You are on page 1of 10

ARMA 21– 1233

Numerical simulation of fractures propagation around the wellbore:


Effect of the surrounding rock's stiffness
Agofack, N. and Cerasi, P.

Downloaded from http://onepetro.org/ARMAUSRMS/proceedings-pdf/ARMA21/All-ARMA21/2479314/arma-2021-1233.pdf by Yangtze University, kai wang on 06 September 2023


SINTEF Industry, Petroleum department, Trondheim, Norway

Copyright 2021 ARMA, American Rock Mechanics Association


This paper was prepared for presentation at the 55th US Rock Mechanics/Geomechanics Symposium held in Houston, Texas, USA, 20-23 June
2021. This paper was selected for presentation at the symposium by an ARMA Technical Program Committee based on a technical and critical
review of the paper by a minimum of two technical reviewers. The material, as presented, does not necessarily reflect any position of ARMA, its
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent
of ARMA is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 200 words; illustrations may not be copied. The
abstract must contain conspicuous acknowledgement of where and by whom the paper was presented.

ABSTRACT: Wells play an important role in subsurface activities such as oil/gas exploitation, carbon dioxide (CO2) and hydrogen
storage. During these activities, wells are submitted to various mechanical and thermal loadings which can lead to the creation of
fractures in the near-well region (cement sheath and rock formation), and thus be potential leakage paths. Analytical and numerical
understanding of the stress distribution that may lead to fractures around the near-well is therefore crucial. Depending on the casing,
cement sheath, formation properties, and the in-situ stresses, the analytical model predicts the induced stresses which can lead to
fracture creation, while the numerical tool is used to model the propagation of these fractures. A modified discrete element method
(MDEM) is used in the numerical simulations. The results show that the fracture creation and propagation not only depend on the
casing pressure, initial in-situ stresses, and pore pressure, but also on the formation rock's mechanical properties such as the Young's
modulus.
observed after mechanical loading on the casing (Anya et
1. INTRODUCTION
al., 2020; Goodwin and Crook, 1992; Skorpa et al., 2019;
Wells are needed in various subsurface activities Skorpa et al., 2018; Vrålstad et al., 2019). The observed
including oil/gas exploitation, greenhouse gas storage, cracks are not only limited within the cement sheath but
etc. The construction of these wells can be simplified as can propagate also into the rock formation nearby.
follows: a hole goes through different formations, a steel
hollow cylinder (called casing) is run into the hole, and a
cement paste in injected into the annular space, which
hardens to become the cement sheath (Figure 1). One of
the main concerns in well construction is to achieve zonal
isolation, which prevents downhole fluids that are under
high pressure to leak and flow up to shallow formations
or to the surface in an uncontrolled manner. However,
during their life, wells are submitted to various
mechanical and thermal loadings which can create radial
fracture in the cement sheath and surrounding, and thus
compromise their integrity. It is therefore important to
investigate the mechanism controlling the fractures
creation and propagation as well as their extension from
the cement sheath to rock formations nearby. It was
shown that the main failure mechanics following loadings
applied to the casing include inner and outer debonding at Figure 1. Well architecture, usually composed of hole, casing
and cement sheath.
the casing/cement and cement/rock interfaces,
respectively, shear damage, radial cracks and axial
More recently, a laboratory experiment on a mini-well
disking (Bois et al., 2012; Bois et al., 2011). Most of these
setup was performed. With this setup it was possible to
failure mechanisms have been confirmed in laboratory
control both the confining (or outer) pressure and the
experiments where crack creation and propagation are
casing (or inner) pressure, as well as the possibility to scan
the setup under pressure with CT (Taghipour et al., 2021). They concluded that the Young's moduli of the cement
A class G cement paste was injected into the annulus and the rock were the key properties controlling the
between the casing and the rock formation (Castlegate cement sheath failure. Gheibi and coworkers (Gheibi et
sandstone in this case). The cement was cured for two al., 2021; Gheibi et al., 2019) used the modified discrete
days at ambient temperature and under 2 MPa of element method (MDEM) to numerically analyze the
confining pressure. In the first experiment (with no effect of the tensile strength of cement and that of the
confining pressure), the confining pressure was released confining pressure on radial fracture propagation. They
after curing, and the casing pressure was increased to showed that under 20 MPa of confining pressure, crack
45 MPa. In the second test, the rock's pore pressure and creation and propagation can only be limited within the
the confining pressure were increased to 5 MPa and cement sheath even at casing pressure as high as 87 MPa.

Downloaded from http://onepetro.org/ARMAUSRMS/proceedings-pdf/ARMA21/All-ARMA21/2479314/arma-2021-1233.pdf by Yangtze University, kai wang on 06 September 2023


10 MPa, respectively. The casing pressure was then The radial cracks, as mostly observed in laboratory
increased to 45 MPa, followed by a reduction of the experiments and simulation, result from the tensile
confining pressure down to 8.5 MPa. A seal rubber sleeve (negative) stress in the material becoming higher (in
was used on the lateral outer surface of the rock to prevent absolute value) than its tensile strength. However, from
communication between the rock's pore fluid and the analytical point of view, the stress distribution around the
confining fluid. The results, given in Figure 2, show borehole is not only a function of the applied casing
continuous fractures from cement sheath into the rock. pressure and in-situ stresses, but also depends on the
More fractures (with smaller opening) are observed in the mechanical properties of the materials behind the casing
experiment under confining pressure, while the (Jaeger et al., 2007). As presented in Figure 1, the well
experiment without confining pressure displays fewer goes through different formations which may have
fractures but with bigger opening. There also seems to be different mechanical properties. The effect of rock
the formation of a micro-annulus at the casing/cement mechanical properties (e.g., stiffness) on the radial crack
interface (possibly during the confining pressure release creation and propagation both in the cement sheath and
after curing, see Figure 2a). the rock itself still needs to be properly investigated and
is therefore the main aim of this paper.

In the following sections, a brief description of the


numerical code is given. The analytical description of
elastic equations of stress and strains distributions around
the well will be given in section §3. The simulation results
and discussion will then be given is section §4, and the
paper will end with some conclusions.

2. NUMERICAL DESCRIPTION
Figure 2. Radial fractures due to casing pressure and after CT
The modified discrete element method (MDEM) code,
scan (under pressure): (a) with no confining pressure, (b)
confining pressure of 8.5 MPa (Taghipour et al., 2021).
developed at SINTEF, is used to simulate fracture
(R = rock, ce = cement, ca = casing, F = fracture). propagation in the near-well. Written in MATLAB and
C/C++, this code was first developed by Alassi (Alassi,
Beside laboratory experiments, other methods have been 2008; Alassi and Holt, 2012) for failure in the context of
used to investigate cement sheath integrity, which include reservoir geomechanics.
theorical modelling (Bois et al., 2012; Thiercelin et al.,
1998), finite elements-based methods (De Andrade and
Sangesland, 2016; Orlic et al., 2018) as well as discrete
elements method. Thiercelin et al. (1998) used
thermoelastic modelling to examine the conditions under
which radial cracking or debonding may appear in the
cement sheath. Their results revealed that the downhole
conditions as well as the thermoelastic properties of
casing, cement and rock all play important roles in the
failure mechanisms. They also underlined that having a Figure 3. An element in MDEM. The red dots show the contacts
high compressive strength cement formulation is not and their allocated ID (Alassi, 2008; Gheibi et al., 2021). A
always a solution to prevent cracks. De Andrade and triangle element (continuum) is made of the centres of three
Sangesland (2016) investigated the effect of casing stand- particles that are connected by springs. A force is transmitted
off (not centralized in the hole) and thermal loading on when a relative displacement occurs between the particles.
cement sheath failure using a finite elements method.
Its use has been extended to reservoir pressurization, ur + iu = (u + iv)e − i (4)
hydraulic fracturing, fault reactivation (Rongved and
Cerasi, 2019) as well as casing pressurization (Gheibi et
 rr +   =  xx +  yy (5)
al., 2019). MDEM is a hybrid code, which can capture  −  rr + 2i r = ( yy −  xx + 2i xy )e2i (6)
both continuum (in elastic regime) and dis-continuum
(when an element fails) behavior of material. An element where ur and u are radial and tangential displacements,
in the code, as presented in the Figure 3, can only be a
triangle in 2D or tetrahedra in 3D, and the boundary  rr ,   and  r the radial, tangential and shear stress,
conditions are given in cartesian coordinates. The code is respectively, and ei = cos + i sin . The traction in the
coupled with the open-source MATLAB Reservoir material is defined by:

Downloaded from http://onepetro.org/ARMAUSRMS/proceedings-pdf/ARMA21/All-ARMA21/2479314/arma-2021-1233.pdf by Yangtze University, kai wang on 06 September 2023


Simulation Toolbox (MRST) code for fluid flow. This
coupling makes it possible to simulate the fracture 2( N − iT ) =  yy +  xx − ( yy −  xx + 2i xy )e2i (7)
propagation and the fluid flow into it.
where N and T are its normal and shear components,
3. ANALYTICAL DESCRIPTION respectively. In these polar coordinates, the complex
variable is given by z = rei , r and  being real numbers
Let us consider the well architecture of Figure 1. The in-
denoting the radius and the Lode angle, respectively. The
situ stresses are usually anisotropic.  1 and  2 are
strains are defined by:
perpendicular to the hole axis while  3 is along the axis. ur 1 u ur
 rr = ;   = +
The material is transversely isotropic around the hole-axis r r  r (8)
(no properties variation in the plane  1 –  2 ). The elastic 1  u 1 ur u 
 r =   + −
formulation is then reduced to 2D. The casing pressure is 2  r r  r 
denoted Pi , and is sometimes referred to as the inner
pressure, and the outer pressure (also referred to as where,  rr ,  and  r the radial, tangential and shear
confining pressure) is denoted P0 . In the configuration of strains, respectively. In the case of a single hollow
(multi-) hollow cylinder(s), the in-situ stresses  1 and  2 cylinder, according to Jaeger et al. (2007), the complex
are replaced by the confining pressure P0 . Jaeger et al. potentials can be given by:
d
(2007) showed that the stress and displacement  ( z ) = cz ;  ( z ) = (9)
distribution within a hollow cylinder can be derived from z
two complex potentials. For example, if  ( z ) and  ( z ) where c and d are real constants, which can be evaluated
are those complex potentials, the displacements ( u , v ) using the outer and inner boundary conditions. When
and stresses (  kl ) in cartesian coordinates are given by multi-concentric hollow cylinders are considered, as
the following expressions: presented in Figure 4, similar forms of complex potentials
can be used for each hollow cylinder as follows:
2G(u + iv) =  ( z ) − z ( z ) − ( z ) (1) dj
j ( z ) = c j z ;  j ( z ) = (10)
z
 xx +  yy = 2   ( z ) +  ( z )  (2)
  where c j and d j are real constants,  j ,  j the complex
 yy −  xx + 2i xy = 2  z  ( z ) + ( z )  (3) potentials in each material, and z the complex variable.
 

where G is the shear modulus of the material, u and v


the displacements in the x- and y-directions, respectively.
 = 1 − i  2 represents the conjugate of the complex
number  = 1 + i  2 ;   and   are the first and second
derivatives of  , z = x + iy is a complex variable, with (1)

i the complex number such as i = −1 .  is a function


2 (2)

of Poisson's ratio  . It is given by  = 3 − 4 for plane (3)

strain (  1  0 ,  2  0 ,  3 = 0 ) and  = (3 − ) / (1 + )
Figure 4. Three concentric hollow cylinders.
for plane stress (  1  0 ,  2  0 ,  3 = 0 ). The analytical
computations given in this section are for the plane strain For a well architecture, the hollow cylinders of Figure 4
problem. In polar coordinates, the displacements and represent the casing, cement sheath, and rock (see Figure
stresses are derived from equations (1)-(3) as follows: 1 on Formation 1's region). The casing has an inner radius
a and outer radius a1 . The inner radius of the rock is a2 3 
c1 = P0 − 2 Pi (23)
and its outer radius b . The cement sheath is then limited 2 1 2 1
by a1 and a2 . The three materials are characterized by where  j are dimensionless constants given by:
their shear moduli and Poisson's ratios (G j , j ) , with
a12 − a 2  2 1
j = 1, 2, 3, standing for casing, cement, and rock, 1 = +
respectively. The pressure Pi is applied at the inner
b2 1 1
surface of the casing, and the pressure Po is applied at the  2   a2
2 =  1 − 2  2 (24)
outer lateral surface of the rock. From equation (10), there  1 1  b
are six unknown constants, which need six independent
 2 

Downloaded from http://onepetro.org/ARMAUSRMS/proceedings-pdf/ARMA21/All-ARMA21/2479314/arma-2021-1233.pdf by Yangtze University, kai wang on 06 September 2023


relations. Continuity of displacements and tractions is 3 =  1 − 2  3
assumed at the two interfaces (casing/cement and  1 
cement/rock). Two equations are then obtained from
displacement continuity, two from traction continuity and  j and  j are also dimensionless constants given by:
b2 − a22 a22 ( 2 + 1 )
two from the boundary conditions at the inner wall of the
casing and the outer surface of the rock. From equations 1 =  2 + 2 (25)
(1), (4) and (10), the displacements in different materials b2 b2  3
are given by: a22 − a12 b2 − a22 a22 ( 2 + 1 )
2 = + (26)
1  d1  b2 b2 b2  3
ur + iu = ( 1 − 1)c1r − r  ; a  r  a1 (11)
2G1   b2 − a22 ( 1 − 2 )
3 = 1 + 2 (27)
ur + iu =
1  d 
( 2 − 1)c2r − 2  ; a1  r  a2 (12) b2 3

2G2  r 
( 1 + 1 ) a12 − 2 ( a12 − a2 )
1  d3  1 = (28)
ur + iu = ( 3 − 1)c3r −  ; a2  r  b (13) b2
2G3  r 
( 2 + 1 ) a12 + 2 ( a22 − a12 )
2 = (29)
From equations (1) - (7), the tractions in the materials are b2
also given by:
a2  a2 
d
N − iT = 2c1 + 21 ; a  r  a1 (14)
 3 = 2 ( 3 + 1 ) 22 − 2 (1 − 2 )  1 − 22  (30)
r b  b 
d
N − iT = 2c2 + 22 ; a1  r  a2 (15)  j is a function of the Poisson's ratio of material j. Using
r
(23) into equations (17) - (22), the other constants can be
d
N − iT = 2c3 + 23 ; a2  r  b (16) successively determined and are given as follow:
r
1 a2 
The continuity of displacements and tractions at the c2 = c1 − Pi + 3 P0 (31)
interfaces and the boundary conditions lead to the 1  1 1  1 b 2
1
following system of equations: a22 ( 2 + 1 ) ( 1 − 2 ) P
c3 = c2 − (32)
b 2
3 3 0

(1 − 1)a c − d1 = 1 (2 − 1)a c − 1 d2


2 2
(17)
d1 = a2 ( Pi − 2c1 )
1 1 1 2
(33)
(2 − 1)a c − d2 = 2 ( 3 − 1)a c − 2d3
2 2

( )
2 2 2 3 (18)
d2 = b P0 − 2a c − 2 b − a c3
2 2
2 2
2 2
2
(34)
2a12c1 + d1 = 2a12c2 + d2 (19)
d3 = b2 ( P0 − 2c3 ) (35)
2a c + d2 = 2a c + d3
2
2 2
2
2 3 (20)
a2 Pi = 2a2c1 + d1 (21) In the case of similar materials ( G1 = G2 = G3 = G and

b2 Po = 2b2c3 + d3 (22)
 1 =  2 =  3 =  ), it can be demonstrated that:
1 b2 P0 − a2 Pi
c1 = c2 = c3 = (36)
where 1 = G1 / G2 and 2 = G2 / G3 are dimensionless 2 b2 − a 2
parameters, G1 , G2 and G3 being the shear moduli of the a2 b2 ( Pi − P0 )
d1 = d2 = d3 = (37)
casing, cement and rock, respectively. The resolution of b2 − a 2
the system given in equations (17) - (22) leads to: and
ur =
(
( − 1) b P0 − a Pi
2 2

r
)

1 a b ( Pi − P0 )
2 2
(38)
(a,d). When increasing the casing pressure, the radial
stress and radial strain increase, while the tangential stress
4G b2 − a 2 2G b2 − a 2 r ( ) and strain decrease. Particularly, for the casing pressure

 rr =
(
( − 1) b P0 − a Pi
2 2

+
)1 a b ( Pi − P0 )
2 2
(39)
up to 7.5 MPa (d), the tangential stress is compressive
(positive) in cement and rock. With further increase of the
4G b2 − a 2 (
2G b2 − a 2 r 2 ) inner pressure, the tangential stress become progressively

  =
(
( − 1) b P0 − a Pi
2 2


)
1 a b ( Pi − P0 )
2 2
(40)
tensile (negative) in the cement sheath and the rock. At
casing pressure of 45 MPa, the tensile stress is as high as
4G b2 − a 2 2G b2 − a2 r 2 ( ) 370 MPa, 10 MPa and 12 MPa in the casing, cement, and
b2 P0 − a2 Pi a b ( Pi − P0 ) rock, respectively. This high tensile stress within the
2 2

Downloaded from http://onepetro.org/ARMAUSRMS/proceedings-pdf/ARMA21/All-ARMA21/2479314/arma-2021-1233.pdf by Yangtze University, kai wang on 06 September 2023


 rr = + (41)
( )
material is responsible for the radial cracks when its
b2 − a 2 b2 − a 2 r 2
absolute value exceeds the material's tensile strength. To
b2 P0 − a 2 Pi a b ( Pi − P0 )
2 2 analyse the effect of rock stiffness on the stress
  = − (42)
b2 − a 2 (
b2 − a 2 r 2 ) distribution, we consider the case in which the casing and
the outer pressures are equal (Pi = Po = 5 MPa). Three
configurations are analysed: (i) all the materials are
Equations (36) - (42) correspond to those usually found similar, (ii) the system is similar to that of Figure 4 with a
for one hollow cylinder ( u = 0 ,  r = 0 ,  r = 0 ). rock softer than cement, (iii) similar to (ii) but with a rock
Unlike the displacement and strains, notice the stiffer than cement. Their properties are summarized in
independence of stress with the material's elastic Table 2. The computation of radial and tangential stresses
properties – equations (41) and (42). Table 1 provides is presented in Figure 6. In the first case, it can be
some materials' properties which are used to compute the observed that there is no variation of stresses within the
analytical results of strains and stresses. They are given material when the inner and outer pressures are equal.
for different casing pressures, at a fixed outer pressure of However, in cases (ii) and (iii), due to different elastic
5 MPa, and as a function of radius from the wellbore properties of the materials, compressible tangential stress
(Figure 5). The tangential strain and radial stress are as high as 36 MPa appeared in the casing, which suddenly
continuous through the interfaces (Figure 5b,c), while the drops at its interface with the cement. One can also notice
radial strain and tangential stress are highly discontinuous the difference in stress distribution depending on the rock

Figure 5. Strains, and stresses as a function of casing pressure for constant pressure applied to the formation: P o = 5 MPa.
being softer or stiffer than the cement. When increasing 4. SIMULATION, RESULTS AND DISCUSSION
only the casing pressure from 5 to 25 MPa (and using
For numerical simulation, the system of Figure 4 has been
properties of Table 2), it can be observed that the tensile
meshed with different domains to be as close as possible
stress appears earlier in the stiffer rock (Figure 7).
to the experimental configuration. Like the analytical
Table 1. Materials properties. description, the simulation is limited here to the 2D case.
The meshed domains are presented in Figure 8 within a
Material Casing Cement Formation
square of 0.5×0.5 m2 in cartesian coordinates. The
a = 30.15 mm a1 = 32.15 mm a2 = 38 mm boundaries are constrained along the x- and y-axes at the
Size left and bottom sides, respectively, while boundaries at
a1 = 32.15 mm a2 = 38 mm b = 76 mm the right and top sides are free for movement. Only x-

Downloaded from http://onepetro.org/ARMAUSRMS/proceedings-pdf/ARMA21/All-ARMA21/2479314/arma-2021-1233.pdf by Yangtze University, kai wang on 06 September 2023


E (GPa) 200 8.0 15.0 displacement (respectively y-displacement) is permitted
on the bottom (respectively on the left) side. The meshed
ν [-] 0.3 0.15 0.28
domain is composed of a hole, a casing, an interface
casing/cement, a cement sheath, a rock, a seal sleeve, a
Table 2. Soft and stiff rock. confining region, and an outer region. The use of an
a = 30 mm Material Casing Cement Rock interface between the casing and the cement is because for
b = 76 mm practical cases (both in laboratory experiments and in the
Thickness:→ 2 mm 6 mm 38 mm
field), there is no perfect bond at this interface and some
E (GPa) 200 8.0 2.0
…softRock of its elements may move differently during a loading.
ν [-] 0.3 0.15 0.28
E (GPa) 200 8.0 30.0
This non-perfect bond can also explain the potential of
…stiffRock micro-annulus creation at casing/cement interface (Bois
ν [-] 0.3 0.15 0.28

Figure 6. Radial and tangential stresses for different material when Pi = Po = 5 MPa.

et al., 2011; Taghipour et al., 2021). It can be observed


that at the crossing of different materials, the element
sizes are affected, which could impact both the
convergence time and the crack patterns. A solution could
be to reduce and keep a constant element size in all
domains. It will improve the mesh, but significantly
increase the number of nodes and thus, the computing
power. A good compromise was found by adapting the
size of elements based on the particular material size, and
as a rule, to refine the mesh from the outer region down
to the casing. The sizes of different materials of Figure 8
and their poroelastic properties are summarized in Table
3. Notice different values of the rock's Young's modulus,
which effect will be analyzed. The casing/cement's
interface, the cement sheath and the rock can fail in tensile
mode during a variation of the casing pressure if the
Figure 7. Tangential stress for two rock's stiffnesses under an induced tensile stress exceeds the given tensile strength.
outer pressure of 5 MPa and a casing pressure of 25 MPa. A standard variation was also applied to the tensile
strength of these three materials and has shown to have (Agofack et al., 2018). A value of 7.5 GPa was assumed
rather limited effect on crack creation and propagation. for a confining pressure of 5-8 MPa. The parameters such
Other materials such as casing, sleeve, confining, and as tensile strength and loading rate were then calibrated.
outer regions were prevented to fail in tensile mode.
A simulation was run with the parameters of Table 3: the
Extremely high tensile strengths were then assigned to
tensile strength of interface casing/cement, cement sheath
them. The permeabilities of casing and sleeve were also
and rock were 0.10, 7 and 2 MPa, respectively. The initial
extremely low to prevent fluid communication from the
pore pressure in these materials was set to 5 MPa. The
hole or from the confining region.
confining pressure was kept at 7.5 MPa and the casing
pressure increased to 45 MPa. It was then decreased to
20 MPa. The result is presented in Figure 9 and is in good

Downloaded from http://onepetro.org/ARMAUSRMS/proceedings-pdf/ARMA21/All-ARMA21/2479314/arma-2021-1233.pdf by Yangtze University, kai wang on 06 September 2023


agreement with those of Figure 2, even though their
confining was 1 MPa higher. From Figure 9 it is also
observed that cracks started in the cement, precisely at its
interface with the casing, before propagating into the
rock. Most of the created cracks did not propagate through
the entire rock, probably due to material sizes in this
simulation, which were slightly higher than those used in
the experiment of Figure 2. Except for pore pressure, no
change was observed (e.g., crack closure) during the
reduction of the casing pressure from 45 MPa to 20 MPa.

Confining region

Rock formation
Cement sheath

Casing

Fracture/crack

Figure 9. Calibration's results at a casing pressure of 45 MPa


and under a confining pressure of 7.5 MPa.

4.2. Results and discussion


With parameters presented in Table 3, three different
simulations were run and only the Young's modulus of the
rock was made different in these simulations (7.5, 15 and
30 GPa). The confining pressure was kept constant at
5 MPa and the casing pressure increased up to 45 MPa.
The results are given in Figure 10 and Figure 11.
Figure 8. Mesh of different materials.
In Figure 10, the left graphs (a, d, g and j) are for a rock's
4.1. Parameter calibration Young's modulus of 7.5 GPa (less than that of the
The experimental results of Taghipour et al. (2021), cement). The middle and the right graphs are for a rock's
presented in Figure 2, were used for calibration. Since the Young's moduli of 15 and 30 GPa, respectively. The
poroelastic properties of the cured cement were not graphs in the first row (a, b and c) correspond to a casing
directly measured during the experiment, they were pressure of 10 MPa, and those at the bottom row
estimated by means of micromechanical modeling as correspond to a casing pressure of 35 MPa. At a given
presented in (Agofack, 2019; Agofack et al., 2020). rock's Young's modulus, the simulation results show that
Similarly, the properties of the Castlegate sandstone (the fractured elements and fracture propagation increase with
rock used in their experiments), were not measured. increase of the casing pressure. For example, for the
Previous investigations showed that its Young's modulus rock's Young's modulus of 7.5 GPa, there is no fractured
is around 2 GPa for unconfined loading tests but can be element at the casing pressure of 10 MPa (Figure 10a),
as high as 13 GPa under high confining pressure while four major cracks are present in the cement sheath
Table 3. Parameters of the different materials used in this simulation.
Interface Confining Outer
Material Hole Casing Casing/cement Cement Rock Sleeve
Region Region

Size: OD (mm) 60.3 64.3 65.0 76.0 152.0 155.0 200.0 500

Young modulus [GPa] 0.28 200 2 8.0 7.5, 15, 30 0.028 2.8 7

Poisson ratio [-] 0.48 0.27 0.15 0.15 0.28 0.48 0.45 0.25

Porosity [-] 0.95 0.02 0.40 0.40 0.29 0.02 0.95 0.3

Downloaded from http://onepetro.org/ARMAUSRMS/proceedings-pdf/ARMA21/All-ARMA21/2479314/arma-2021-1233.pdf by Yangtze University, kai wang on 06 September 2023


Permability [m2] 1×10-5 1×10-21 1×10-16 1×10-16 2×10-15 1×10-23 1×10-5 1×10-13

Tensile strength σt [MPa] 1020 1020 0.10 7 2 1020 1020 1020

and rock at casing pressure of 15 MPa (d), and many more correctly calibrate the model to recreate the observed
cracks at 35 MPa (j). The rock's Young's modulus also fracturing from the experiments. Once this is achieved, it
strongly affects the crack initiation and propagation. For is a straightforward matter to upscale dimensions to field
example, at a given casing pressure of 10 MPa, there are well sizes. In addition, the MDEM code was designed for
no cracks when the rock's Young modulus is 7.5 GPa, anisotropic in-situ stresses in cartesian coordinates, where
while many cracks have appeared in the cases of higher the main fractures propagate along the direction of the
rock's Young's moduli (b, c). It is therefore observed that maximum horizontal stress. For isotropic boundary
the higher the rock's Young's modulus the earlier cracks stresses, in theory, the cracks can start at any point around
appear in the rock and/or cement sheath. This observation the wellbore. However, even with the use of a confining
agrees with the analytical solution presented in Figure 7, domain, it was found that the x- and y-axis were
which showed that the tensile stress appears earlier when systematically the main directions of first cracks initiation
the rock is stiffer. A softer rock indeed is more deformable and propagation (e.g., see Figure 10b, d).
and thus permits less buildup of the tensile stress, as
observed in the stiff rock. Also note that for the rock's 5. CONCLUSIONS
Young's modulus of 30 GPa, there is almost no crack in
the cement sheath for the casing pressure up to 15 MPa, A combined analytical and numerical approach was used
while many cracks have appeared in the rock (c, f). A in this paper to investigate the effect of rock stiffness on
damaged region can also be observed at the interface crack initiation and propagation around the borehole.
cement/rock (for stiffer rock) and increases with the Equations of multi-hollow cylinders were derived, and the
casing pressure (c, f, i and l). stress and strain distributions were computed. The results
showed that under certain casing pressure, the tangential
At the casing pressure of 45 MPa (Figure 11), it is stress becomes negative (tensile) and is responsible of
observed that some boundary effects appeared in the rock radial cracking. In addition, it was observed that the stiffer
with higher Young's modulus. These boundary effects the rock the earlier the tensile stress appeared with
manifest themselves by formation of a damaged region increase of the casing pressure. For simulations, the
within the rock, which could have probably increased MDEM code was used. The parameters were calibrated
with higher casing pressures (Figure 11n). Overall, the on the experiments of Taghipour et al. (2021). The results,
simulations agree very well with the experimental results which agreed very well with the experimental
(Figure 2). Some compromises were however needed to observations showed that under constant confining
obtain satisfactory and semi-quantitative match between pressure, radial cracks increased with the applied casing
the simulation results and the laboratory experiments. pressure. At fixed casing and confining pressures, radial
These can be summarized by stating that the MDEM cracks increased with the rock's Young's modulus. When
software was conceived for porous rocks and therefore it the rock's stiffness is higher than that of the cement and
was not straightforward to include steel casing and rubber its tensile strength lower, the cracks started in the rock
sleeve, without having rigorous interfaces between them
before propagating into the cement sheath.
and the geomaterials. Interface transitions are best
captured with a detailed mesh, however computing time
becomes excessive if the whole domain is discretized in 6. AKNOWLEDGEMENTS
detail, especially as fracturing starts to develop. It is thus This publication has been produced with support from the
very tempting to reduce the modelling domain extent, European Union’s Horizon 2020 research and innovation
although this is not easily done without compromising program, grant no. ENER/H2020/764531/SECURe. The
accurate representation of applied stresses, especially in authors would also like to acknowledge Idar Larsen for
the isotropic stress case. Eventually, it was possible to guidance and advices during the numerical simulations.
Downloaded from http://onepetro.org/ARMAUSRMS/proceedings-pdf/ARMA21/All-ARMA21/2479314/arma-2021-1233.pdf by Yangtze University, kai wang on 06 September 2023

Figure 10. Effect of rock's stiffness on fracture around the well under a confining pressure of 5 MPa and for casing pressure up to
35 MPa.
In Proceedings of the SPE Annual Technical
Conference and Exhibition, New Orleans, 4–
7 October 2009.
9. De Andrade, J. and S. Sangesland.
2016. Cement Sheath Failure Mechanisms:
Numerical Estimates to Design for Long-
Term Well Integrity. Journal of Petroleum
Science and Engineering. 147: 682-698.
10. Gheibi, S., N. Agofack, and S.
Sangesland. 2021. Modified discrete element

Downloaded from http://onepetro.org/ARMAUSRMS/proceedings-pdf/ARMA21/All-ARMA21/2479314/arma-2021-1233.pdf by Yangtze University, kai wang on 06 September 2023


method (MDEM) as a numerical tool for
cement sheath integrity in wells. Journal of
Petroleum Science and Engineering. 196.

Figure 11. Effect of rock's stiffness on fractures around the well 11. Gheibi, S., Sangesland S., Vrålstad T. 2019. Numerical
under the confining and casing pressures of 5 MPa and 45 MPa, modeling of radial fracturing of cement sheath caused
respectively. by pressure tests. Proceedings of the ASME 2019 38th
International Conference on Ocean, Offshore and Arctic
Engineering OMAE2019, Glasgow, June 9-14, 2019.
REFERENCES
12. Goodwin, K.J. and R.J. Crook. 1992. Cement Sheath
1. Agofack, N. 2019. Micromechanical and Stress Failure. SPE Drilling Engineering. 6: 291-296.
homogenization modelling for a hydrating oilwell
cement paste. In Proceedings of the 53rd US Rock 13. Jaeger, J.C., N.G.W. Cook, and R.W. Zimmerman.
Mechanics /Geomechanics Symposium, New York, 23– 2007. Fundementals of rocks mechanics. Fouth Edition,
26 June 2019. Blackwell Publishing.

2. Agofack, N., S. Ghabezloo, and J. Sulem. 2020. Chemo- 14. Orlic, B., A. Chitu, L. Brunner, M. Koenen, J.
poro-elastoplastic modelling of an oilwell cement paste: Wollenweber, and G.-J. Schreppers. 2018. Numerical
Macroscopic shrinkage and stress-strain behaviour. investigations of cement interface debonding for
Cement and Concrete Research. 132: 106046. assessing well integrity risks. 52nd US Rock Mechanics
/Geomechanics Symposium, Seattle, Washington, USA,
3. Agofack, N., S. Lozovyi, and A. Bauer. 2018. Effect of 17–20 June 2018.
CO2 on P- and S-wave velocities at seismic and
ultrasonic frequencies. International Journal of 15. Rongved, M. and P. Cerasi. 2019. Simulation of Stress
Greenhouse Gas Control. 78: 388-399. Hysteresis Effect on Permeability Increase Risk Along
A Fault. Energies. 12: 3458.
4. Alassi, H.T. 2008. Modeling reservoir geomechanics
using discrete element method: Application to reservoir 16. Skorpa, R., B. Werner, and T. Vrålstad. 2019. Effect of
monitoring. Norwegian University of Science and Mud on Cement Sheath Integrity. In Proceedings of the
Technology. PhD Thesis. 144p. SPE Norway One Day Seminar, Bergen, 14 May 2019.

5. Alassi, H.T. and R. Holt. 2012. Relating discrete 17. Skorpa, R., T. Øia, A. Taghipour, and T. Vrålstad. 2018.
element method parameters to rock properties using Laboratory set-up for determination of cement sheath
classical and micropolar elasticity theories. integrity during pressure cycling. In Proceeding of the
International Journal for Numerical and Analytical 37th International Conference on Ocean, Offshore and
Methods in Geomechanics. 36: 1350-1367. Arctic Engineering, ASME, Madrid, June 17-22, 2018.

6. Anya, A., H. Emadi, and M. Watson. 2020. Computed 18. Taghipour, A., A. Ghaderi, P. Cerasi, and S. Gheibi.
Tomography Study of Annular Cement Mechanical 2021. Novel experimental setup for study wellbore
Response Under Cyclic Hydraulic Stress. Proceedings cement and formation integrity. (Under review at)
of the 8th Unconventional Resources Technology Journal of Petroleum Science and Engineering.
Conference, Austin, 20-22 July 2020. 19. Thiercelin, M.J., B. Dargaud, J.F. Baret, and W.J.
7. Bois, A.-P., A. Garnier, G. Galdiolo, and J.-B. Laudet. Rodriquez. 1998. Cement Design Based on Cement
2012. Use of a Mechanistic Model To Forecast Cement- Mechanical Response. SPE Drilling & Completion. 266-
Sheath Integrity. SPE International Conference on CO2 273.
Capture, Storage, and Utilization, New Orleans, 10–12 20. Vrålstad, T., R. Skorpa, and B. Werner. 2019.
November 2010. Experimental Studies on Cement Sheath Integrity
8. Bois, A.-P., A. Garnier, J. Saint-Marc, and N. Aimard. During Pressure Cycling. In Proceeding of the
2011. How To Prevent Loss of Zonal Isolation Through SPE/IADC International Drilling Conference and
a Comprehensive Analysis of Microannulus Formation. Exhibition, The Hague, 5-7 March 2019.

You might also like