You are on page 1of 13

1.

INTRODUCTION
The petroleum industry drills inclined and/or horizontal
wellbores for many purposes. The wellbores are
completed using casing and cement to maintain wellbore
integrity. Moreover, these completed wellbores often
experience severe conditions that threaten the integrity
of the wellbore: eccentric casing; large changes in
wellbore pressure; large changes in wellbore temperature
and so on. These conditions often intensify near
wellbore stress states, which, in turn, influence wellbore
stability, sand production, zonal isolation, and hydraulic
fracturing. Most oil companies spend significant amount
of money to prevent the loss of wellbore integrity.
Therefore, proper modeling of the stress states around
wellbores is an important step in evaluating the
effectiveness of a drilling/completion strategy.
For modeling stress states around wellbores, this work
mainly focuses on two objectives: (1) develop
comprehensive analytical and numerical models for an
inclined cased wellbore; (2) apply the developed models

1
A commercial finite element analysis software program
made by Dassault Systmes S.A.
to study causes of the cement sheath failure. The
following section introduces some background
information relevant to the objectives.
1.1. Developing Models for an Inclined Cased
Wellbore
Generally, several processes are required to model the
state of stress of an inclined cased wellbore. Three main
processes are considered as shown in Fig. 1: Phase 2, 3
and 4.
In this figure, Phase 1 shows the natural state of rock
formation under in situ stresses before drilling.
Phase 2 illustrates the process of drilling an inclined
wellbore. Drilling itself causes several complicated
processes. However, these can be simplified into two
basic processes: plain excavation and wellbore pressure
of the drilling mud. Bradley calculated the stress
distribution around an inclined open wellbore through
this simplification along with the assumption of linear
elasticity of the formation, and applied them to wellbore
failure [1]. However, this elastic model had crucial
limitations. It ignored the coupling between the rock
matrix and the fluid contained inside. Moreover, the

ARMA 10-142

Mechanical Behavior of Concentric Casing, Cement, and Formation
Using Analytical and Numerical Methods

Jo, H.
BJ Services Company, Houston, Texas, USA
Gray, K.E.
The University of Texas at Austin, Austin, Texas, USA

Copyright 2010 ARMA, American Rock Mechanics Association
This paper was prepared for presentation at the 44
th
US Rock Mechanics Symposium and 5
th
U.S.-Canada Rock Mechanics Symposium, held in
Salt Lake City, UT June 2730, 2010.
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 300 words; illustrations may not be copied. The
abstract must contain conspicuous acknowledgement of where and by whom the paper was presented.
ABSTRACT: The primary goal of this research is to develop comprehensive analytical and numerical models for stress
distributions around an inclined, cased wellbore by considering all wellbore processes, including amendments to models of other
works. Most previous research has focused on individual wellbore processes rather than a comprehensive treatment which
considers all wellbore processes simultaneously: a more realistic approach developed in this paper. To achieve this goal, this work
utilizes an elastic approach by coupling a poroelastic, undrained condition and a steady state condition for stresses induced by
wellbore temperature variations. The superposition principle is used to develop a comprehensive model, which is then applied to
cement sheath failure. ABAQUS
1
is utilized for numerical solutions to verify the comprehensive analytical model. These
comprehensive models show analogous stress distribution results to those of previous models at each individual wellbore process
when using the plane strain condition. However, ABAQUS model results show stress differences because the general plane strain
model for the analytical solution and full 3D model for numerical solution are used. While there are differences between analytical
and numerical solutions as noted, the comprehensive analytical model is a good alternate to costly numerical software programs,
and it provides an improvement to plane strain models.

elastic model was not a 3D stress model because it used
a plane strain condition. To consider the coupling effect
of an inclined open wellbore, Cui et al. used Biot's fully
coupled poroelastic theory and Detournay et al.'s
poroelastic plane strain model. They developed a
poroelastic stress distribution for an inclined open
wellbore and applied it for wellbore failure [2, 3, 4, 5, 6,
7]. Nevertheless, this poroelastic model utilized the
plane strain condition, and its solution was expressed in
the Laplace-transform domain. To apply an inverse
Laplace transform for the solution, the poroelastic model
used a numerical inverse Laplace transform algorithm
[8], which requires long convergent times. Interesting
results of the poroelastic model can be summarized as
follows: the elastic model is more conservative than the
poroelastic model, and the results of the poroelastic
model approach the elastic model as time approaches
infinity [9].
Phase 3 depicts the well completion process involving
the placement of casing and cement. The mechanism of
how casing and/or cementing affects the stress
distribution of an inclined cased wellbore is complicated
and mostly unknown. The results of a few previous
studies are applicable to general cases [10]. Chenevert et
al. [11] made a semi-analytical model for the stress
distribution induced by cement shrinkage and compared
its results with experimental data by Cooke et al. [12].
Prohaska et al. developed a more intensive and reliable
model than the previous work [13]. Recently, Zhou et al.
collected the previous data and made a comprehensive
model [14]. However, these analytical models are not
widely applicable since they are based on limited
experimental results. Thus, no generally applicable
analytical models for stress distribution induced by
casing and/or cementing are available. To compensate
for limitations of these analytical models, Gray et al.
developed a reliable numerical model by using
ABAQUS [10]. However, this present work ignores the
effect of the cement shrinkage and assumes that the
cement is a hydraulic column before it is set.
Phase 4 represents tectonic stress variation and/or
wellbore temperature variation after a well is completed.
There exist few analytical and/or numerical models that
address those factors listed in Phase 4. Li developed a
simple analytical model, which considered a well with
only casing [15], for the stress distribution induced by
tectonic stress variation after well completion. Atkinson
et al. developed a more extended analytical model which
considered a well with both casing and cement. However,
the researchers mistook tectonic stress variations as the
in situ stresses [16]. In addition, Carter et al. developed a
numerical model [17] and recently, Morita et al.
performed experiments to develop a more realistic
analytical model [18]. For the stress distribution induced
by temperature variation, Timoshenko et al. introduced a
general solution [19]. Moreover, Ghosn derived the
solution for the coupling between rock matrix and
temperature [20] and Shahri tried to solve the stress
distribution around an inclined cased wellbore under
high temperature and high pressure using ANSYS
2
[21].
Although separate models treating each phase in Fig. 1
have been developed, the comprehensive and combined
effects of all the phases on the stress distribution around
an inclined cased wellbore have not been intensively
considered. Moreover, some previous work applied the

2
A commercial finite element analysis software program
made by ANSYS, Inc.
Fig. 1. Processes of drilling and well completion, and related previous studies
incorrect boundary conditions, as mentioned above, for
an inclined cased wellbore and, consequently, developed
erroneous models. In order to address these errors, the
present work develops an analytical and numerical
model that considers the comprehensive and combined
effects of all three phases. In addition, most previous
analytical models used the plane strain condition to
transform a 3D problem, which is non-solvable
analytically, into a 2D problem, which is solvable
analytically. Even though the plane strain condition (i.e.
axial strains along a wellbore are assumed to be zero)
can be applicable to most wellbores, non-zero axial
strains occur along real wellbores. To compensate for
this limitation, this research uses the generalized plane
strain condition developed by Duncan et al. [22] and
Zhenye et al. [23].
While only numerical results were available previously,
this research develops an analytical model for the stress
distribution induced by wellbore temperature variation.
Although the analytical model assumes a steady state
condition, it is economically beneficial because it does
not require costly numerical programs.
1.2. Cement Failure
Failure theories are constantly being developed. They
can be categorized into two main areas: tensile failure
and shear failure. Specifically, several criteria for shear
failure have been proposed, while tensile failure has a
simpler mechanism. Among them, Mohr-Coulomb
criterion and Drucker-Prager criterion [24] are most
commonly used. However, the former underestimates
material strength by ignoring the effect of the
intermediate stress, whereas the latter overestimates
material strength [25]. To resolve this problem, this
research uses Mogi-Coulomb criterion [25].
Handin [26] and Goodwin et al. [27] performed
experiments to investigate cement failure under high
pressure and/or high temperature environments.
Moreover, Thiercelin et al. [28] and Bosma et al. [29]
utilized numerical models to investigate cement failure
and to design a cement sheath under high pressure
and/or high temperature environments. These results
were based on experiments or numerical programs
which require significant time and expense; alternative
models are needed to save time and expense. This
research applies the developed analytical model for the
cement failure under high pressure and/or high
temperature.

2. DEVELOPING MODELS
2.1. Fully Coupled Poroelasticity
This research distinguishes between partially coupled
and fully coupled poroelasticity. Partially coupled
poroelasticity ignores the interaction between solid
(solid stress) and fluid (pore pressure). However, fully
coupled poroelasticity considers that interaction. This
concept was first formulated by Biot [30] and developed
by several others. Since rock is a porous medium, a fully
coupled poroelastic approach is required. Nevertheless,
the complexity of poroelastic solutions caused by the
coupling between solid and fluid makes elastic
approaches preferable.
This research applied the theory of fully coupled
poroelasticity to an inclined cased wellbore which is a
combined system of elastic and porous materials. In
addition, it showed that a poroelastic, inclined cased
wellbore system under undrained conditions became an
elastic one with undrained properties instead of normal
elastic properties, if it is assumed that the wellbore is
perfectly cased and cemented and does not allow any
leaking fluid from the rock formation. The complete
derivation is shown in the dissertation by Jo [31].
2.2. Governing Equations
Several governing equations were required to analyze an
inclined cased wellbore system since the system had
solid-fluid coupling and solid-temperature coupling.
These governing equations and couplings are
schematically represented in Fig. 2. Horizontal arrows
represent coupling directions. Upper boxes describe
governing equations for each coupling. Lower boxes
show the governing equations for each area which have
no direct relation with the couplings. If these equations
are solved simultaneously under certain boundary
conditions for an inclined cased wellbore system, the
stress distribution of the system can be obtained and the
mechanical behavior of casing, cement and rock can be
analyzed and predicted. These analyses and predictions
are the desired goal of this work.
However, there were technical problems in analytically
solving these equations under general conditions.
Numerical techniques or programs such as ABAQUS,
DIANA
3
, ANSYS, I-DEAS
4
, etc. may be used to solve
them. Thus, some special but reasonable boundary
conditions were used in order to solve the governing
equations simultaneously and analytically. These
boundary conditions decouple the set of equations and
make them analytically solvable.
To attain the goal, the undrained poroelastic condition
and steady state wellbore temperature variation were
introduced. The undrained poroelastic condition
decouples solid-fluid coupling as stated above. Also, the
steady state wellbore temperature variation decouples
solid-temperature coupling. For this reason, if a steady

3
A commercial finite element analysis software program
made by TNO DIANA
4
A commercial integrated design and engineering analysis
software made by Siemens PLM Software
state is assumed, all time variation dependent terms
disappear. Thus, only one term remains in the
generalized Fourier's law [20]. This equation can be
generally solved in terms of only spatial variables; for
example r, and z. The solution can be substituted into
the Duhamel-Neuman relation, which represents the
solid-temperature coupling. The relation is reduced to an
elastic equation which is solvable with ease. Therefore,
the system becomes decoupled and analytically solvable.

3. STRESS DISTRIBUTION AROUND AN
INCLINED CASED WELLBORE
Based on the results of the previous section, this research
used elastic approaches to analyze an inclined cased
wellbore. However, this elastic model is distinct from
previous models [16, 22] because this research
developed an analytical, elastic model including
generalized plane strain, steady state temperature effect
on the stress distribution, and correction for the
erroneous modeling procedure of previous research [16].
To calculate the stress distribution induced after drilling,
casing and cementing, some assumptions are required as
follows.
3.1. Assumptions
The following assumptions are required for developing a
comprehensive model:
(i) The coordinate system used in this research is a
local borehole coordinate system. Fig. 3 and 4
show the details.
(ii) For simplicity, the directions of in situ principal
stresses coincide with the global coordinates.
(iii) The casing, cement and borehole are concentric
circles.
(iv) The casing, cement and borehole are assumed to
be perfectly bonded to each other at each interface.
The perfectly bonding mathematically means that
the continuity of stresses and displacements is
satisfied at each interface.
(v) Casing and cement were homogeneous, isotropic,
linearly elastic materials and rock was a
homogeneous, isotropic, linearly poroelastic
material.
(vi) The rock formation had an infinite boundary and it
was a porous medium which was 100% saturated
with one phase pseudo fluid.
(vii) Casing and cement were assumed to be nonporous.
(viii) Drilling was assumed to happen instantaneously.
Since enough time (3 ~ 5 hours) had passed
between drilling and well completion, poroelastic
effects of pore pressure induced by drilling on the
stress distribution could be ignored [3, 9].
(ix) The effects of invasion and cement shrinkage were
negligible. Since reasonable analytical models for
the cement shrinkage were not available, this
research assumed that cement slurry was a
hydraulic column before it was set.
(x) The poroelastic undrained condition was applied
after well completion.
(xi) Generalized plane strain condition was used,
which has many versions [23]. However, this
research used the constant axial strain-generalized
plane strain condition [23].
Fig. 2. Governing equations for an inclined cased wellbore system
Fig. 3. Global coordinate system and in situ principal stresses

(xii) The stresses induced by wellbore temperature
variation were assumed to be steady state, i.e. not
a function of time.
(xiii) The coupling between temperature and fluid
contained inside of rock was ignored because
perfectly bonded casing and cement at each
interface sealed the wellbore, in other words, there
was no fluid behavior.
(xiv) No history of previous phases affected any phases.
This assumption becomes reasonable when a
system has small displacements (or deformations)
induced by each phase. Since the system of an
inclined cased wellbore can have small
deformations by each wellbore process, this
assumption can be applicable to this system. This
assumption is required because the superposition
principle is valid under it.
In this paper, the sign convention for stress is positive in
compression following the convention used in rock
mechanics.
3.2. Methodology
This research investigated the process of drilling and
well completion and found that most previous models
improperly analyzed the stress distribution for an
inclined wellbore. For example, while previous work
imposed in situ stresses after well completion, this
research imposed in situ stresses before drilling. Since
real rock formations are under in situ stresses before
drilling, this model simulates the system more
realistically by considering the effects of drilling in the
presence of in situ stresses. Fig. 1 shows the process of
drilling and well completion.

Fig. 4. In situ principal stresses and wellbore coordinate
system [(x, y, z) = (x
w
, y
w
, z
w
)]

At Phase 1, the rock formation is under in situ principal
stresses (or in situ stresses). It was assumed that the
magnitude and direction of in situ stresses were known.
Although methods of measuring in situ stresses have
been reported, accurate measurement of the in situ
stresses remains a challenging area in geoscience.
At Phase 2, a generalized plane strain condition is used.
All detailed procedures may be found in the dissertation
of Jo [31].
At Phase 3, the influence of casing and cementing, such
as cement shrinkage on the stress distribution, should
have been considered. However, this research ignored
these contributions because they were considered second
order effects compared to primary factors, such as
changes in wellbore temperature and pressure; most
results are applicable only under special conditions; and
reliable models were not available [14]. The study on the
influence of cement shrinkage is a challenging
geomechanical research area [10].
At Phase 4, the stress distribution caused by tectonic
stress variation was derived by modifying Li's model
[15]. Details of the model are explained in Jos
dissertation as well as the stress distribution induced by
steady state temperature variation [31].
Through phases 1 to 4, the stress distribution induced by
each factor was obtained. To combine each stress
distribution caused by each factor, it was assumed that
the system was linear elastic and the history of all phases
was independent of each other. That is, the principle of
linear superposition for the stress was applied.
Considering that most geological materials exhibit non-
linear behavior, this superposition principle may not
always be appropriate. However, the superposition is
useful and reasonable in a practical respect. The
superposition can be simply mathematically represented
as follows.

disturbed initial induced
o o o A + = (1)
Therefore,
initial
for phase 2 is in situ stress and
disturbed

for phase 2 is stress disturbed by drilling.
induced
for
phase 2 is second order tensor summation according to
Eq. (1). For phase 3,
initial
becomes
induced
for phase 2
and
disturbed
is stress disturbed by casing and cementing.
By applying this sequential process, the stress
distribution of an inclined cased wellbore at subsequent
phases can be obtained.

4. APPLICATIONS (CEMENT FAILURE)
The comprehensive analytical model developed through
the above processes is applicable to many wellbore
phenomena. For illustration purposes, cement failure is
considered here. While many factors are involved, this
example illustrates the models utility.
Generally speaking, there are two kinds of failures:
tensile failure and shear failure. For cement, tensile
failure occurs when a stress on the cement exceeds the
tensile strength of the cement. Similarly, shear failure
happens when a stress surpasses the shear strength of the
cement. However, since cement behaves differently
under tension and compression, separate failure criteria
are needed for each type of failure.
4.1. Tensile Failure
The tensile failure is referred to as Mode I fracture
(opening fracture). In addition, the tensile strength of
cement is about ten times less than the compressive
strength [32].
A tensile failure criterion can be described as follows:

f o p
T P s o
(2)
where
p
is the least principal stress, P
o
is the pore
pressure in the cement and T
f
is the tensile strength of
cement [1, 33].
It is assumed that that P
o
=0 because the cement is
assumed to be non-porous. Thus,

0 s +
f p
T o
(3)
If a location in the cement area satisfies Eq. (3), tensile
failure occurs at that location.
Using this tensile failure criterion, this research
investigates the influence of Young's modulus of the
cement, the temperature of the wellbore and wellbore
pressure. Both analytical and numerical models
(ABAQUS models) are used to obtain the results.
4.2. Shear Failure
The shear failure has been studied by many researchers
and has many criteria. Among them, this work uses the
Mogi-Coulomb criterion, which was experimentally
verified for describing rock failure [25].
While the Mohr-Coulomb criterion consists of the shear
stress () and the normal stress (
n
), and Drucker-Prager
criterion uses the octahedral shear stress (
oct
) and the
octahedral normal stress (
oct
), the Mogi-Coulomb
criterion was developed in terms of the octahedral shear
stress (
oct
) and mean stress (
m,2
) calculated as follows.
( ) ( ) ( )
2
1 3
2
3 2
2
2 1
3
1
o o o o o o t + + =
oct
(4)

2
3 1
2 ,
o o
o
+
=
m
(5)
In addition, Fig. 5 depicts the Mogi-Coulomb failure
envelope. The envelope can be expressed
mathematically as follows.

2 , m oct
b a o t + =
(6)
where a and b are Mogi strength parameters. Generally,
they are obtained experimentally and given for failure
analysis.
This research investigates the influence of the same
factors as those of the tensile failure case on the cement
shear failure.
4.3. Assumption
The following assumptions are required for the failure
criteria to be applied to the system properly:
(i) Cement is assumed to be homogeneous, linear,
isotropic, and elastic. Although this assumption is
simple and has some limitations when applied to
real situations, it is practical and appropriate for
many cases.
(ii) The failure of cement experiences no plastic
behavior.
Fig. 5. Mogi-Coulomb failure envelope [25]

(iii) A linear failure envelope is used for Mogi-
Coulomb failure. Higher order envelopes are not
used because of their complexity.
(iv) Only failure in the cement is assumed to occur in
order to ignore the coupling among failures of
casing, cement and rock formation.
(v) Each factor affecting cement failure acts
independently of all of the others.

4.4. Methodology
Based on these assumptions, the following steps are
applied to investigate the tensile and shear failure of
cement:
(i) Calculate stresses in the cement in the area of
interest. The stresses were calculated by using the
developed analytical and ABAQUS models.
(ii) Calculate the least principal stress at every
location in the cement area.
(iii) Investigate whether tensile failure in the cement
has occurred by checking whether or not Eq. (3) is
satisfied.
(iv) Calculate
oct
and
m,2
in the cement area of interest.
(v) Plot
oct
versus
m,2
.
(vi) Draw a linear failure envelope for shear failure.
(vii) Investigate whether shear failure of the cement has
occurred.
(viii) Perform steps (i) through step (vii) to study the
influence of each of the following parameters:
(a) Young's modulus of the cement
(b) Wellbore temperature variation
(c) Wellbore pressure
The tensile failure was investigated using steps (ii) to
step (iii). The shear failure was studied using steps (iv)
to step (vii).

5. RESULTS
Table 1 summarizes the input data. For the analytical
analyses, MATLAB
5
is used for the analytical model
whereas ABAQUS is utilized for the numerical analyses.
5.1. The Effect of Youngs modulus of the Cement
on Cement Failure
For Young's modulus of the cement, this research
considers the coupling between the tensile and
compressive strength of the cement and Young's
modulus of the cement. Thus, this study uses Lacy's
empirical correlations between the unconfined
compressive strength (UCS) and static Young's modulus
as follows [34]:

5
A commercial software for technical computing developed
by The MathWorks
TM

) ( 485 . 2 2787 . 0 ) (
2
ksi E E C UCS
s s o
+ = (7)
where E
s
is static Young's modulus (10
6
psi).
Even if Lacy's correlation was largely based on weaker
sands, it is useful to this study. The tensile strength of
cement (T
f
) is assumed to one tenth of UCS [32]. In
addition, Mogi strength parameters can be expressed as
follows [25]:

1 3
2 2
+
=
q
C
a
o
(8)

1
1
3
2 2
+

=
q
q
b (9)
where
( )
( ) |
|
sin 1
sin 1

+
= q
and | is the angle of internal
friction of cement, which is normally assumed to be 30
o
.
For tensile failure, Fig. 6 describes the minimum value
of
p
+T
f
in the cement for a given Young's modulus.
Hereafter, large solid dots represent analytical results
and small hollow dots represent numerical results. The
location where tensile failure initially occurs is not
considered. Only the value related to the initial tensile
failure is considered. This diagram shows that the
minimum value of
p
+T
f
declines and changes from a
positive to a negative value as Young's modulus
increases. This change in sign indicates that an initial
tensile failure occurs in the cement. In other words, both
the analytical and ABAQUS calculated results show that
the chance of the cement tensile failure increases as the
Young's modulus increases. In addition,
p
+T
f
values for
the analytical solution showed a linear relationship with
Young's modulus of the cement, while
p
+T
f
for the
ABAQUS solution showed a quadratic relationship. The
specific reason for this result is not understood. However,
the difference between the analytical and ABAQUS
models may cause this phenomenon: generalized plane
strain model versus full 3D model.
Fig. 6. The effect of Young's modulus of cement on tensile
cement failure
For shear failure, Mogi-Coulomb shear failure criterion
is utilized [25]. In other words, if both analytical and
numerical results are plotted in (
m,2
,
oct
) space and the
plotted points lie on or over the given failure envelope,
then this indicates that shear cement failure occurs at the
interface between casing and cement. Fig. 7 shows the
influence of Young's modulus on shear failure at the
interface between casing and cement. Since the tensile
and compressive strength of the cement and Young's
modulus of the cement are coupled, the failure envelope
changes whenever Young's modulus of the cement
changes. Both analytical and numerical results show that
as Young's modulus increases, the results move away
from the corresponding failure envelopes. Therefore, as
Young's modulus increases, the chance of shear failure
decreases, while the chance of the tensile failure
increases. These interesting phenomena may not be
general because the calculations are specific to Lacy's
empirical correlations. This observation can be
explained through following equations: Eq. (10) and (11).
The slope of AT
f
/AE
2
is smaller than the slope of a/AE
2
.

2
0657 . 0
1 3
2 2
s
o
E
q
C
a
+
= (10)

2
02787 . 0
10
s
o
f
E
C
T =

(11)
However, even though the relationship between tensile
and shear failure and Youngs modulus of cement
depends on the correlations between the modulus and the
strength of the cement, the results of this work can
provide some good guidelines for the selection of
cement to avoid the cement failure after the cement is set.
In addition, the analytical results corresponding to each
Young's modulus of cement are closer to the failure
envelope than the numerical results. This means that the
analytical results reach the shear failure faster than the
numerical results.
Table 1. Input data for cement failure


Casing inner radius (R
1
) 0.1778 m (7 in)
Casing outer radius (R
2
) 0.1905 m (7.5 in)
Wellbore radius (R
3
) 0.2159 m (8.5 in)
Young's modulus of casing (E
1
) 206.8 GPa (30 MMpsi)
Young's modulus of cement (E
2
) 13.8 GPa (2 MMpsi)
Young's modulus of rock (E
3
) 10.3 GPa (1.5 MMpsi)
Poisson's modulus of casing (
1
) 0.3
Poisson's modulus of cement (
2
) 0.2
Poisson's modulus of rock (
3
) 0.2
Depth 1828.8 m (6000 ft)
Wellbore pressure (P
w
) 34.5 MPa (5000 psi)
Azimuth 0
o

Inclination 0
o

Reservoir pressure 19 MPa (2760 psi, or 0.46 psi/ft)
Max. horizontal in situ stress (
H
) 31 MPa (4500 psi , or 0.75 psi/ft)
Min. horizontal in situ stress (
h
) 26.9 MPa (3900 psi , or 0.65 psi/ft)
Vertical in situ stress (
v
) 41.4 MPa (6000 psi , or 1 psi/ft)
Changed Max. horizontal in situ stress (
H2
) 36.2 MPa (5250 psi )
Changed Min. horizontal in situ stress (
h2
) 31.4 MPa (4550 psi)
Changed Vertical in situ stress (
v2
) 48.3 MPa (7000 psi)
Thermal conductivity of casing (K
1
) 52W/mK (0.100 Btu(IT) inch/sec/ft
2
/
o
F)
Thermal conductivity of cement (K
2
) 0.9W/mK (0.00173 Btu(IT) inch/sec/ft
2
/
o
F)
Thermal conductivity of rock (K
3
) 1.6W/mK (0.00308 Btu(IT)inch/sec/ft
2
/
o
F)
Linear thermal expansion coefficient of casing (
T1
) 0.000012 1/ K (0.00000667 1/
o
R)
Linear thermal expansion coefficient of cement (
T2
) 0.0000144 1/ K (0.000008 1/
o
R)
Linear thermal expansion coefficient of rock (
T3
) 0.0000125 1/ K (0.00000694 1/
o
R)
Temperature at wellbore (T
,1
) 90
o
C (194
o
F)
Temperature at reservoir (T
,4
) 75
o
C (167
o
F)
Mogi strength parameter (a) 9.8 MPa (1421.5 psi)
Mogi strength parameter (b) 0.47
Cement density 1889.6 kg/m
3
(15.77 ppg)
Unconfined Compressive Strength (UCS) of cement 41.6 MPa (6030.8 psi)
Tensile strength of cement (T
f
) 4.16 MPa (603.08 psi)
Interesting area for cement failure R
2


Fig. 8. The effect of wellbore temperature on tensile cement
failure

5.2. The Effect of the Wellbore Temperature
Variation on Cement Failure
Fig. 8 compares analytical tensile results caused by
wellbore temperature variation with those of numerical
models. As wellbore temperature increases, the
possibility of tensile cement failure decreases. The
interesting part is that analytical results show a linear
relationship between wellbore temperature and the
minimum value of
p
+T
f
, but numerical results show a
logarithmic relationship. Even though further research
has not yet been performed, this discrepancy may be
caused by the difference of the two models in dealing
with stresses caused by temperature variation.
Fig. 9 shows the effects of wellbore temperature on
shear cement failure. The chance of shear cement failure
decreases as wellbore temperature increases for both
analytical and numerical results. However, the numerical
results have a smaller slope than the analytical results
and show different values from those of the analytical
results. Similar to the previous subsection, the analytical
results for tensile and shear failure reach failure faster
than the numerical results.
5.3. The Effect of the Wellbore Pressure on
Cement Failure
Fig. 10 describes the effects of wellbore pressure on
tensile cement failure. Contrary to the previous
subsections, the analytical and numerical results show
very similar behavior. Also, as the wellbore pressure
increases, the chance of tensile cement failure increases.
Moreover, the minimum of
p
+T
f
has a linear
relationship with wellbore pressure. This linearity may
be the result of attributing elastic properties to the
cement.
Fig. 11 characterizes the aftermath of wellbore pressure
on shear failure. Like the tensile failure, the analytical
and numerical results show analogous results except for
Pw = 0. Similar to the previous subsections, the
analytical results reach failure faster than the numerical
results. Similar to the tensile failure, the shear failure
shows the linearity with respect to wellbore pressure.
Fig. 7. The effect of Young's modulus of cement on shear cement failure

Fig. 10. The effect of wellbore pressure on tensile cement failure
6. CONCLUSION AND FUTURE WORK
6.1. Analytical Model
This research shows results in accord with the results of
the previous research [16, 17] when taking the plane
strain condition. Fig. 12 shows the comparison among
them. The bottom of the figure shows input conditions
for the comparison. u is the angle from the direction of

xx
. The values of

at each interface for the each


research are compared.
For the comprehensive models, even if further research
on the effects of other wellbore conditions on cement
failure is required, the analytical model shows similar
results to the numerical model and can be a good tool to
quickly obtain the stress distribution without needing
numerical simulators. Moreover, this analytical model
can be extended to multiple, concentric, inclined cased
wellbores thanks to the simple superposing principles. In
addition, this analytical model can be summarized as
having following characteristics:
(i) Uniqueness
(a) Fully coupled poroelasticity is used.
(b) Solutions using simple superposition
principles are used. This superposition extends
an inclined cased wellbore model to multiple,
concentric, inclined cased wellbores.
(c) Generalized plane strain is utilized.
(d) First analytical model considering all drilling
and well completion processes for an inclined
cased wellbore, including tectonic stress
variation and temperature variation is
developed.

(ii) Limitation
(a) Plasticity is not considered.
(b) Solution is not full 3D but pseudo 3D
(generalized plane strain condition).
Fig. 9. The effect of wellbore temperature on shear cement failure
(c) The coupling between temperature and fluid
contained in rock is ignored.

Fig. 12. Comparison of the results of the previous research and
this study


6.2. Cement Failure
First, as Young's modulus increases, the chance of shear
failure decreases while the chance of tensile failure
increases. These interesting phenomena are caused by
the difference of slopes of Aa/AE
2
and AT
f
/AE
2
. The
difference of slopes among them is due to the coupling
among the tensile and compressive strength of the
cement, and Young's modulus of the cement.
Second, the analytical and numerical models induced by
wellbore temperature variation show distinct results for
tensile and shear failure. The disagreement might be
caused by the different modeling technique. That is, the
analytical models are based on the generalized plane
strain which is pseudo 3D, whereas the numerical
models use a finite box model which is full 3D.
Third, the effects of wellbore pressure show very similar
behavior between analytical and numerical results.
Moreover, both analytical and numerical models show
almost identical linear relationships between the tensile
and shear failure and wellbore pressure, which is caused
by the use of the linear elastic approximation.
Consequently, the following comprehensive conclusions
can be made. First, the analytical models are more
conservative than the numerical models, observed by the
result that in every case, the analytical results always
reach failure faster than the numerical results. That is,
the analytical models may be more preferable to the
numerical models in order to safely design cements for
an inclined cased wellbore system. In addition,
analytical models are faster and require less expense
than numerical models. Second, this research can
introduce a comprehensive map to determine which
factor and to what degree will significantly affect the
Fig. 11. The effect of wellbore pressure on shear cement failure
tensile and shear cement failure, and provide the proper
guidelines for the selection of cements.
6.3. Future Research
Although these analytical and numerical models give
beneficial and helpful results and insights into
understanding mechanical behavior of concentric casing,
cement, and rock formation systems, these models are
based on elastic behavior. However, real casing, cement
and rock behave elastically and plastically. That is, they
are non-linear materials. Thus, further studies
considering the plasticity of materials are needed. Also,
the behavior of some materials depends on a time
history: visco-elastic or visco-plastic materials. This
time-dependent behavior should be considered, too.
In addition, this research ignores the influence of well
completion, such as the effects of cement shrinkage on
cement failure, because reasonable and generally
applicable experimental data are lacking. However, the
initial state of stress of the cement at set may be critical
in predicting cement integrity. Therefore, further studies
on well completion processes are needed.
Finally, these analytical models ignore the coupling
between temperature and fluid contained in rock because
of its complexity. Developing models that consider the
coupling is needed.

ACKNOWLEDGEMENT
Ametek/Chandler Engineering, BJ Services, Chevron,
ExxonMobil and Schlumberger are acknowledged for
sponsoring the Life-Of-Well Rock, Fluid, and Stress
Systems Research Program which funded this research.

REFERENCES

1. Bradley, W.B. 1979. Failure of inclined boreholes.
Journal of Energy Resources Technology, Transactions
of the ASME, 101(4):232-239.
2. Cui, L., Y. Abousleiman, A. H.-D. Cheng and J.-C.
Roegiers. 1999. Time-dependent failure analysis of
inclined boreholes in fluid-saturated formations.
Journal of Energy Resources Technology (American
Society of Mechanical Engineers), 121(1):31-39.
3. Cui, L., A.H.-D. Cheng and Y. Abousleiman. 1997.
Poroelastic solution for an inclined borehole. Journal of
Applied Mechanics, ASME, 64(1):32-38.
4. Cui, L., S. Ekbote and Y. Abousleiman. 1999. Effect of
pore fluid conditions at the borehole wall on borehole
stability, in rock mechanics for industry. In Kranz-R.L.
Scott G.A. Amadei, B. and P.H. Smeallie, editors, 37th
U.S. rock mechanics symposium proceedings, volume
1, pages 187-194, A.A. Balkema, Rotterdam.
5. Cui, L., S. Ekbote, Y. Abousleiman, M. M. Zaman and
J. C. Roegiers. 1996. Borehole stability analyses in
fluid saturated formations with impermeable walls.
International Journal of Rock Mechanics and Mining
Sciences and Geomechanics Abstracts, 35:582-583,
doi:10.1016/S01489062(98)00077-1.
6. Cui, L. and Y. Abousleiman. 2001. Time-dependent
poromechanical responses of saturated cylinders.
Journal of Engineering Mechanics, 127(4):391-398.
7. Ekbote, S., Y. Abousleiman, L. Cui and M. Zaman.
Analyses of inclined boreholes in poroelastic media.
International Journal of Geomechanics, 4(3):178-190,
2004.
8. Stehfest, H. 1970. Acm algorithm 368: Numerical
inversion of Laplace transform[d5]. Communications of
the Association for Computing Machinery (ACM),
13:47-49.
9. Detournay, E. and A.H.-D. Cheng. 1988. Poroelastic
response of a borehole in a non-hydrostatic stress field.
International Journal of Rock Mechanics and Mining
Science & Geomechanics Abstracts, 25(3):171-182.
10. Gray, K.E., E. Podnos and E. Becker. 2007. Finite
element studies of near-wellbore region during
cementing operations: Part I. In SPE(106998-MS), 31
March - 3 April, Oklahoma City, Oklahoma, U.S.A.
11. Chenevert, M.E. and L. Jin. 1989. Model for predicting
wellbore pressures in cement columns. In SPE(19521-
MS), 8-11 October 1989, San Antonio, Texas.
12. Cooke Jr., C.E., M.P. Kluck, and R. Medrano. 1983.
Field measurements of annular pressure and
temperature during primary cementing. Journal of
Petroleum Technology, 35(8):1429-1438, August.
13. Prohaska, M. and D.O. Ogbe. 1993. Determining
wellbore pressures in cement slurry columns. In SPE
(26070-MS), Western Regional Meeting, 26-28 May,
Anchorage, Alaska.
14. Zhou, D. and A.K. Wojtanowicz. 2000. New model of
pressure reduction to annulus during primary
cementing. In IADC/SPE Drilling Conference, 23-25
February 2000, New Orleans, Louisiana, (59137-MS).
15. Li, Y. 1991. On initiation and propagation of fractures
from deviated wellbores. PhD thesis, The University of
Texas at Austin.
16. Atkinson, C. and D.A. Eftaxiopoulos. 1996. A plane
model for the stress field around an inclined, cased and
cemented wellbore. International Journal For
Numerical and Analytical Methods in Geomechanics,
20(8):549-569.
17. Carter, B.J., P.A. Wawrzynek and A.R. Ingraffea. 1994.
Effects of casing and interface behavior in hydraulic
fracture. In Proceedings 8th international conference
on Computer Methods and Advances in Geomechanics,
1994.
18. Morita, Y., N. Kasahara, H. Hikida and Y. Ito. 2005.
Collapse resistance of tubular strings under geotectonic
load. SPE (95691-MS).
19. Timoshenko, S.P. and J.N. Goodier. 1970. Theory of
Elastiicity. 3
rd
ed. McGraw Hill.
20. Ghosn, A.H. 1981. Some Problems In One And Two-
Dimenstional Coupled Thermoelasticity. PhD thesis,
The Department of Mechanical Engineering at
Louisiana State University.
21. Shahri, M.A. 2005. Detecting and modeling cement
failure in high pressure/high temperature wells, using
finite-element method. Master's thesis, Texas A&M
University, December.
22. Duncan Fama, M. E. and M. J. Pender. Analysis of the
hollow inclusion technique for measuring in situ rock
stress. 1980. International Journal of Rock Mechanics
and Mining Science & Geomechanics Abstracts,
17(3):137-146.
23. Zhenye, W. and L. Shiping. 1990. The generalized
plane strain problem and its application in three-
dimensional stress measurement. Int. J. Rock Mech.
Min. Sci. & Geomech. Abstr., 27(1):43-49.
24. Drucker, D.C. and W. Prager. Soil mechanics and
plastic analysis or limit design. Providence, R.I.:
Division of Applied Mathematics.
25. Al-Ajmi, A. 2006. Wellbore stability analysis based on
a new true-triaxial failure criterion. PhD thesis, KTH
Land and Water Resource Engineering.
26. Handin, J. 1965. Strength of oil well cements at
downhole pressure-temperature conditions. SPE(1300).
27. Goodwin, K.J. and R.J. Crook. 1992. Cement sheath
stress failure. SPE Drilling Engineering, 7(4):291-296.
28. Thiercelin, M.J., B. Dargaud, W.J. Baret and J.F.
Rodriquez. 1998. Cement design based on cement
mechanical response. SPE Drilling&Completion,
(52890-PA):266-273, December.
29. Bosma, M., K. Ravi, W. Driel, and G. Schreppers. 1999.
Design approach to sealant selection for the life of the
well. In SPE, (SPE 56536), 36 October, 1999. Houston,
Texas
30. Biot, M.A. 1941. General theory of three-dimensional
consolidation. Journal of Applied Physics, 12(2):155-
164,.
31. Jo, H. 2008. Mechanical Behavior of Concentric and
Eccentric Casing, Cement, and Formation Using
Analytical and Numerical Methods. PhD thesis, The
University of Texas at Austin.
32. Nelson, E.B. and D. Guillot. 2006. Well Cementing.
Schlumberger, 2nd edition edition.
33. Jaeger, J.C., R. Zimmerman, N.G. Cook, N.G.W. Cook
and R.W. Zimmerman. 2007. Fundamentals of Rock
Mechanics. John Wiley & Sons, Limited.
34. Lacy, L.L. 1997. Dynamic rock mechanics testing for
optimized fracture designs. In SPE Annual Technical
Conference and Exhibition, 5-8 October 1997, San
Antonio, Texas, (38716-MS).
35. Dassault Systmes. ABAQUS Version 6.7
Documentation.

You might also like