propagation of hydraulically induced
dt ∂t ∂x
fractures—a continuum damage mechanics approach, Int. J. Rock Mech. Substitution of Darcy’s law, eq. (A2) and rheological eqs (A3) and
Min. Sci. Geomech. Abstr., 31, 221–229. (A4) into eq. (A5) gives
Valko, P.P. & Economides, M.J., 1999. Fluid-leakoff delineation in high-
ρf d pe ρ f cs d p f ϕρ f cs d pt
permeability fracturing, SPE Prod. Facil., 14, 117–130. − ϕcϕ + ϕρ f c f + ϕ 2 −
Warpinski, N.R., 1991. Hydraulic fracturing in tight, fissured media, J. 1−ϕ dt 1−ϕ dt 1 − ϕ dt
Petrol. Tech., 43, 146–209.
∂ ρ f k ∂p f
Yang, Y.L. & Aplin, A.C., 2010. A permeability-porosity relationship for − = 0. (A6)
mudstones, Mar. Petrol. Geol., 27, 1692–1697. ∂x μ ∂x
Yao, Y., Gosavi, S.V., Searles, K.H. & Ellison, T.K., 2010. Cohesive fracture We approximate the total pressure with lithostatic stress pt =
mechanics based analysis to model ductile rock fracture, paper ARMA10–
−ρs gz, where z is the depth. Hence, d pt /dt = 0 and d p f = −d pe .
140, in 44th US Rock Mechanics Symposium and 5th U.S.-Canada Rock
With this in mind, we can rewrite the hydraulic equation for slightly
Mechanics Symposium, Salt Lake City, UT, USA, pp. 1–8.
dϕ = −ϕcϕ d pe , (A4)
will develop around structural heterogeneities in the rock matrix. The similarity transformation
We define irreversible time-independent deformation of rock that √
occurs after stresses reach the failure criterion as plastic irrespec- ξ = x/ at, (C2)
tive of the micromechanism of such deformation, which is often the
where a = 4k0 / c f μϕ0 , can be applied to eq. (C1) resulting in
interplay of several micromechanical processes, including microc-
racking, grain crushing, pore collapse and frictional sliding on an c̃ dϕ d ϕ0 c f k dϕ
−2ξ = , (C3)
array of fissures (Vajdova et al. 2004; Fortin et al. 2007). This be- cϕ dξ dξ ϕ cϕ k0 dξ
haviour is characterized by path-dependence and hysteresis. We use
simple and widely used Mohr–Coulomb criterion, which describes where we have made use of eq. (10). Introducing the normalized
not only failure of intact rock that takes place along a plane due to porosity
a shear stress in that plane, as in a fault, but also describes yielding f = ϕ/ϕ0 , (C4)
in rocks and soils (Yu 2006; Jaeger et al. 2007):
and substituting eq. (3) for permeability into (C3) yields
κσθ − σr = Y ∗ + (1 − κ) p f , (B1)
c̃ d f d c f n−1 d f
where κ and Y ∗ are the following functions of cohesion, Y, and
κ = (1 + sinφ) / (1 − sinφ) , Next, we follow the approach of Fan & Economides (1995) and
(B2) consider simultaneously the invaded and reservoir regions separated
∗
Y = 2Y cosφ/ (1 − sinφ) . by a moving boundary, as fracturing fluid progresses into the forma-
In response to an increase in fluid pressure, the failure region spreads tion. The materials in both regions have the same functional form of
into the shell. Outside the failure region, material is still in the permeability (3) and pore space compressibility. However, the fluid
elastic state. Due to symmetry, the elastoplastic boundary will be a viscosities and compressibilities are different, which influences the
cylindrical surface with radius c. The solution to the force balance pressure and porosity evolution in each region. In the following
equation and specific elastic or plastic constitutive laws yield the we use subscripts ‘i’ and ‘r’ to distinguish the porosity, pressure,
stresses and velocities within the elastic and plastic parts of the shell permeability, fluid viscosity and compressibility in the invaded and
(Carter et al. 1986). reservoir zones. In each of the two regions, (C5) must be satisfied.
Further, we assume that shell is composed of an incompressible The initial and boundary conditions are:
solid material based on the scaling argument that the compressibility
of solid mineral grains is negligible relative to total compressibility. pei = pfrac at x = 0 and t > 0, (C6)
The porosity equation is obtained from the equation:
per = p0 at t = 0 and x > s(t), (C7)
dϕ dV p
= ,
ϕ Vp where x = s(t) is the moving interface between the regions. At the
interface, continuity of pressures and flow velocities is assumed,
where V p is the pore volume. For a cylindrical void, V p = πa , 2
which leads to two more conditions
therefore,
1 dϕ vr pei = per at x = s(t), (C8)
=2 .
ϕ dt r r=a
ki ∂ pei kr ∂ per ds
Substitution of the plastic velocity vr into the last equation gives = = at x = s(t). (C9)
ϕi μi ∂ x ϕr μr ∂ x dt
the effective compressibility in elastoplastic rocks in the form
Recasting boundary and initial conditions in terms of the new
cϕ = β p̃ −q , (8)
similarity variables must be done separately for each type of rock
where p̃ = (1 + sinφ)(1 + tanφpe /Y ), q = (1 + cscφ)/(1 + sinψ) rheologies.
and Ys = Y cosφ/(1 + sinφ).
This equation holds only if −Y cotφ ≤ pe < −Ys . The upper limit
corresponds to the plasticity onset and the lower limit gives pressure C1: Linearly elastic rocks
for the onset of fully plastic pore collapse when the failure region
occupies the entire body. Physically, this limiting state is unstable First, we consider purely linear elastic rocks in which the effective
and cannot be described by the same type of model. compressibility is given by (4). In this case (C5) reduces to
df d df
− 2ξ c∗ = f n−1 , (C10)
APPENDIX C: SIMILARITY dξ dξ dξ
T R A N S F O R M AT I O N T O H Y D R AU L I C
where
E Q UAT I O N C O U P L E D W I T H
D E F O R M AT I O N c̃ β
c∗ = = + 1. (C11)
cf (1 − ϕ0 f ) c f
Eqs (1) and (2) describe fluid flow in a deforming porous rock,
and constitute a system of two equations for two unknowns. For Using (5) and (C4) conditions (C6)–(C9) can be rewritten as
convenience, we can eliminate the effective pressure in eq. (2) using
f i = exp (β p) at ξ = 0, (C12)
eq. (1) and express the hydraulic equation in terms of porosity as
follows: in the invaded region and
ct dϕ ∂ k ∂ϕ
= . (C1) f r = 1 at ξ = ∞, (C13)
ϕcϕ dt ∂ x μϕcϕ ∂ x
1526 V. M. Yarushina, D. Bercovici and M. L. Oristaglio
in the reservoir region. At the moving boundary between the regions The leak-off rate is given as
f i = f r = f inter , (C14) k ∂ pe f in−1 k0 c f,i ϕ0 1 d f i
v= =− √ . (C20)
μi ∂ x x=0 2β μi t dξ ξ =0
c f,i d f i c f,r d f r ϕ0 ds √ Substitution of (C16) into (C20) gives the total leak-off
f in−2 = f rn−2 = −2β t = −2βη,
μi dξi μr dξr k0 dt coefficient
(C15) √ 1 k0 c̃i ϕ0
Ct = v t = exp (β p) η̃,
where p = p0 − pfrac > 0 and η > 0 is an unknown constant. β π μi
√
Similarity variables range between 0 and ξi = c f,i μi η in the in-
√ where
vaded region and between ξr = c f,r μr η and infinity in the reservoir
region. π μi
η̃ = 1 − 1 + βη exp c̃i μi η2
When the initial porosity of the reservoir is small (ϕ0
1), (C10) c̃i
can be integrated in a closed form together with the boundary con-
−1
dξ dξ dξ
× erf cr∗ ξr 1 − erf μr c̃r η , (C17)
where
√
at c f,r μr η < ξr < ∞. To determine the unknown position of the c̃ f 1/q
interface, η, and the unknown porosity at the interface, f inter , we use c∗ = = + 1. (C22)
cf (1 − ϕ0 f ) p0 c f
(C15). The position of the interface is given as an implicit solution
to an algebraic equation: Boundary conditions are:
f i = ( pfrac / p0 )−q at ξ = 0,
c̃i μr
π βη 1 − erf μr c̃r η fr = 1 at ξ = ∞,
μi c̃r
and the moving boundary
c̃i √