Post on 23-Dec-2016
transcript
ORIGINAL ARTICLE
Monitoring above-zone temperature variations associatedwith CO2 and brine leakage from a storage aquifer
Mehdi Zeidouni • Jean-Philippe Nicot •
Susan D. Hovorka
Received: 15 June 2013 / Accepted: 15 January 2014
� Springer-Verlag Berlin Heidelberg 2014
Abstract CO2 injection in saline aquifers induces tem-
perature changes owing to processes such as Joule–
Thomson cooling, endothermic water vaporization, exo-
thermic CO2 dissolution besides the temperature discrep-
ancy between injected and native fluids. CO2 leaking from
the injection zone, in addition to initial temperature con-
trast due to the geothermal gradient, undergoes similar
processes, causing temperature changes in the above zone.
Numerical simulation tools were used to evaluate temper-
ature changes associated with CO2 leakage from the stor-
age aquifer to an above-zone monitoring interval and to
assess the monitorability of CO2 leakage on the basis of
temperature data. The impact of both CO2 and brine
leakage on temperature response is considered for three
cases (1) a leaky well co-located with the injection well, (2)
a leaky well distant from the injector, and (3) a leaky fault.
A sensitivity analysis was performed to determine key
operational and reservoir parameters that control the tem-
perature signal in the above zone. Throughout the analysis
injection-zone parameters remain unchanged. Significant
pressure drop upon leakage causes expansion of CO2
associated with Joule–Thomson cooling. However, brine
may begin leaking before CO2 breakthrough at the leakage
pathway, causing heating in the above zone. Thus, unlike
the pressure which increases in response to both CO2 and
brine leakage, the temperature signal may differentiate
between the leaking fluids. In addition, the strength of the
temperature signal correlates with leakage velocity unlike
pressure signal whose strength depends on leakage rate.
Increasing leakage conduit cross-sectional area increases
leakage rate and thus increases pressure change in the
above zone. However, it decreases leakage velocity, and
therefore, reduces temperature cooling and signal. It is also
shown that the leakage-induced temperature change covers
a small area around the leakage pathway. Thus, tempera-
ture data will be most useful if collected along potential
leaky wells and/or wells intersecting potential leaky faults.
Keywords CO2 geological storage � Leakage �Temperature monitoring � Well � Fault
Introduction
Deep geological formations are used widely for fluid dis-
posal and are the main candidates for storage of CO2 as a
means of reducing greenhouse gas emissions to the atmo-
sphere. Safety of storage projects depends on the contain-
ment of CO2 within the target zone. However, the
confining system overlying the aquifer may include leak-
age pathways that permit injected and/or native fluids to
leak to subsurface formations and/or to the surface. Tem-
perature measurements at the injection zone and an over-
lying permeable zone can be acquired accurately at high
resolution at a low cost, from which information on CO2
leakage may be inferred. Temperature monitoring at the
subsurface has been used widely in the oil and gas industry
to track gas migration (Marinaro et al. 2011) and to mon-
itor hydraulic fracturing (Hoang et al. 2011). The thermal
response may have the potential to determine the fate of
injected CO2 during and after CO2 injection for seques-
tration purposes. Temperature has been monitored to assess
Electronic supplementary material The online version of thisarticle (doi:10.1007/s12665-014-3077-0) contains supplementarymaterial, which is available to authorized users.
M. Zeidouni (&) � J.-P. Nicot � S. D. Hovorka
Bureau of Economic Geology,
The University of Texas at Austin, Austin, TX, USA
e-mail: mehdi.zeidouni@beg.utexas.edu
123
Environ Earth Sci
DOI 10.1007/s12665-014-3077-0
CO2 storage performance in demonstration projects,
including Cranfield field, Mississippi, USA (Hovorka et al.
2013) and Ketzin, Germany (Liebscher et al. 2013). CO2
leakage from the injection zone will go through processes
that may induce a detectable temperature signal in the
above-zone monitoring interval (AZMI). These processes
are the same as those causing temperature changes in the
injection zone, namely the Joule–Thomson (JT) effect,
CO2-brine mutual dissolution, and heat exchange with the
surrounding rock-fluid system to which can be added the
geothermal gradient contributing to the cooling of leakage
fluids toward the surface.
Pressure propagates in the reservoir and decreases as
radial distance from the wellbore or wellfield increases.
Under reservoir adiabatic conditions and at the limit
DP ? 0, the process is isenthalpic, and temperature change
due to pressure reduction can be estimated by the JT
coefficient:
lJT ¼ limDP!0
DT
DP
� �H
¼ oT
oP
� �H
ð1Þ
Under the practical P–T conditions encountered for CO2
geological storage applications, the JT coefficient is a
positive number (Han et al. 2010). Therefore, pressure
reduction will be accompanied by thermal cooling. The
temperature reduction accompanied by fluid expansion is
referred to as JT cooling, even though the adiabatic condi-
tion assumption may not be fully valid. Thermal effects can
also be caused by mutual dissolution of CO2 and brine and
heat losses to underburden/overburden. At conditions suit-
able for CO2 storage (P 10–40 MPa and T 30–80 �C), the
enthalpy change (DH) associated with CO2 dissolution in
brine is negative, indicating an exothermic process. On the
other hand, vaporization of water into a CO2-rich phase is an
endothermic reaction and causes cooling (Han et al. 2010).
Assuming single-phase, steady-state flow and constant
thermophysical properties, Mathias et al. (2010) derived an
analytical solution to evaluate JT thermal cooling during
CO2 injection in a saline aquifer. They were able to closely
match numerical results obtained using the TOUGH2
numerical simulation package. Assuming constant ther-
mophysical properties of CO2 (density, viscosity, specific
heat capacity, and JT coefficient), their analytical solution
marginally overestimates temperature decline.
The thermal signature of CO2 injection in saline aquifers
caused by the above-mentioned processes is well studied in
the literature. Andre et al. (2010, 2011) investigated ther-
mal processes triggered by CO2 injection, including the JT
effect, heat of vaporization, and heat of dissolution. The JT
effect was identified mainly in the dry-out zone, whereas
CO2 dissolution was found to cause subtle large-scale
thermal effects. Andre et al. (2010) concluded that the
impact of thermal processes has to be considered mainly in
three types of scenarios: (1) for aquifers at PVT conditions
close to the critical point of CO2 at Pcrit = 7.38 MPa,
Tcrit = 31.04 �C; (2) for reservoirs at PVT conditions
favorable to a good expression of the JT coefficient, such as
in depleted gas fields (Oldenburg 2007); and (3) in the
presence of a large discrepancy between injection tem-
perature and reservoir temperature. Han et al. (2010)
investigated thermal effects within the injection well by
considering the effect of adiabatic compression (JT effect),
frictional energy loss, and heat exchange between injected
CO2 and the rock and fluids surrounding the injection-well
path from the surface. Adiabatic heating may be observed
under moderate injection rates, whereas high injection rates
maximize heat losses due to friction and reduce heat
exchange with the surroundings. Generally, the JT coeffi-
cient has an inverse relation with the density of CO2.
Tracking the CO2 plume using temperature measure-
ments in the injection zone has also been investigated. A
visible temperature signal may be detectable upon CO2
arrival at an observation well in the injection zone, as
shown by Hovorka et al. (2006) for the Frio CO2 injection
pilot test. Bielinski et al. (2008) investigated the possibility
of using temperature-measurement devices for CO2 moni-
toring in the injection zone. They studied the effect of
various parameters on the time-dependent temperature
signal measured at an observation well at some distance
from the injection point. They showed that a detectable
temperature effect at the observation well could be
obtained from which information on the shape of the CO2
plume (and, thus, about permeability distribution) could be
achieved. Hurter et al. (2007) examined the suitability of
thermal methods to monitor CO2 leakage. They discussed
the dynamic temperature signal associated with CO2
leakage and movement of CO2 in the porous rock. They
suggested that CO2 leakage can produce significant cooling
that may be easily detectable. Singh et al. (2012) have also
shown importance of cooling in the injection zone for CO2
injected in a gas field.
In relatively static conditions, long after the end of the
CO2 injection period, thermal equilibrium prevails and
temperature gradient is controlled only by heat conduc-
tivity of the system. Heat conductivity of the porous
medium is a function of heat conductivity of the rock, as
well as its containing fluids. Because a rock filled with
brine has larger conductivity than a rock filled with CO2 or
both CO2 and brine, the presence of CO2 will make the
porous media less heat conductive. As a result, heat will be
more preserved, and temperature increase caused by
reduced heat conduction in the reservoir containing CO2
will be observed as an effect known as the blanketing effect
(Hurter et al. 2007; Wangen 1995; Zhang 1993). During
the injection period, dynamic-flow conditions prevail. Heat
will be transferred both by advection (owing to migration
Environ Earth Sci
123
of injected CO2 into the reservoir) and by conduction.
Given that CO2 migration in the reservoir is faster, the
thermal signature is controlled mainly by heat advection.
Heat conduction will affect the thermal signature only
slightly by reducing heat conductivity at the two-phase
(CO2-rich and aqueous phases) region. As a result, at
locations of increased CO2 saturation, heat transport may
be hindered.
Temperature can be monitored using point measure-
ments, wireline-deployed instruments that produce logs of
temperature versus depth, or fiber-optic distributed tem-
perature sensing (DTS) cables. In spite of its lower reso-
lution compared to conventional temperature sensors (e.g.
quartz gauges), DTS technology has the advantage of
providing continuous temperature measurements both in
time and in space along the cable. Upon careful calibration,
DTS can provide resolution and accuracy as small as 0.02
and 0.3 �C, respectively, as reported for a gas-hydrate
monitoring application at *1,200 m depth (Henninges
et al. 2005; Bielinski et al. 2008).
In this study, the usefulness of temperature data in
detecting leakage is investigated. First the numerical simu-
lation tool (CMG-GEM 2012) to be employed in modeling
the thermal response coupled with fluid flow is briefly dis-
cussed. Next, the capability of the CMG-GEM simulator in
modeling thermal response is assessed by comparing its
results with those reported using TOUGH2 (Pruess et al.
1999) for a benchmarking example problem. Then the
example problem is modified to account for leakage to the
overlying formation. Finally, the thermal response due to
leakage is investigated in three scenarios: (1) leakage
occurring through a leaky well co-located with the injection
well, (2) leakage through a leaky well located at a distance
from the injection well, and (3) leakage through a fault at a
distance from the injection well. Sensitivity of the temper-
ature signal to various parameters is investigated for the first
scenario.
Thermal simulation using CMG-GEM simulator
In modeling thermal processes in the reservoir, the CMG-
GEM (2012) compositional simulator is used. CMG-GEM
calculates temperature distribution by solving the energy
balance equation, coupled with flow equations. The energy
balance equation includes convection, conduction, and heat
loss to the surroundings. Enthalpy of the phases should be
calculated for the energy balance equation and is obtained
by:
H ¼ DHE þ H� ð2Þ
where DHE is excess enthalpy of the fluid and H* is ideal
gas enthalpy at zero pressure and temperature. Excess
enthalpy can be calculated from the equation of state (EOS)
on the basis of the following:
DHE ¼ RTðZ � 1Þ þZV
1
ToP
oT
� �� P
� �dV ð3Þ
where R is the universal gas constant and Z is the
compressibility factor; and P, V, and T are pressure, molar
volume, and temperature, respectively. Given EOS type
and parameters, DHE can be obtained. The ideal gas
enthalpy, H*, for each fluid is the sum of ideal gas
enthalpies of the fluid components weighted by their mole
fractions. The enthalpy of each component is obtained by
the following 5th-degree polynomial (API Technical Data
Book, Petroleum Refining 1983):
H� ¼X5
j¼0
ajiTj ð4Þ
The coefficients aji are referred to as enthalpy coeffi-
cients that are listed in Table 1 for CO2 and brine.
As a result of changes in composition of the aqueous and
CO2-rich phases upon mutual dissolution, enthalpy of the
phases changes accordingly. For the range of injection-
zone temperatures for CO2 storage, enthalpy of the aqueous
phase decreases upon dissolution of CO2 (heat is gener-
ated), and enthalpy of the CO2-rich phase increases as
water vaporizes into the CO2-rich phase and heat is con-
sumed (Koschel et al. 2006). Therefore, temperature is
likely to increase at the dissolution front, and cooling may
occur at the vaporization front close to the wellbore. Heat
loss to the surrounding rock is incorporated using the
method of Vinsome and Westerveld (1980) and is con-
trolled by the cross-sectional area for heat loss, rock heat
capacity, rock mass density, and rock thermal conductivity.
Cubic EOS, including PR (Peng and Robinson 1976), is
not accurate in calculating solubility of CO2 in the aqueous
phase. For the CMG-GEM (2012) simulator, solubility is
obtained by applying the thermodynamic equilibrium.
Fugacity of CO2 in the CO2-rich phase is evaluated on the
basis of PR EOS, and fugacity of CO2 in the aqueous phase
is calculated based on Henry’s law:
fCO2¼ xCO2
HCO2ð5Þ
where xCO2and HCO2
are the mole fractions of CO2 in the
aqueous phase and CO2 Henry’s constant, respectively.
HCO2is a function of pressure, temperature, and salinity.
Variation of Henry’s constant with pressure and tempera-
ture is obtained on the basis of the Harvey (1996) corre-
lation. The effect of salinity on Henry’s constant is
obtained as described by Bakker (2003). Thermodynamic
equilibrium is applied to modeling of H2O vaporization in
the CO2-rich phase, for which the fugacity of H2O in the
Environ Earth Sci
123
CO2-rich phase is calculated based on PR EOS. Fugacity of
water in the aqueous phase is calculated on the basis of the
water mole fraction in the aqueous phase combined with
equations given by Canjar and Manning (1967) and Saul
and Wagner (1987).
Benchmarking
To validate results of the numerical simulation performed
by CMG-GEM, they are compared against those reported
in the literature by Han et al. (2012). They presented results
for CO2 injection in an idealized homogeneous, 2-D radial
zone using TOUGH2-ECO2N (Pruess 2005; Pruess et al.
1999). For their example problem (see Table 2 for detailed
specifications), the reservoir is bounded top and bottom by
no flow boundaries and extends 5 km radially. The depth to
the top of the injection zone is 900 m, its thickness is
182.9 m and is divided into 30 layers. Reservoir porosity is
10 %, and its lateral and vertical permeabilities are 10 and
1 mD, respectively. Aqueous- and CO2-rich-phase relative
permeability curves are obtained on the basis of models
given in Table 3. Reservoir initial temperature is 41.7 �C,
and its initial pressure varies over a 6.89- to 8.9-MPa range
from top to bottom, changing hydrostatically with depth.
CO2 is injected at a rate of 50 kton/year (69,185 sm3/day)
Table 1 Enthalpy coefficients for CO2 and brine used in CMG-GEM simulation
a0i a1i a2i a3i a4i a5i
CO2 0.097 0.158 -0.034e-03 0.148e-6 -0.966e-10 2.074e-14
Brine -1.930 0.447 -0.022e-3 0.030e-6 -0.057e-10 0.028e-14
Table 2 Descriptions of
example problems
* sm3, m3 at surface conditions
of 1 atm pressure and 25 �C
temperature; Empty cells, same
properties as left column; –,
property not applicable
Parameter Han et al.
(2010)
Base
case
Co-located
leaky well
Not-co-located
leaky well
Leaky
fault
Thermal gradient (�C/m) 0 0 0.03 0.03
Thermal conductivity [W/(m K)] 2.51
Rock density (kg/m3) 2,650
Rock heat capacity [J/(kg.K)] 1,000
Injected CO2 temperature (�C) 41.7
Injection-zone initial temperature (�C) 41.7
Injection-zone porosity (fraction) 0.1
Injection-zone lateral permeability (m2) 1e-14
Injection-zone vertical permeability (m2) 1e-15
Injection-zone thickness (m) 182.9 55
Reservoir radial extent (km) 5 24
Brine salinity (wt%) 15
Initial pressure at 1,000-m depth (MPa) 8.03
Injection-well radius (m) 0.3
Injection rate (kton/year, sm3*/day) 50, 69,185
Injection period (years) 2
Seal thickness (m) – – 5
AZMI thickness (m) – – 10
AZMI porosity (fraction) – – 0.1
AZMI lateral permeability (m2) – – 1e-14
AZMI vertical permeability (m2) – – 1e-15
Leaky-well radius (m) – – 0.3
Leaky pathway lateral permeability (m2) – – 1e-14
Leaky pathway vertical permeability (m2) – – 1e-14
Leaky pathway porosity (fraction) – – 0.1
Leak-injector distance (m) – – 0 100 100
Leaky fault width (m) – – – – 2
Environ Earth Sci
123
for 2 years in a 24.4-m-thick interval 12.2 m from the
bottom of the reservoir. Injected CO2 temperature is
identical to the initial reservoir temperature (41.7 �C), and
brine salinity is 15 wt%. Rock density and heat capacity
are 2,650 kg/m3 and 1,000 J/(kg K), respectively. Total
thermal conductivity is 2.51 W/(m K).
Compositional properties of aqueous and CO2-rich
phases used in CMG-GEM versus TOUGH2-ECO2N are
shown for a range of pressures and temperatures
encountered in simulating the Han et al. (2012) example
problem (Figure 1, ESM only). A good match is observed
for aqueous- and CO2-rich-phase viscosities and densi-
ties, as well as mutual solubilities. The temperature
change versus distance from the injection well at the
uppermost layer where CO2 is injected is shown in Fig-
ure 2 (ESM only) after 2 years of injection. In spite of
closely matching compositional properties, the tempera-
ture response given by TOUGH2 differs from that
obtained by CMG-GEM. TOUGH2 provides greater
heating, which corresponds to dissolution of CO2 and
brine and CMG-GEM provides greater JT cooling.
However, temperature responses have similar features,
displaying cooling close to the wellbore and heating
toward the CO2 front. Nevertheless, CMG-GEM is used
because of its more versatile compositional module
applicable to a broader range of temperature and pressure
conditions. Its use is also justified by a computed tem-
perature signal possibly smaller than the actual field
signal as approximated by TOUGH2 results. Conclusions
from this paper are therefore conservative in terms of
using the temperature signal that may be slightly stronger
than described here. Note that salt precipitation (Zeidouni
et al. 2009) is not considered here, which may affect the
thermal response near the wellbore. The pressure and
temperature range in this benchmarking example from
Han et al. are close to the critical point of CO2 where
small changes in pressure/temperature can cause large
changes in CO2 properties. So the thermal effects
reported tend to be the largest possible. The thermal
effects for the base case problem and leakage scenarios
studied in this paper will also tend to be the largest as
these cases are introduced based upon the Han et al.’s
example properties.
Base case
For the purpose of this work the parameters of the example
problem were modified to a situation in which CO2 is
injected over the whole thickness of the injection zone.
Also depth to the top of the injection zone was increased to
1,000 m and reduced injection-zone thickness to 55 m
(Table 2). Thus, the reservoir now extends from 1,000 to
1,055 m depth. While the reservoir depth was changed, the
pressure variation with depth is kept the same as that for
the benchmark model. Thus, at 1,000 m depth the pressure
is the same as that for the benchmark model at
*8.03 MPa. The reservoir radius was also increased from
5 to 24 km and the number of grids in radial direction from
60 to 400. The grid size in vertical direction remains
unchanged. This example is referred to as the base case, the
thermal response corresponding to which is shown in
Figure 3 (ESM only). After 2 years of injection, the max-
imum temperature change for the base case is 0.7 �C,
compared with 1.4 �C for the benchmark example. When
vertical gridding is neglected and the model is treated as a
single layer, more thermal heating and less cooling caused
by mutual dissolutions will be observed. The uppermost
layer of the base case will go through more vaporization
(and, thus, more cooling) than that of the single-layer case.
This is due to easier flow of the fresh CO2 caused by
increased permeability with respect to the CO2-rich phase
in the uppermost layer. Vaporization and dissolution fronts
for the uppermost layer of the base case will then be ahead
of those for the single-layer case. This implies more dis-
solution and greater heating of the uppermost layer could
be expected. However, the CO2-laden brine in the two-
phase region of the uppermost layer flows continuously by
gravity to the deeper layers, despite its slightly elevated
temperature, as it is being replaced by the buoyant CO2-
rich phase. As a result, less heating will be observed in the
uppermost layer.
The thermal response in Figure 3 (ESM only) can be
further compared with the analytical solution given by
Mathias et al. (2010), considering JT cooling as the only
thermal process. They derived the following equation for
the temperature variation near the wellbore:
Tðr; tÞ ¼
aqlf
4phkln 1� cDqt
phr2
� �þ T0 for
cDqt
phþ r2
w\r2
aqlCO2
4phkln
r2w
r2
� �þ Tw for
cDqt
phþ r2
w� r2
8>><>>:
ð6Þ
where T0 and Tw are initial temperature and injection-
well temperatures, respectively; q is volumetric injection
rate; rw is injection-well radius; lf is CO2-rich-phase
viscosity; h and k are injection-zone thickness and (lateral)
Table 3 Relative-permeability curves
Aqueous phase: van Genuchten function (1980)
kra ¼ffiffiffiffiffiS�p
1� 1� S�½ �1=k� �k
� �2
where: S� ¼ Sa�Swirr
1�Swirr, k = 0.457,
Swirr = 0.2
CO2-rich phase: Corey curve (1954)
krg ¼ 1� S n
1� S2
where: S ¼ Sa�Swirr
1�Swirr�Sgc, n = 2, Sgc = 0.05
Environ Earth Sci
123
permeability, respectively; and r and t are radial distance
and time. The constant cD is given by
cD ¼qf cf
/ð1� SwirrÞqf cf þ /Swirrqwcw þ ð1� /Þqscs
ð7Þ
where q and c are density and specific heat capacity,
respectively; subscripts f, w, and s denote the CO2-rich
phase, aqueous phase, and rock, respectively; and Swirr
and / are irreducible water saturation and porosity,
respectively. For the analytical model constant JT coef-
ficient of 3 K/MPa and CO2 heat capacity of 1,016 J/
(kg K) were used, considering pressure and temperature
conditions at the injection well. The analytical model
displays less cooling (due to ignored vaporization) and no
heating (because of neglected dissolution) because it does
not consider the thermal effects corresponding to mutual
dissolution of CO2 and brine. The temperature minimum
will be closer to the CMG-GEM solution when the
vaporization of water in CO2 is neglected (Figure 3, ESM
only).
Defining a temperature change below which the thermal
response may be indistinguishable from instrument and
background noise is essential. On the basis of temperature
measurements at the Cranfield project (Nunez-Lopez and
Hovorka 2012) shown in Figure 4 (ESM only), it was
determined that temperature changes [0.15 �C (0.27 F)
may be considered a signal. Capability of acquiring high
resolution DTS data has also been demonstrated at Ketzin
site where the temperature has been extensively monitored
(Reinsch et al. 2013). The area of the thermal pulse is
therefore defined where the temperature change is
[0.15 �C in either direction from the initial temperature.
The modeled thermal pulse after 2 years of injection is
shown in Figure 5 (ESM only) and compared to the pres-
sure pulse and the location of the CO2 plume. The pressure-
change cutoff in defining the pressure pulse is 1 kPa as
suggested by the resolution of pressure sensors currently
used in petroleum industry and background noises such as
tidal effects (Zeidouni 2011 and references therein; Nogues
et al. 2011; Zeidouni and Pooladi-Darvish 2012). Whereas
the pressure pulse reaches 12.1 km from the injection well,
the temperature pulse does not extend beyond 42 m after
2 years. The CO2 plume at this time covers a 190-m radius
area. The thermal pulse covers a much smaller region than
the pressure signal or the CO2 plume (Figure 5, ESM only)
mainly because the large heat capacity of the rock and
brine weakens the temperature signal accompanied by CO2
flow into the reservoir.
Radius of the area covered by the temperature and
pressure pulses and the CO2 plume is plotted versus time in
Figure 6a (ESM only). The CO2 plume and pressure-pulse
radii are larger than the temperature pulse by one and four
orders of magnitudes, respectively.
Co-located leaky-well scenario
To investigate temperature response to leakage from the
storage formation, the base case is modified to allow for
leakage to an above-zone monitoring interval (AZMI)
above the injection horizon. The AZMI is separated from
the injection zone by an otherwise sealing interval. Unlike
the base case the geothermal gradient is accounted for by
considering 0.03 �C/m of temperature increase with depth
and 41.7 �C at the base of the reservoir at 1,053 m depth.
For this case (see Table 1 for detailed properties), thickness
of AZMI and seal is 10 and 5 m, respectively. All petro-
physical properties of AZMI are identical to those of the
injection zone. For convenience, a leaky well located at the
same location as the injection well is considered. This
maximizes the leakage rate to the AZMI while reducing
computational cost and it can be realistic as the injection
well itself may be leaky. The 10-m-thick AZMI is divided
into five layers. The leakage pathway has lateral and ver-
tical permeabilities of 10 mD with a 0.3-m radius.
Significant pressure drop occurs when CO2 flows from
the seal to the AZMI. After 2 years, the pressure drop from
the upper layer of the injection zone to the lower layer of
the AZMI is 2.40 MPa. As a result, the maximum JT
cooling effect occurs at the base of the AZMI, which is also
in closest thermal contact with the injection zone. Tem-
perature variation (after 2 years of injection) with distance
is shown for the base of the AZMI and compared with
those in the top layer of the injection zone and the seal
(only one layer) in Figure 7 (ESM only). The top layer of
the injection zone is initially at lower temperature than the
injected CO2. As a result, the thermal cooling caused by JT
effect appears as a local minimum followed by further
reduction in temperature with radial distance from the
injection well to reach the initial reservoir temperature.
Temperature values in the seal as well as in the AZMI are
characterized by an increase with radial distance from the
leak. As CO2 leaks toward the AZMI, it undergoes
expansion accompanied by significant JT cooling. There-
fore, what enters the AZMI (and is in communication with
the seal) is cooler than the rock and native fluids. Heat
transfer of the leaked CO2 with the rock and fluid system
leads to cooling of the AZMI and of the sealing interval. It
is visible that far from the leak the temperature in the seal
slightly increases which is due to heat transfer from the
injection-induced slightly elevated temperature in the
injection zone. Note that owing to the opposite effect of
pressure and temperature on CO2 viscosity, mobility of the
leaking CO2 changes only slightly. To present results in the
form resembling field DTS data, temperature change (T–Ti)
was plotted versus depth and time at the location of the
well (Fig. 1) for the 2-year injection period. The temper-
ature increases for all the layers of the injection zone
Environ Earth Sci
123
except for the bottom layer where the temperature change
is negligible. This is because the injected CO2 is at same
temperature as the bottom layer of the injection zone but at
higher temperature than the upper layers. The maximum
temperature increase of *1.5 �C is at the uppermost layer
of the injection zone. In the AZMI, significant cooling is
observed, especially at its base, and cooling becomes more
pronounced with time causing temperature reduction of
*5.7 �C. The temperature change in AZMI’s base already
exceeds the detection limit (0.15 �C) 7 h after start of CO2
leakage. The leakage rate after 2 years (that reaches a
maximum quickly after the start of injection) is 1.55 m3/
day which is 0.7 % of the injection rate at reservoir con-
dition (218.4 m3/day).
The area covered by the temperature pulse in the AZMI
is plotted versus time and is compared with that covered by
the CO2 plume and the pressure pulse in Figure 6b (ESM
only). Results are similar to those given for the injection
zone. As was already concluded in the base case, the area
covered by the temperature pulse is much smaller than that
covered by the CO2 plume and pressure pulse. After
2 years of leakage, the temperature pulse reaches only
13 m from the leak, as compared with 17 and 8,400 m for
CO2 plume and pressure pulse, respectively.
Sensitivity analysis
In the following, sensitivity of the temperature pulse to
different AZMI parameters is investigated. In analyzing
temperature response in the injection zone, Han et al.
(2012) identified the following as key parameters affecting
temperature change: porosity, vertical and horizontal per-
meabilities, injection rate, and injected CO2 temperature.
These parameters are varied for the AZMI and their effect
on the temperature pulse is investigated. Other important
parameters are leakage-pathway permeability, leakage
interval (length of leakage pathway) and rock heat capac-
ity. Throughout the analysis injection-zone parameters
remain unchanged.
Leakage-pathway permeability (kl)
Increasing leakage-pathway permeability increases leakage
rate, which enhances thermal advection in the AZMI.
Increasing leak permeability from 10 to 100 mD increases
the radius of the temperature pulse from 13 to 26 m
(Fig. 2a). Also, increasing thickness of the seal separating
the injection zone and the AZMI increases the leakage
interval. As a result, the leakage rate decreases, leading to a
reduced extent of the temperature pulse.
AZMI porosity
Varying AZMI porosity (assuming all pores contribute to
flow) may affect the extent of the thermal pulse. Higher
porosity limits CO2 migration and reduces heat advection
in the upper zone as a result of which temperature signal
should cover a smaller area. Simultaneously, higher
porosity allows for less heat adsorption by rock grains.
Heat adsorption by rock grains is a function of contact
surface area as well as grain volume. Based on Carman–
Kozeny equation (Carman 1956) increasing porosity at
constant permeability means increased contact surface
area per unit grain volume. The increased contact area
increases heat convection between the fluids and rock
grains. As a consequence, fluids and rock grains reach
equilibrium more quickly. It follows that the thermal
pulse covers a smaller area in response to increasing the
porosity. On the other hand, CMG-GEM already assumes
local temperature equilibrium and therefore, the local
temperature difference between the fluids and rock as
well as surface area does not appear in the governing
energy balance equation coupled with the flow equation
(see CMG-GEM 2012, Appendix F). The local tempera-
ture equilibrium assumption can be justified considering
large ([0.2) Biot number for other applications of heat
transfer coupled with flow in porous media (Satik and
Yortsos 1996; Javadpour et al. 2003). Thanks to smaller
heat adsorption by rock grains, temperature signal may
cover a larger area in response to increasing the porosity.
It follows that, in single-phase flow, these competing
processes may cancel each other out (Han et al. 2012;
Fig. 1 Temperature change (from baseline) at location of leaky well
(and injection well) versus time and depth. Temperature in the
injection zone does not vary with time as it is in immediate contact
with (and therefore determined by) the injection-well temperature
Environ Earth Sci
123
Fig. 2 Sensitivity of temperature-pulse extent in AZMI for various parameters of a leak permeabilities, b AZMI porosity, c horizontal and
vertical permeabilities, d injected CO2 temperature, e injection rate, f rock heat capacity, g geothermal gradient, and h AZMI thickness
Environ Earth Sci
123
Oldenburg and Pruess 1998). In the two-phase CO2-brine
system, increased porosity will further contribute to
reducing the thermal signature at the CO2 front by
reducing heat conductivity of the system. The area cov-
ered by the temperature pulse (Fig. 2b) for porosities of
0.1 and 0.3 suggests that the temperature pulse is almost
insensitive to porosity. However, the pulse covers a
slightly smaller area for a higher porosity, showing that
advective force and heat conductivity influence the extent
to which the temperature pulse can travel.
AZMI permeability (kv and kh)
Figure 2c shows the effect of increasing AZMI perme-
ability on the temperature-pulse extent. Both horizontal
and vertical permeabilities are increased by factor of 10.
The temperature-pulse area is decreased at early time but
increased at late-time period. In general, higher perme-
ability leads to a lower pressure gradient, which reduces
the JT cooling effect. Because pressure will diffuse to a
larger area, the temperature pulse travels a longer distance.
After a sufficiently long time, a larger area should be
covered by the temperature pulse. However, at earlier times
the temperature response will be too small to be detected.
The temperature pulse therefore covers a smaller area for
the higher permeability case at early time. Decreasing the
temperature-detection limit from 0.15 to 0.05 �C shortens
this early time period from 500 to 460 days (results not
shown for brevity).
Injected CO2 temperature
Injected CO2 may be at different temperature than the
brine-saturated injection zone. Non-isothermal modeling of
the wellbore is required to determine the temperature of
CO2 at the sandface. Lu and Connell (2008) and Han et al.
(2010) studied the non-isothermal behavior of CO2 injec-
tion wells by solving coupled heat, mass and momentum
equations using various fluid and thermodynamic proper-
ties. They have shown that CO2 generally undergoes
cooling followed by heating as the depth increases, a
process which has been also observed in the field as
reported for Ketzin project (Liebscher et al. 2013). As a
result the sandface temperature at the injection well may be
warmer or cooler than that of the injection zone. Injection
of cooler CO2 makes leaking CO2 even cooler, leading to a
stronger temperature signal, and, as a result, the area
covered by the temperature pulse will be larger. The effect
of changing injected CO2 temperature on the temperature-
pulse radius is shown in Fig. 2d. After 2 years of injection,
the temperature signal covers 30 m, for an injection tem-
perature of 38 �C, as compared with 13 m, for a 41.7 �C
injection temperature.
Injection rate
Increasing the (sandface) injection rate leads to an
increased leakage rate and faster advection of the thermal
signal. Therefore, the temperature pulse will cover a larger
area in response to injection at a higher rate. Figure 2e
shows that doubling the injection rate increases the radius
of the temperature pulse by 38 %.
Heat capacity of the rock
Increasing heat capacity of the rock implies that more heat
is required to increase its temperature, making the tem-
perature signal weaker and the area covered by the tem-
perature pulse smaller. Figure 2f shows the temperature-
pulse radius for various values of a rock’s heat capacity.
Increasing the rock capacity [from an initial value of
1,000 J/(kg K) given by Han et al. 2012] by 100 J/(kg K)
reduces the radius of the temperature pulse by only
*0.2 m. Decreasing the heat capacity by 100 J/(kg K)
increases the radius by 0.2 m as well.
Temperature gradient
The temperature-pulse extent is compared with that when
the thermal gradient is neglected in Fig. 2g. Based on this
figure, the thermal pulse weakens when the geothermal
gradient is considered. This is mainly because when geo-
thermal gradient is neglected, the AZMI is warmer and
leakage-related cooling is more pronounced. Thus, a case
in which the cooling due to JT effect just counterbalances
the lower temperature in the overlying aquifer due to the
geothermal gradient is possible. In such a case, leakage-
induced temperature signal will be negligible.
AZMI thickness
Thicker AZMI accommodates more vertical than lateral
flow of the leakage fluid while allowing for higher leakage
rate owing to its increased capacity. Thus, radial extent of
the temperature pulse will be reduced at the beginning due
to decreased lateral flow. But, after some time the increased
leakage rate will generate a larger thermal pulse. This is
shown in Fig. 2h where temperature pulse extent is com-
pared to a case with AZMI thickness of 20 m. At the end of
the injection period, the pulse extent is increased by 10 %
when doubling the thickness of AZMI.
Not-co-located leaky-well scenario
Because the leaky well is unlikely to be co-located with the
injection well, investigation of the temperature signal to
Environ Earth Sci
123
leakage through a well at a distance from the injection well
is important. Both CO2 and brine may be leaking in this
situation, making the thermal response more complex. To
investigate this problem, the co-located leaky-well exam-
ple above was modified by considering a 100-m distance
between the injection well and the leaky well (see Table 1
for case properties). Cartesian gridding is used for this
configuration, although it makes numerical calculations
more costly. To reduce the computational cost, the system
is made symmetric and only a quadrant of the system is
modeled. The system involves two identical leaky wells at
equal distance from the injection well. The quadrant model
then includes one leaky well (half the size of the leaky well
in the full system) and one injection well injecting
12.5 kton/year, or *17,300 sm3/day, which is � of the
central injection well of the full system. The same mesh
size is used in both x and y directions, covering the first
100 9 100 m2 around the well with a 2 9 2 m2 grid and
increasing it gradually beyond this area to ensure infinite
outer boundaries.
The HYBRYD refinement option is used to model the
cylindrical shape of the leaky well in Cartesian mesh. This
option allows refinement of a Cartesian grid block into a
cylindrical grid within the center plus an outer block,
whose outer boundary fits the shape of the parent grid.
Using 2 9 2 m2 for the parent grid of the leaky well, the
radius of the leaky well will be 0.37 m. The vertical
gridding and all other reservoir and leak properties are
identical to those of the co-located leaky-well case. The
total number of grid cells is 124,215 (91 9 91 9 15). A
thermal gradient of 0.03 �C/m is used.
The temperature variation at the base of AZMI is shown
in the Fig. 3. Note that temperature increases slightly at the
beginning owing to leakage of warmer brine from the
injection zone to the AZMI. After *6 months of injection,
CO2 reaches the leaky well, and the leakage fluid is no
longer a single-phase brine but a mix of CO2-rich and
aqueous phases (see Fig. 3 for aqueous- and CO2-rich-
phase leakage velocities). As a result of leakage and
expansion of the CO2-rich phase, *5 �C of cooling will be
observed in the AZMI. The pressure reduces from
10.26 MPa at the upper layer of injection zone to 8.36 MPa
at the lower layer of the AZMI at the end of the injection
period. The leakage rate at this time is 0.95 m3/day which
is 1.7 % of the injection rate at reservoir conditions
(55.8 m3/day). Temperature change (T–Ti) along the leaky-
well path is shown in Fig. 4 versus time and depth. In the
injection zone, slight cooling is observed at the base and
top of the reservoir, the latter of which is due to heat
exchange with the cooling seal, and the former corresponds
to leakage of warmer brine from the injection zone to the
AZMI. Although thermal cooling in the leaky well occurs
throughout the leakage pathway in the seal and the AZMI,
highest cooling within the AZMI occurs at its base of
owing to maximal CO2 expansion. The temperature change
in the AZMI (at the location of the leak) becomes detect-
able 1 day after start of CO2 leakage. The pressure
reduction from lower layer of the AZMI to its upper layer
is only *0.1 MPa compared to 1.9 MPa from the upper
layer of the injection zone to the lower layer of the AZMI
at the end of the injection period (Fig. 5, right panel). The
extent to which the CO2-rich phase may expand depends
on pressure decline and, thus, Darcy velocity of the leak.
Everything else being equal, the pressure drop (and phase
velocities) is larger if the leakage conduit is smaller. The
temperature variation along the leaky well is evaluated
considering various well radii at the end of injection period
(Fig. 5, left panel). As the size of the conduit increases, the
velocity of the CO2-rich phase decreases and CO2 goes
through less pressure reduction, resulting in less cooling.
Temperature response correlates well with CO2-rich-phase
velocity (Fig. 5). The higher the velocity, the more pro-
nounced the thermal cooling. The velocity of the fluids
decreases toward the top of the AZMI because of the upper
closed boundary. Note that, because of the pressure decline
along the leakage path, CO2 solubility also decreases,
leading to CO2 exsolution from the aqueous phase, which
may cause cooling effect if the pressure drop is significant
and/or the aqueous-phase leakage rate is high. Given the
conditions of this scenario, however, the change in aque-
ous-phase CO2 concentration upon leakage is negligible, as
is CO2 exsolution.
Leaky fault scenario
Running a temperature log or DTS system along wells
believed to be leaking is a conventional use of thermal
monitoring, and it has been shown to be effective in leak
0
0.5
1
1.5
2
2.5
34
35
36
37
38
39
40
0 0.5 1 1.5 2
Vel
oci
ty (
m/d
ay)
Tem
per
atu
re (
C)
Time (yr)
Temperature
Gaseous phase velocity
Aqueous phase velocity
Fig. 3 Temperature change at base of AZMI in response to mixed
leakage of brine and CO2. Leakage rates of aqueous and CO2-rich
phases through leaky well also shown
Environ Earth Sci
123
detection under certain conditions. However, in spite of
possibly large temperature perturbations at the location of
the leaky well (as shown earlier), it may be difficult to
detect well leakage based on temperature data at observa-
tion wells distant from the leaky well. Another setting to be
assessed is a fault providing a planar pathway for leakage.
Locations of faults are generally known and determined
based on a combination of geological, drilling, and seismic
data. Geometries in which potential observation wells
intersect faults include vertical wells crossing dipping faults
or non-vertical wells emplaced to intersect faults. Leakage
along the fault may cause a detectable temperature signal.
An observation well intersecting the fault may then provide
a temperature anomaly that can be translated to leakage.
In the following section, the temperature response caused
by fault leakage is investigated. The not-co-located leaky-
well scenario was modified to account for thermal effects of
fault leakage. All properties of the injection zone, AZMI,
and sealing interval, as well as injection rate and mesh size,
are identical to those of the not-co-located leaky-well case.
The system includes two identical and parallel leaky faults
at equal distances (100 m) from the injection well. Again,
symmetry allows for modeling using a quadrant. The 2-m-
wide fault zone, is laterally conductive, and is assumed to be
vertical. Fault permeability is 10 mD in all directions (see
Table 1 for detailed specifications).
Temperature variation versus depth at the base of the
AZMI at cell nx = 51 and ny = 1 is shown in Fig. 6 (left
panel). Injection well is located at cell nx = 1 and ny = 1.
Brine leakage before CO2 breakthrough causes a larger
thermal signal than it does in the not-co-located leaky-well
case. However, the temperature rise caused by brine leak-
age remains insignificant. In addition, unlike the not-co-
located leaky well, in which continuous temperature
Fig. 4 Variation of temperature change with time and depth along
leaky well. Note that unlike the co-located leaky well case, the
temperature in the injection zone slightly varies with time
Fig. 5 Variation of temperature (solid curves on left panel), CO2-rich-phase velocity (dashed curves on left panel), and pressure (dotted curves
on right panel) along leaky well at end of injection period (2 years) considering various leaky-well radii
Environ Earth Sci
123
decline was observed after CO2 breakthrough, the tem-
perature does not monotonically decrease. This is due to a
quick pressure buildup in the AZMI, which quickly reduces
the pressure drop over the leakage path (and therefore
reduces CO2-rich-phase velocity, as shown in Fig. 6). The
pressure drop from the upper layer of the injection zone to
lower layer of the AZMI is shown in Figure 10 (ESM only)
for both leaky fault and not-co-located leaky well scenar-
ios. The pressure drop for the leaky fault scenario reaches a
maximum after 5 month of injection (*1 month before
CO2 breakthrough at the fault) after which it declines
quickly toward the initial pressure difference between
AZMI and injection zone (*0.1 MPa). The pressure in the
AZMI starts declining after CO2 breakthrough but it
remains far from the initial difference for the not-co-
located leaky-well case. Note that in spite of the reduced
pressure drop for the leaky-well scenario, the CO2 leakage
velocity keep increasing (Fig. 3) due to increased relative
permeability of CO2-rich phase in consequence of its
increased saturation in AZMI.
The reduced pressure drop leads to less JT cooling and
non-monotonic temperature change after CO2 breakthrough.
This behavior is also shown in Fig. 7, in which a temperature
decline from *0.5 to 0.75 year followed by increase in
temperature is observed in the AZMI. Before CO2 break-
through, temperature in the AZMI increases as a result of
warmer brine leakage. In the uppermost layer of the AZMI,
temperature increased by 0.16 �C after 6 months, when CO2
leakage to the AZMI began. The thermal signal induced in
the AZMI by brine leakage exceeds the temperature
threshold (0.15 �C) 160 days after start of brine leakage,
while induced cooling due to CO2 leakage becomes
detectable 55 days after start of CO2 leakage. CO2 leakage
rate after 2 years of injection reaches 9.4 m3/day (*16.4 %
of injection rate at reservoir conditions).
The sensitivity of the temperature response to fault-zone
width is examined in Fig. 8. The fault-zone width is varied
from 1 to 3 m. JT cooling is more pronounced in the 1-m-
thick fault zone due mainly to larger pressure drops cor-
responding to higher CO2-rich-phase velocities.
Discussion
The thermal signature was modeled by considering some
idealizations and simplifications. It was assumed that both
leaky well and fault are homogeneous. However, assuming
that the fault is homogeneously permeable for its entire
length may be an oversimplification of the fault structure,
although it may be considered an end-member fault structure.
As shown earlier, such a structure may not generate a sig-
nificant temperature change. In the field, fault internal
structure is observed to be highly heterogeneous and is dif-
ficult to access. Spatial variations in porosity and perme-
ability and related non-uniform architecture are needed to
account for fault heterogeneity. One way to represent this
heterogeneity may be using multiple leaky wells contained in
the fault plane. Representing the fault zone using discrete
leaky wells will result in similar thermal behavior, and a large
cooling signal may be observed along the fault as a result.
As leaking CO2 continues to rise and expand toward
shallower depths, it may convert from supercritical fluid to
liquid phase (Pruess 2011). In this study, phase change was
not considered which occurs when there is a large interval
between injection formation and shallow zones. Such phase
change will be associated with further cooling similar to
that experienced from movement of natural gas in wells
and pipelines. As shown by Pruess (2011), liquid CO2
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
39.7
39.8
39.9
40
40.1
0 0.5 1 1.5 2
Vel
oci
ty (
m/d
ay)
Tem
per
atu
re (
C)
Time (yr)
Temperature
Gaseous phase velocity
Aqueous phase velocity
Fig. 6 Temperature variation at base of AZMI at the location of the
leak, cell nx = 51 and ny = 1. Injection well is located at cell nx = 1
and ny = 1
Fig. 7 Variation of temperature change with time and depth along
vertical well completed at fault center (nx = 51, ny = 1) throughout
all layers. Injection well is located at nx = 1, ny = 1
Environ Earth Sci
123
starts boiling as it goes through additional pressure decline,
leading to significant temperature reduction. Expansion of
the gas phase as it enters shallower depths causes further
cooling. Three-phase (aqueous, and liquid and gaseous
CO2-rich phases) flow may evolve consequently. Such a
three-phase flow is slower (owing to relative permeability
of the phases), allowing for stronger heat exchange with the
wall rocks. As a result, liquid CO2 heats up and may start
boiling, leading to increased CO2 flow rates again. Within
the three-phase zone, liquid and gas saturations, and
therefore CO2 leakage rates, may show strong fluctuations
(Pruess 2011). Also, creation of CO2 bubbles could sup-
press leakage rate because it increases viscosity of the brine
(Wang and Clarens 2012). The combined effects of these
processes should be studied to determine the temperature
signal along the leakage pathway toward the surface.
Impact of geochemical reactions was neglected in this
study. As discussed by Han et al. (2012), geochemical
interactions between CO2, rock, and native brine may
cause a visible thermal signature. Further research to
account for the thermal response associated with geo-
chemical interactions should be pursued. The analysis in
this study was limited to the injection period. Further
investigation is required to examine temperature variations
during the post-injection period, when CO2 and brine
backflow from the AZMI may occur.
Conclusions
The thermal response to CO2 injection was simulated using
the CMG-GEM simulator. Simulating a benchmark prob-
lem, the thermal results of which have been reported on the
basis of TOUGH2-ECO2N, it was found that temperature
responses of TOUGH2 and CMG-GEM have similar fea-
tures, providing cooling close to the well bore and heating
toward the CO2 front. However, the magnitude of the
temperature response is different in spite of closely
matching PVT properties used in CMG-GEM and
TOUGH2. This difference is rooted in differently calcu-
lating the enthalpy in CMG-GEM versus TOUGH2. As a
result, the temperature response corresponding to mutual
dissolutions given by TOUGH2 is larger than that given by
CMG-GEM.
A scenario with a leaky injection well (co-located leaky-
well case) was defined so that temperature behavior in the
injection zone and the AZMI can be examined. At a given
time, moving away from the injection well into the injec-
tion zone, temperature drops owing to the JT effect, before
increasing to equilibrate with the unperturbed temperature
of the injection zone. Leaking CO2 undergoes a significant
pressure drop and JT cooling upon leakage into the AZMI.
The CO2 entering the AZMI may therefore be cooler than
the AZMI brine. As a result, temperature in the AZMI may
Fig. 8 Variation of temperature (solid curves on left panel), CO2-rich-phase velocity (dashed curves on left panel) and pressure (dotted curves
on right panel) along cell nx = 51, ny = 1 at end of injection period (2 years) considering various widths of fault zone
Environ Earth Sci
123
increase monotonically with radial distance from the
injection well. Sensitivity analysis in the co-located leaky
well case revealed that the area covered by the temperature
pulse (defined by a temperature change of [0.15 �C)
depends more on boundary conditions than on static pe-
trophysical parameters of the injection zone or the AZMI.
The area is highly sensitive to injection-well parameters,
including injection rate and injection temperature, leakage-
pathway parameters, such as leakage conduit permeability,
and the geothermal gradient.
The temperature signal was further investigated for a
leaky well some distance from the injection well (not-co-
located leaky-well case). Slight heating was observed ini-
tially owing to leakage of warmer brine from the injection
zone to the AZMI. After CO2 breakthrough, JT cooling
dominated the thermal signature, causing monotonic tem-
perature decline at the leaky well. Temperature variation
along the leaky well was evaluated by considering various
well radii, indicating that as the size of the leakage pathway
increased, pressure drop decreased and CO2 experienced
less expansion, resulting in less cooling. Replacement of
the leaky well by a 2-m-wide leaky fault resulted in less
cooling, although more heating due to brine leakage was
observed prior to CO2 breakthrough at the fault.
In summary, it was shown that the temperature pulse
caused by CO2 leakage is local and restricted to the
immediate vicinity of the leakage pathway. However,
leakage can result in large temperature changes along the
leakage pathway. Running temperature logs along suspi-
cious wells and/or wells intersecting suspicious faults may
then provide useful information on the leakage potential
and amount.
Compared to temperature, the pressure pulse covers a
considerably larger area proving to be more useful with
sparsely spaced sensors. Unlike the temperature signal
which depends on the leaking fluid velocity, the strength of
the pressure signal is determined by the leakage rate.
Increasing the leakage conduit cross-sectional area
increases the leakage rate and thus increases the pressure
change in the AZMI. However, it decreases the flow
velocity, and therefore, reduces temperature cooling and
signal. It was shown that AZMI temperature may increase
for brine leakage and decrease for leakage of CO2. How-
ever, either CO2 or brine leakage induces pressure increase
in the AZMI. In addition, pressure changes in the AZMI
may be induced by geomechanical effects and not related
to any leakage (Kim and Hosseini 2013). However, tem-
perature changes in AZMI are only due to leakage.
Acknowledgments This project was funded in part by EPA STAR
grant R834384. The authors wish to thank three anonymous reviewers
as well as Axel Liebscher of GFZ German Research Centre for
Geosciences for their careful reviews of the manuscript and sugges-
tions for improvements. The authors thank Computer Modelling
Group Ltd. for providing the option of free access to the CMG sim-
ulation package. Amin Badamchizadeh and Vijay Shrivastava are
acknowledged for their help in applying CMG-GEM thermal mod-
eling of CO2 storage. Thank you also to Lana Dieterich who edited
the manuscript. Publication authorized by the Director, Bureau of
Economic Geology.
References
Andre L, Azaroual M, Peysson Y, Bazin B (2011) Impact of porous
medium desiccation during anhydrous CO2 injection in deep
saline aquifers: up scaling from experimental results at labora-
tory scale to near-well region. Energy Proced 4:4442–4449.
doi:10.1016/j.egypro.2011.02.398
Andre L, Azaroual M, Menjoz A (2010) Numerical simulations of the
thermal impact of supercritical CO2 injection on chemical
reactivity in a carbonate saline reservoir. Transp Porous Media
82(1):247–274. doi:10.1007/s11242-009-9474-2
Bakker RJ (2003) Package FLUIDS 1. Computer programs for
analysis of fluid inclusion data and for modelling bulk fluid
properties. Chem Geol 194(1–3):3–23
Bielinski A, Kopp A, Schutt H, Class H (2008) Monitoring of CO2
plumes during storage in geological formations using tempera-
ture signals: numerical investigation. Int J Greenhouse Gas Cont
2(3):319–328. doi:10.1016/j.ijggc.2008.02.008
API Technical Data Book, Petroleum Refining (1983) 4th edn.
Washington
Canjar LN, Manning FS (1967) Thermodynamic properties and
reduced correlations for gases. Gulf Publishing Co., Houston
Carman PC (1956) Flow of gases through porous media. Butter-
worths, London 1
CMG-GEM (2012) CMG-GEM version 2012 user’s guide. Computer
modelling group Ltd. http://www.cmgl.ca/software/gem.htm
Han WS, Stillman GA, Lu M, Lu C, McPherson BJ, Park E (2010)
Evaluation of potential nonisothermal processes and heat
transport during CO2 sequestration. J Geophys Res 115(B7):
B07209. doi:10.1029/2009JB006745
Han W, Kim K-Y, Park E, McPherson B, Lee S-Y, Park M-H (2012)
Modeling of spatiotemporal thermal response to CO2 injection in
saline formations: interpretation for monitoring. Transp Porous
Media 93(3):381–399. doi:10.1007/s11242-012-9957-4
Harvey AH (1996) Semiempirical correlation for Henry’s constants
over large temperature ranges. AIChE J 42(5):1491–1494.
doi:10.1002/aic.690420531
Henninges J, Huenges E, Burkhardt H (2005) In situ thermal
conductivity of gas-hydrate-bearing sediments of the Mallik
5L-38 well. J Geophys Res-Sol Ea 110(B11)
Hoang H, Mahadevan J, Lopez H (2011) Interpretation of wellbore
temperatures measured using distributed temperature sensors during
hydraulic fracturing. In: SPE hydraulic fracturing technology
conference, SPE 140442, Woodlands, TX, 24–26 January 2011
Hovorka SD, Benson SM, Doughty C, Freifeld BM, Sakurai S, Daley
TM, Kharaka YK, Holtz MH, Trautz RC, Nance HS, Myer LR,
Knauss KG (2006) Measuring permanence of CO2 storage in
saline formations: the Frio experiment. Environ Geosci
13(2):105–121. doi:10.1306/eg.11210505011
Hovorka SD, Meckel TA, Trevino RH (2013) Monitoring a large-
volume injection at Cranfield, Mississippi—project design and
recommendations. Int J Greenhouse Gas Cont 18:345–360
Hurter S, Garnett A, Bielinski A, Kopp A (2007) Thermal signature of
free-phase CO2 in porous rocks: detectability of CO2 by
temperature logging. In: SPE Offshore Europe, SPE 109007,
Aberdeen, Scotland, UK, 4–7 September 2007
Environ Earth Sci
123
Javadpour F, Pooladi-Darvish M, Jeje A, Shen LC (2003) Effect of
transient temperature on MWD resistivity logs. Petrophysics
44:435–445
Kim S, Hosseini SA (2013) Above-zone pressure monitoring and
geomechanical analyses for a field-scale CO2 injection project in
Cranfield, MS. Greenhouse Gas Sci Technol. doi:10.1002/ghg.1388
Koschel D, Coxam JY, Rodier L, Majer V (2006) Enthalpy and
solubility data of CO2 in water and NaCl(aq) at conditions of
interest for geological sequestration. Fluid Phase Equilib
247(1–2):107–120. doi:10.1016/j.fluid.2006.06.006
Liebscher A, Moller F, Bannach A, Kohler S, Wiebach J, Schmidt-
Hattenberger C, Weiner M, Pretschner C, Ebert K, Zemke J
(2013) Injection operation and operational pressure–temperature
monitoring at the CO2 storage pilot site Ketzin, Germany—
design, results, recommendations. Int J Greenhouse Gas Control
15:163–173
Lu M, Connell LD (2008) Non-isothermal flow of carbon dioxide in
injection wells during geological storage. Int J Greenhouse Gas
Cont 2(2):248–258. doi:10.1016/S1750-5836(07)00114-4
Marinaro G, Etiope G, Gasparoni F, Furlan F, Bruni F (2011) Gas
seepage detection and monitoring at seafloor. In: 10th offshore
Mediterranean conference and exhibition, Doc. ID 2011-045,
Ravenna, Italy, 23–25 March 2011
Mathias SA, Gluyas JG, Oldenburg CM, Tsang CF (2010) Analytical
solution for Joule–Thomson cooling during CO2 geo-sequestra-
tion in depleted oil and gas reservoirs. Int J Greenhouse Gas
Cont 4(5):806–810. doi:10.1016/j.ijggc.2010.05.008
Nogues JP, Nordbotten JM, Celia MA (2011) Detecting leakage of
brine or CO2 through abandoned wells in a geological seques-
tration operation using pressure monitoring wells. Energy
Procedia 4:3620–3627
Nunez-Lopez V, Hovorka SD (2012) Subsurface monitoring of large-
scale CO2 injection at SECARBC’s phase III Cranfield Site. In:
carbon management technology conference, SPE 151504,
Orlando, Florida, USA, 7–9 February 2012
Oldenburg CM (2007) Joule–Thomson cooling due to CO2 injection
into natural gas reservoirs. Energy Convers Manage
48(6):1808–1815. doi:10.1016/j.enconman.2007.01.010
Oldenburg CM, Pruess K (1998) Layered thermohaline convection in
hypersaline geothermal systems. Transp Porous Media
33(1–2):29–63. doi:10.1023/A:1006579723284
Peng DY, Robinson DB (1976) A new two-constant equation of state.
Ind Eng Chem Fundam 15:59–64
Pruess K (2005) TOUGH2 fluid property module for mixtures of
water, NaCl, and CO2. Lawrence Berkeley National Laboratory,
Berkeley
Pruess K (2011) Integrated modeling of CO2 storage and leakage
scenarios including transitions between super- and subcritical
conditions, and phase change between liquid and gaseous CO2.
Greenhouse Gases Sci Tech 1(3):237–247. doi:10.1002/ghg.24
Pruess K, Oldenburg C, Moridis G (1999) TOUGH2 user’s guide,
version 2.0. Report LBNL-43134, Lawrence Berkeley National
Laboratory, Berkeley
Reinsch T, Henninges J, Asmundsson R (2013) Thermal, mechanical
and chemical influences on the performance of optical fibres for
distributed temperature sensing in a hot geothermal well.
Environ Earth Sci 70(8):3465–3480
Satik C, Yortsos YC (1996) A pore-network study of bubble growth
in porous media driven by heat transfer. J Heat Transfer
118(2):455–462. doi:10.1115/1.2825866
Saul A, Wagner W (1987) International equations for the saturated
properties of ordinary water substance. J Phys Chem Ref Data
16(4):893–901
Singh AK, Baumann G, Henninges J, Gorke UJ, Kolditz O (2012)
Numerical analysis of thermal effects during carbon dioxide
injection with enhanced gas recovery: a theoretical case study for
the Altmark gas field. Environ Earth Sci 67(2):497–509
Vinsome PKW, Westerveld J (1980) A simple method for predicting
cap and base rock heat-losses in thermal reservoir simulators.
J Can Petrol Technol 19(3):87–90
Wang S, Clarens AF (2012) The effects of CO2-brine rheology on
leakage processes in geologic carbon sequestration. Water
Resour Res 48(8):W08518. doi:10.1029/2011wr011220
Wangen M (1995) The blanketing effect in sedimentary basins. Basin
Res 7(4):283–298. doi:10.1111/j.1365-2117.1995.tb00118.x
Zeidouni M (2011) Analytical and inverse models for leakage
characterization of CO2 storage, Chapter 5, PhD dissertation,
University of Calgary, Calgary, Canada
Zeidouni M, Pooladi-Darvish M (2012) Leakage characterization
through above-zone pressure monitoring: 1—inversion
approach. J Petrol Sci Eng 98–99:95–106
Zeidouni M, Pooladi-Darvish M, Keith D (2009) Analytical solution
to evaluate salt precipitation during CO2 injection in saline
aquifers. Int J Greenhouse Gas Cont 3(5):600–611
Zhang YK (1993) The thermal blanketing effect of sediments on the
rate and amount of subsidence in sedimentary basins formed by
extension (vol 218, pg 297, 1993). Tectonophysics 225(4):551.
doi:10.1016/0040-1951(93)90317-D
Environ Earth Sci
123