You are on page 1of 11

SPE-173864-MS

Study of Thermal Variations in Wells During CO2 Injection


Halvor Lund, SINTEF Energy Research; Malin Torsæter, SINTEF Petroleum Research; Svend Tollak Munkejord,
SINTEF Energy Research

Copyright 2015, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Bergen One Day Seminar held in Bergen, Norway, 22 April 2015.

This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents
of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written
consent of the Society of Petroleum Engineers 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 acknowledgment of SPE copyright.

Abstract
Carbon dioxide (CO2) can be injected into the subsurface for the purpose of enhanced oil recovery (EOR),
possibly with geological CO2 storage. This procedure brings about a set of unique challenges with respect
to well construction, operation and remediation. As compared to normal production, CO2 injection
imposes lower temperatures and stronger temperature variations on wells. This is especially prevalent if
the injection is not continuous. Downhole temperature variations will result in expansion or contraction
of casings and well barrier materials, which can cause them to crack or de-bond at interfaces. To avoid
leakage paths through wells it is therefore important to understand within which temperature intervals it
is safe to operate.
In the present paper we describe a heat-conduction model for calculating heat transfer from the well
to the casing, annular seal and rock formation. These materials have dissimilar thermal properties, and will
behave differently upon downhole temperature variations. The model is discretized using a finite-volume
method developed especially for accurate calculation of heat conducted radially in a well. This allows us
to predict temperatures and temperature variations at various locations in and around a given well during
CO2 injection.
To verify the numerical model, we compare our simulation results with the time-varying temperatures
measured in our laboratory. Good agreement is found between the numerical predictions and the measured
data. Simulation results are presented for different combinations of formations and well barrier materials
(cement and alternative types of annular sealants) to display their effect on the well temperature. It is
found that by replacing cement with an annular sealant material with higher thermal conductivity, the
temperature difference between the casing/seal interface and the seal/rock interface can be significantly
reduced. A high-conductivity formation such as halite/rock salt can also reduce thermal gradients in the
well materials.

Introduction
Ensuring well integrity essentially means to avoid leakage from a well during its entire life cycle. This is
critically dependent on maintaining the integrity of annular cement placed between casings or between
casings and rock formations. These annular cement sheaths in wells provide mechanical stability – and
2 SPE-173864-MS

they should ideally ensure zonal isolation by preventing fluid flow from one geological zone to another
(or up to the surface).
Several recent studies show that the integrity of annular cement is difficult to maintain, and that annular
pressure problems in wells become more frequent as they age (Kellingray, 2007; Vignes and Aadnøy,
2008). For a well-formed annular cement sheath, the permeability is typically as low as 1 ⫻ 10⫺5 mD
(Nelson and Guillot, 2006), and annular fluid or pressure transmission is therefore dependent on the
formation of leakage paths (or “microannuli”). Typical mechanisms for microannuli formation are poor
cement placement, cement shrinkage during hardening, or mechanical, chemical or thermal failure of the
set cement material. The latter type of annular cement failure typically manifests itself as either radial
cracking or debonding at casing-cement or cement-rock interfaces.
Downhole temperature variations have long been suspected to be a major cause for annular cement
debonding and cracking. The variations occur as hot hydrocarbons are produced with an on/off pattern
through colder formations, or as a result of on/off injection of fluids that are colder or warmer than the
wellbore itself. When subjected to thermal variations, the components of a well (e.g. casing steel, annular
cement and near-well rock formations) will repeatedly expand and contract. Since all the materials have
different thermal expansion coefficients, stresses are likely to develop in these materials. For the annular
cement sheath, the most damaging stresses are tensile stresses, as the tensile strength of cement is only
one tenth of its compressive strength. Radial cement cracking is caused by the development of tensile
hoop stresses, also referred to as circumferential stresses, which are normal stresses along the tangential
(azimuth) direction. Debonding at cement interfaces is caused by the development tensile radial stresses.
Several experimental studies have been made to investigate the resistance of annular cement sheaths
towards temperature variations. Carpenter et al. (1992) found that thermal cycling is destructive to the
cement-casing bond for various cement types, and that just a single thermal cycle with amplitude 56 °C
could damage the bond. Similar trends, displaying that both pressure and temperature variations aggravate
the casing-cement bond, have been observed by Boukhelifa et al. (2005) and Bois et al. (2012). Detailed
experimental studies of annular cement debonding and cracking have also recently been performed by
Albawi et al. (2014) and De Andrade et al. (2014). Both these studies show that only a few thermal cycles
in the range ⫺20 °C to 120 °C cause cement debonding and cracking that is visible in computed
tomography (CT) images.
The above mentioned works indicate that cement, with its low tensile strength, might not be the ideal
annular sealant material. Thus, current industry effort is directed towards the development of new sealant
material types that have higher tensile strengths and that are more deformable. These are developed not
only for use in annuli, but also for permanent well plugging and abandonment (P&A) operations. Such
emerging sealant material types might one day replace cement, and they comprise e.g. sand slurries
(Svindland, 2004), polymer-based sealants (Beharie, 2012) and low melting point metal alloys (Harrison,
2011). Even if their use in annuli or for plugging is likely to be beneficial for maintaining well integrity,
little field experience has so far been gathered. Further studies are therefore needed to understand the
long-term properties of these new materials, and how their use in well annuli can alter the thermal
situation in a well.
This paper presents a heat-conduction model for studying how the choice of annular sealant material
affects temperature gradients in the well and near-well region. Focus is on studying the temperature ranges
relevant for CO2 injection – which is typically performed for Enhanced Oil Recovery (EOR) or permanent
storage purposes. The expansion of CO2 causes cooling, possibly involving phase change, and this creates
especially difficult thermal cycling situations. To avoid formation of leakage paths it is therefore
important to understand within which temperature intervals it is safe to operate, and how these vary with
the choice of annular sealant material.
SPE-173864-MS 3

Conduction model and numerical procedures

Heat conduction
The heat transport in solid materials is due to heat conduction. In this section we will describe a general
model for conduction through concentric layers with different properties. We consider a case with n
concentric layers of different materials, as illustrated in Fig. 1. Layer k stretches from radius rk to radius
rk-1, and has certain material properties, listed in Table 1. For example, ␳k is the material density of layer
k. This general model allows us to describe a number of different symmetric well configurations including
e.g. tubing, packer fluid, casing, drilling mud, cement and stone formation.

Figure 1—A heat conduction model with n concentric layers and a heat flux Q. Layer k stretches from rk-1 to rk.

Table 1—Properties of each layer in the model depicted in Fig.


1.
Description Symbol SI unit

Density ␳ kg/m3
Heat capacity cp J/(kg K)
Thermal conductivity k W/(m K)

It should be noted that our model does not include the effects of convection, which may be present in
fluid layers such as packer fluid and drilling mud. However, if these fluids are sufficiently viscous,
conduction will be of greater significance than convection. For wells where the length is much larger than
radial distances, convection is also expected to be small (Faghri et al., 2014).
To have a fully defined model, one also needs to prescribe boundary conditions for the innermost and
outermost layer, i.e. at radii r0 and rn. There are two options that are most relevant: Prescribing a
temperature, and prescribing a heat flux. For a well, with flow with a certain velocity and temperature
inside the innermost layer, one typically uses a correlation to calculate the heat flux. In this work we will
instead prescribe the temperature applied by the temperature cycling rig described in the next section. For
the outermost layer, one will typically prescribe the temperature of the formation. This requires that the
layer has a large enough outer radius, so that the temperature can be assumed to be constant. In other
words, the formation temperature at the outer boundary should be unaffected by the heat transferred
to/from the well.
The time evolution of temperature is given by the radial heat equation (Cannon, 1984),
(1)

where K(r) is the thermal conductivity at radius r, given by


4 SPE-173864-MS

(2)

and similarly for ␳(r) and cp(r). Using this formulation, we neglect any heat transferred along the pipe
(in the axial direction). Heat is nevertheless transported along the pipe by the fluid inside the tubing. With
given initial and boundary conditions, we may solve Eq. (1) numerically using the methods described in
the following section.
Numerical discretization
We solve Eq. (1) numerically using a finite volume scheme. The domain is divided into cells of finite
volume, where cell Ci stretches from ri-1/2 to ri⫹1/2. The finite volume scheme can then be derived by
integrating Eq. (1) over each cell,
(3)

where dV ⫽ 2␲r dr for polar coordinates. Since we are working in two dimensions, the “volume” in
this case is actually an area. We then assume that pcp is constant within each cell, which yields
(4)

The second term describes the heat fluxes into and out of each cell, so we define the heat flux as
(5)

We also define ⌰i, as the average temperature in cell Ci, i.e.


(6)

where ri ⬅ ½ (ri-1/2 ⫹ ri⫹1/2) and ⌬ri ⬅ ri⫹1/2 – ri-1/2. Using Eq. (5) and Eq. (6) in Eq. (4) then leads
to a semi-discrete formulation
(7)

The numerical flux Fi⫹1/2 approximates the exact flux f at the interface between cell Ci, and Ci⫹1 based
on the average temperatures ⌰, and ⌰i⫹1,
(8)

Much of the challenge in deriving a finite volume scheme lies in approximating these numerical fluxes
based on the average temperatures ⌰i.
Numerical flux function
We aim to derive an expression for the numerical flux function in steady state, which means that the heat
flux is constant in time. Using this assumption, we can solve for the exact temperature profile in all layers
by using Eq. (5), assuming that each layer has a constant thermal conductivity. This yields
(9)

for the temperature in cell Ci, where To,1 is the temperature at the radius r0 i, and C is a constant. We
can then find the average temperature ⌰i in cell Ci using Eq. (6), and equivalently for cell Ci⫹1. By
requiring that the temperature be continuous, by setting r0 i ⫽ r0,i⫹1 and T0,i ⫽ T0,i⫹1, we can solve for
the numerical flux Fi⫹1/2 ⫽ –C,
SPE-173864-MS 5

(10)

Time integration
The semi-discrete finite-volume formulation (7) contains a time derivative which also needs to be
discretized. The easiest, most straightforward approach is to use an explicit scheme such as the Forward
Euler scheme. A disadvantage with explicit schemes is a limit on the time step length. In this work we
rather use the Crank-Nicolson scheme (Press et al., 2007), which is a second order implicit scheme that
can handle arbitrarily large time steps.

Experimental verification of the model


To verify the numerical model, the simulation results were compared to temperatures previously measured
in the laboratory by Albawi (2013). These experiments consisted in heating and cooling a downscaled well
sample with Castlegate sandstone formation, casing pipe made of steel (type ST 52-3) and Portland G
cement in the annular space between them. The sample was manufactured in a furnace at 66 °C, where
the cement was allowed to harden. The dimensions and material properties of the setup are shown in Table
2, and the geometry is illustrated in Fig. 2.

Table 2—Material properties and dimensions in the experimental setup.


Radial segment Thermal conductivity Specific heat capacity
Item [cm] Density [kg/m3] [W/(m K)] [J/(kg K)]

Air 2.5-2.7 1.1614 0.026 1007


Casing (ST 52-3) 2.7-3.0 7850 40 500
Cement (Portland Class G) 3.0-4.0 1917 0.72 780
Sandstone (Castlegate) 4.0-10.0 2600 2.0 1000

Figure 2—Illustration of the experimental setup, as seen from above. Layer thicknesses are to scale.

Computer-controlled heating and cooling of the sample was performed via a copper rod centred inside
the casing, with air filling the annulus between the thermal rod and the casing. The rod was heated/cooled
at its bottom during the test by means of a commercially available thermal platform. The heating/cooling
rate was 2.3 °C/min, and the rod temperature was cycled between –20 °C and 120 °C. Outer surfaces of
the sample, other than the bottom of the copper rod, were thermally insulated during the test.
6 SPE-173864-MS

The sample was instrumented with two thermocouples on its top surface. One was located at the
junction between the casing and the annular cement sheath, while the other was located at the interface
between the cement sheath and the rock. More details on the experimental setup and procedure is
described in Albawi et al. (2014) and similar experiments using a slightly modified setup have also been
performed by De Andrade et al. (2014).
Comparison of experimental and simulation results
The air between the central copper rod and the steel casing will start to move when temperature gradients
are applied, leading to convection of heat in addition to conduction. The significance of convection
compared to just conduction is challenging to quantify without using three-dimensional simulations,
which is outside the scope of this work. We will therefore model the air layer using only conduction,
which should be a reasonable assumption considering its relatively small thickness. The outer edge of the
sample, i.e. the outer edge of the sandstone, was assumed to be perfectly insulated, so the flux was
prescribed to be zero at this side of the domain. At the inner edge of the air layer, we set the temperature
that was applied by the copper rod.
Fig. 3 shows a comparison between the experimental and simulation results. The solid lines show the
temperature applied by the copper rod, as well as the temperature at the interface between casing and
cement (denoted casing/cement), and between cement and formation (denoted cement/formation). The
results match each other relatively well, when one considers the simplifying assumptions we have made,
such as one-dimensionality, no convection in the air layer, constant thermal conductivity of air, and a
perfectly insulated sample.

Figure 3—Comparison between experiment (solid lines) and model (dashed lines).

Cement/formation temperature
The experimental results show that the cement/formation temperature lies below the casing/cement
temperature both when the temperature is rising and when it is falling. During a period with falling
temperature, such as from ~7 to~12 hours, one would expect the casing/cement temperature to drop before
the cement/formation temperature, since the former is closer to the cold copper rod. Possible explanations
for these results could be lack of insulation, location of temperature sensors or possibly measurement
errors.
Heat loss
In Fig. 3, we notice that the difference between the experimental and simulated temperatures seems to
increase over time. This is most probably due to the fact that the sample is not perfectly insulated. Since
SPE-173864-MS 7

the average applied temperature (around 60 °C) is above the ambient temperature, the sample will lose
some heat to the surroundings during the 48 hours the experiment lasts, which will lower the measured
temperature.
Simulation results: Effect of different annular sealant materials
A number of alternative annular sealant materials exist that could possibly replace cement. In this section,
we wish to investigate what effect these materials may have on thermal gradients in CO2 injection wells.
We consider an injection case where CO2 is injected from a ship over a period of 24 hours, which could
be a typical injection time for a ship (Vermeulen, 2011). It is assumed that the fluid has a temperature of
10 °C when reaching the depth of the annular seal, and that the reservoir temperature is 120 °C. We
assume that all materials are at a temperature of 120 °C before injection starts.
The well is modelled as described in Table 3, with steel tubing, an insulating packer fluid, steel casing,
annular seal and formation. The thickness of each layer has been chosen to represent what may be a typical
CO2 injection well (Renpu, 2011). Our simulations consider four different annular sealant materials,
described in Table 4: Cement, a sand slurry, a thermally setting polymer and a bismuthtin alloy. The
material properties have been obtained from commercial suppliers of such materials. These materials have
widely varying properties, especially in thermal conductivity, where the bismuth-tin alloy is more than
100 times more conductive than the polymer. To include the effect of different formations surrounding the
annular seal, we consider three different possibilities: sandstone, halite/rock salt and shale, described in
Table 5. Sandstone and shale have relatively similar properties, whereas rock salt stands out with a rather
high thermal conductivity. The material properties are based on typical values found in the literature.
Some materials, such as cement and rock, might have thermal properties that depend on pressure and
saturation, which varies from well to well.

Table 3—Material properties and dimensions for the layers composing the well
Thermal conductivity Specific heat capacity
Item Radial segment [cm] Density [kg/m3] [W/(m K)] [J/(kg K)]

Tubing (ST 52-3) 8.5-9.0 7850 40 500


Packer fluid (Halliburton, 2012) 9.0-11.1 1400 0.26 4000
Casing (ST 52-3) 11.1-12.2 7850 40 500
Annular seal 12.2-15.5 See Table 4
Formation 15.5-⬁ See Table 5

Table 4 —Annular sealant material properties


Item Density [kg/m3] Thermal conductivity [W/(m K)] Specific heat capacity [J/(kg K)]

Cement 1917 0.72 780


Sand slurry 2600 2.0 1000
Thermally setting polymer 1030 0.16 1600
Bismuth-tin alloy 8560 19.0 167

Table 5—Formation properties


3
Item Density [kg/m ] Thermal conductivity [W/(m K)] Specific heat capacity [J/(kg K)]

Sandstone 2600 2.0 1000


Rocksalt 2170 7.1 931
Shale 2350 1.0 1300
8 SPE-173864-MS

Fig. 4 shows the temperature over time during an injection, for four locations: Inside of casing (CO2
temp.), the interface between casing and annular seal (casing/seal), the interface between annular seal and
formation (seal/formation), and the formation far from the well. For all annular sealant materials, the
temperature drops from 120 °C until the injection stops after 24 hours. As expected, the bismuth-tin alloy
(Fig. 4c) conducts heat very well, which leads to a very small temperature difference between the two
sides of the annular seal. With a rock salt formation (dashed lines), the temperature of the annular seal only
drops by about 20 °C before increasing again. With a sandstone (solid lines) or shale (dotted lines)
formation, the temperature drops somewhat more, by 40 °C to 55 °C.

Figure 4 —Temperature during and after injection of CO2 at 10 °C into well initially at 120 °C, for various annular sealant materials, and
three different formations: Sandstone (solid lines), halite/rock salt (dashed lines) and shale (dotted lines).

In the opposite case, the low conductivity polymer (Fig. 4d) gives a very large temperature difference
over the annular seal, of the order of 60 °C. This could potentially be a problem if it leads to large thermal
stresses. Cement and sand slurry have a similar temperature response. The temperature of the seal/
formation interface drops by about 20 ° C to 50 ° C, whereas the casing/seal interface temperature drops
by approximately 30 °C to 60 °C, depending on the formation. Rock salt gives the smallest temperature
drop, whereas shale gives the largest.
SPE-173864-MS 9

Discussion
With the present results, it may seem that a high-conductivity annular sealant material, such as a
bismuth-tin alloy, may be beneficial for reducing thermal stresses due to the contraction of the casing
steel. A high-conductivity material will help keep the casing temperature high during injection by
providing good thermal coupling to the surrounding formation rock. This may reduce the risk of
debonding and cracking of the annular sealant material due to a contracting casing. The thermally setting
polymer, with its low thermal conductivity, will allow a low temperature in the casing, leading to
significant contraction of the casing steel. The same arguments apply to the formation, where a high
conductivity formation such as rock salt will contribute to reducing temperature variations and thereby
thermal stresses. It is important to note that the materials’ neutral temperature, i.e. the temperature at
which there are no stresses, will affect which temperature variations are problematic.
The more thermally insulating the annular sealant material is, the bigger the temperature difference
becomes between the casing/seal and the seal/formation interfaces. Whether or not this causes the annular
sealant material to crack or debond depends on its deformability and its thermal expansion coefficient as
compared to steel and rock.
Cement
Cement has a thermal expansion coefficient similar to that of most rocks, but significantly lower than most
types of steel. Debonding of cement at the steel interface is therefore likely upon thermal cycling. When
steel is expanding during heating it will expose the annular cement sheath to strong tensile forces, and
when the steel further contracts during cooling it will contract more than the surrounding cement sheath.
As the cement sheath is rigid and does not easily deform, it is likely to radially crack during heating and
to debond at the cement-casing interface during cooling. This is in agreement with what was observed by
both Albawi (2013), Albawi et al. (2014) and De Andrade et al. (2014).
Sand slurry
The sand slurry can deform upon thermal cycling, and unless it dries out it will retain its bond to both rock
and casing at all times. Well integrity failure through cracking or debonding is thereby not an issue if a
sand slurry is used in the annular space between casing and rock formations.
Bismuth-tin alloy
BiSn has a very similar thermal expansion coefficient to steel, and these two materials are therefore not
likely to debond upon thermal cycling. BiSn also has a high tensile strength, so radial cracking of this
material is not likely. When BiSn is used as an annular sealant it is therefore most likely that debonding
occurs at the BiSn/rock interface. Rocks have thermal expansion coefficients lower than metals, meaning
that they will expand and contract less upon thermal cycling. This means that thermal expansion of BiSn
and casing can cause the rock formation near the BiSn-interface to permanently consolidate. This might
induce significant microannuli if the temperature in the well is subsequently lowered, and the radius of
the BiSn/casing metal sheath decreases.
Thermosetting polymer
The polymer material is thermally insulating, causing a big difference in temperature between the
casing/polymer interface and the polymer/rock interface. This is likely to introduce thermal stresses in the
material that can cause it to fail with time. To predict such failure it would be necessary to test the
mechanical properties of the polymer material, and to map whether they change due to the thermal history
the material has been exposed to.
Well types
Different well types and locations have different requirements when it comes to thermal insulation. In
Arctic wells it can be necessary to ensure good thermal insulation in the uppermost part of the well if this
has low temperatures due to e.g. permafrost bearing formations. If not, the flow of hot hydrocarbons
10 SPE-173864-MS

through the well can cause the permafrost to melt, thereby destabilizing the soil. Thermal insulation is also
beneficial in this case to reduce the formation of wax and hydrates in the tubing. In geothermal wells it
is typically beneficial to thermally insulate the uppermost part of the well (to avoid losing geothermal heat
to the surroundings), while the bottom part of the well should be completed in such a way that maximal
heat transfer is possible (to optimally extract the geothermal heat).
Conclusions
We have implemented a numerical model for radial heat conduction in a well. The model handles layers
with different thermal properties, such as tubing, casing, annular seal and rock formation. We have
verified the model against experimental data from a down-scaled well subjected to varying casing
temperature. Good agreement was observed, considering the simplifications made in the mathematical
model and the uncertainties in the experiment.
We have performed simulations of thermal cycling relevant for CO2 injection into a well with various
annular sealant materials and rock formations. The simulations showed that the temperatures are
significantly dependent on the thermal properties of both the annular seal and the rock formation. Large
temperature variations may be expected during CO2 injection, and this may lead to significant thermal
stresses and possible damage to the annular seal. This should be taken into account for the design and
operation of CO2 injection wells.
In the future we plan to couple the heat conduction model to a well flow model, in order to simulate
CO2 injection. This may allow evaluation of how different injection schemes affect the thermal gradients
in the well.

Acknowledgement
This publication has been produced in the Ensuring well integrity during CO2 injection project and the
BIGCCS Centre. The authors acknowledge the support of the following partners: ConocoPhillips, Gassco,
Shell, Statoil, TOTAL, GDF SUEZ and the research programmes CLIMIT and Centres for environment-
friendly energy (FME) of the Research Council of Norway (233893 and 193816).

NOMENCLATURE
Symbol Quantity Dimension Unit

cp Heat capacity L2/t2 T J/(kgK)


f, F Heat flux mL/t3 W/m
r Radius L m
°
T Temperature T C
K Thermal conductivity mL/Tt3 W/(m K)
␳ Density m/L3 kg/m3
⌰ Cell-averaged temperature T K

References
A. Albawi. Influence of thermal cycling on cement sheath integrity. Master’s thesis, Norwegian
University of Science and Technology, Faculty of Engineering Science and Technology, Depart-
ment of Petroleum Engineering and Applied Geophysics, 2013.
A. Albawi, J. De Andrade, M. Torsster, N. Opedal, A. Stroisz, and T. Vralstad. Experimental set-up
for testing cement sheath integrity in arctic wells. 2014. OTC 24587, presented at the Arctic
Technology Conference, Houston, Texas, USA.
C. Beharie. Thermaset® - an alternative plugging material. Plugging & Abandonment Workshop,
Stavanger, June 14 2012.
A. -P. Bois, A. Garnier, G. Galdiolo, and J.-B. Laudet. Use of a mechanistic model to forecast
cement-sheath integrity. SPE Drilling & Completion, 27(2), 2012. SPE 139668.
SPE-173864-MS 11

L. Boukhelifa, N. Moroni, S. G. James, S. Le Roy-Delage, M. J. Thiercelin, and G. Lemaire.


Evaluation of cement systems for oil and gas well zonal isolation in a full-scale annular geometry.
In IADC/SPE Drilling Conference, Dallas, Texas, USA, 2005. doi: 10.2516/ogst/2012087. IADC/
SPE 87195.
J. R. Cannon. The One-Dimensional Heat Equation. Encyclopedia of Mathematics and its Applica-
tions. Cambridge University Press, 1984. ISBN 9780521302432.
R. Carpenter, J. Brady, and C. Blount. Effects of temperature and cement admixes on bond strength.
Journal of Petroleum Technology, pages 880 –885, 1992.
J. De Andrade, M. Torsster, J. Todorovic, N. Opedal, A. M. Stroisz, and T. Vralstad. Influence of
casing centralization on cement sheath integrity during thermal cycling. Society of Petroleum
Engineers, Texas, USA, 4-6 March 2014. IADC/SPE 168012, IADC/SPE Drilling Conference and
Exhibition.
A. Faghri, Y. Zhang, and J. R. Howell. Advanced mass and heat transfer, 2014. http://www.thermal-
fluidscentral.org/e-resources/download.php?id⫽57.
Halliburton. N-SOLATE® High Performance Insulating Packer Fluids, 2012. URL http://www.hal-
liburton.com/public/bar/contents/Data_Sheets/web/Sales_Data_Sheets/H05923.pdf. [Online; ac-
cessed 29 September 2014].
L. Harrison. Seal of approval: Wells tested for better sealing with fusible metal alloy. New Technology
Magazine, 2011.
D. Kellingray. Cementing - Planning for success to ensure isolation for the life of the well. SPE
Distinguished Lecturer Series, 2007. SPE 112808-DL.
E. Nelson and D. Guillot. Well Cementing. Developments in petroleum science. Schlumberger, 2006.
ISBN 9780978853006.
W. H. Press, S. A. Teukolsky, W. T. Vetterling, and B. P. Flannery. Numerical recipes. Cambridge
University Press, third edition, 2007. ISBN 978-0-521-88068-8.
W. Renpu. Advanced Well Completion Engineering. Elsevier Science, 2011. ISBN 9780123858696.
A. Svindland. Particulate matter plug for plugging a well. Patent: US 6 715 543 B1, 2004.
T. N. Vermeulen. Knowledge sharing report - CO2 liquid logistics shipping concept (LLSC): Overall
supply chain optimization. Tebodin Netherlands B.V., The Hague, 2011.
B. Vignes and B. Aadnøy. Well integrity issues offshore Norway. 4-6 March 2008. SPE/IADC
112435, presented at the SPE/IADC Drilling Conference and Exhibition held in Orlando, Florida.

You might also like