International Journal of Greenhouse Gas Control 108 (2021) 103306

Contents lists available at ScienceDirect

International Journal of Greenhouse Gas Control

journal homepage: www.elsevier.com/locate/ijggc

Effect of geological heterogeneities on reservoir storage capacity and

migration of CO2 plume in a deep saline fractured carbonate aquifer
M. Adeel Sohal a, *, Yann Le Gallo b, Pascal Audigane c, J. Carlos de Dios d, Sean P. Rigby a
a
Department of Chemical and Environmental, Engineering, University of Nottingham, United Kingdom
b
Geogreen, 2 rue des Martinets, 92569, Rueil Malmaison, France
c
Bureau de Recherches Géologiques et Minières (BRGM), France
d
Fundación Ciudad de la Energía, Avenida del Presidente Rodríguez Zapatero, 24492, Cubillos del Sil, Spain

A R T I C L E I N F O A B S T R A C T

Keywords: In a reservoir characterization study of the Hontomín deep saline aquifer, the impact of geological heteroge­
CO2 injection neities on reservoir storage capacity and the migration of the CO2 plume is explored. This work presents, for the
Storage capacity first time, very long-term (up to 200 years) simulations of CO2 injection into the naturally fractured Sopeña
Geological heterogeneities
Formation, of the lower Jurassic age, at Hontomín. CO2 injection was simulated as a dual permeability case with
Injection pressure
Limestone
Eclipse compositional software. The matrix permeability of the carbonate reservoir is quite low (0.5 mD) and
Scenario-based modeling thus fluid flow through the fractures dominates. The reservoir is dissected by eight normal faults which limited
its south-east extension and divided it into several segments. The effect of geological heterogeneities was tested
through scenario-based modeling and variation of parameters characterizing heterogeneity within realistic limits
based on other similar formations. This modeling approach worked well in Hontomín where the database is
completely scarce. The plume migration, the reservoir storage capacity, and pressure, were each influenced in
diverse ways by incorporating particular types of heterogeneities. The effect of matrix heterogeneities on
reservoir storage capacity was substantial (by factors up to ~2.8×), compared to the plume migration. As the
reservoir matrix permeability heterogeneity increased, the reservoir storage capacity markedly decreased, whilst
an increase in porosity heterogeneity significantly increased it. The vertical gas migration in the homogeneous
base case was relatively larger compared to the heterogeneous cases, and gas accumulated underneath the
caprock via hydrodynamic trapping. It was also observed that, in heterogeneous cases, gas saturation in rock
layers from top to bottom was relatively high compared to the base case, for which most of the gas was stored in
the topmost layer. In contrast, the impact on storage capacity and plume movement of matrix vertical to hori­
zontal permeability ratio in the fractured carbonate reservoir was small. The impact of the transmissibility of
faults on reservoir pressure was only observed when the CO2 plume reached their vicinity.

a result of hydrodynamic trapping, solubility trapping, and mineral

1. Introduction trapping because of stratigraphic and structural heterogeneities (Bachu
et al., 1994; Doughty, 2010; Gaus, 2010; Gershenzon et al., 2015; Hesse
Greenhouse gas emissions to the atmosphere can be mitigated by et al., 2008; Iglauer et al., 2015; Iglauer et al., 2011; Kumar et al., 2005;
capturing and storing the carbon dioxide in subsurface geological for­ Lindeberg and WesselBerg, 1997; Metz et al., 2005; Naylor et al., 2011;
mations, in order to stabilize its atmospheric concentration. Typically, Pentland et al., 2011; Spycher et al., 2003; Xu et al., 2004). Different
CO2 is injected in a supercritical (scCO2) state and the minimum depth types of storage sites are available to store CO2, namely coal beds, saline
required to store it in this phase is approximately 800 m (Al-Khdheeawi aquifers, depleted hydrocarbon reservoirs, and geochemical trapping in
et al., 2017). Although it is stored as scCO2, it moves upward due to basalts. Saline aquifers have the largest storage capacity among the
buoyancy (density difference between formation brine and supercritical listed storage sites (Lackner, 2003). The main concern of CO2 seques­
phase) as described in the literature (Dai et al., 2014; Flett et al., 2007; tration in saline aquifers is to minimize the risk of leakage at a maximum
Hassanzadeh et al., 2009). During its upward and lateral movement in a injection rate/volume. To determine the characteristic extent of the
reservoir, CO2 is mainly trapped in three different ways, in particular as plume flow, the residence time of CO2 volume, and reservoir storage

* Corresponding author.
E-mail address: [email protected] (M.A. Sohal).

https://doi.org/10.1016/j.ijggc.2021.103306
Received 5 August 2020; Received in revised form 8 March 2021; Accepted 13 March 2021
Available online 31 March 2021
1750-5836/© 2021 Elsevier Ltd. All rights reserved.
M.A. Sohal et al. International Journal of Greenhouse Gas Control 108 (2021) 103306

Nomenclature Kfz Fracture permeability in z-direction

Km Matrix permeability
Bw Water formation volume factor (bbl/STB) Kmx Matrix permeability in X-direction
CV Coefficient of variation Kmy Matrix permeability in Y-direction
Cw Water isothermal compressibility (bar− 1) Kmz Matrix permeability in Z-direction
CO2 Carbon dioxide kPa Kilo Pascal
μw Water viscosity (cP) Kv/kh Vertical to horizontal matrix permeability ratio
Kh Horizontal permeability φf Fracture porosity
Kv Vertical permeability φm Matrix porosity
Kf Fracture permeability LGR Local grid refinement
Kfx Fracture permeability in x-direction TM Transmissibility
Kfy Fracture permeability in y-direction

capacity demands a detailed investigation of the subsurface reservoir. migration (spread) depends on reservoir storage capacity, and, there­
This might be available when CO2 is stored in depleted oil reservoirs but fore, it was not possible to ignore one and calculate the other. In the
it seldomly exists in saline aquifer cases. Therefore, scenario-based work described below, the reservoir pressure is plotted against the
modeling is the best substitute during the initial assessment of an injected cumulative mass of CO2 and time to compare the reservoir
aquifer storage site when the database is scarce, as is often the case when storage capacity at the same pressure in different heterogeneous cases.
CO2 is to be stored in relatively unexplored saline aquifers. In that sit­ In this simulation study, the effect of heterogeneities on the migra­
uation a deterministic field-scale reservoir model cannot be very accu­ tion of the CO2 plume and reservoir storage capacity was explored with
rate with respect to geological heterogeneities and fluid flow due to the Eclipse (E300) software using the dual permeability option. There is no
uncertainties associated with the initial input parameters. Thus, time-lapse seismic dataset available in the Hontomín case to validate the
scenario-based models taking expected site-specific geological hetero­ CO2 plume movement in different layers of the Sopeña Formation
geneities into account are useful as part of the reservoir characterization against the simulation model. Predicting the CO2 flow behavior in
and for planning the suitable injection and monitoring strategies. The naturally fractured carbonate reservoirs is quite challenging compared
motivation behind this work is to find the answer to the following to other lithologies especially when the database is completely scarce.
questions in the simulation study. Therefore, scenario-based modeling can reveal the best field develop­
ment strategy for long-term CO2 storage as well as the potential chal­
o How geological heterogeneities affect the reservoir storage capacity? lenges that could be envisaged in different possible geological settings. A
o How sensitive is the plume migration to geological heterogeneities? gradual increase in CV value to greater than one provided an insight into
the effect of the level of heterogeneity variation on plume migration and
The impact of different types of geological heterogeneities on the storage capacity. Moreover, the influence of the variation of individual
migration of the CO2 plume and reservoir storage capacity has mostly matrix and fracture permeabilities at the same rate of migration and
been investigated through conceptual reservoir models via parametric storage of CO2 were explored and compared.
studies. The coefficient of variation (CV) of heterogeneities in those
simulation models is always less than one (Ghanbari et al., 2006). The 1.1. Geological model description
CV is a statistical measure of the dispersion of data points in a data series
around the mean. It represents the ratio of the standard deviation to the The geological model of the Hontomín site is a structural dome and
mean, and it is a useful statistic for comparing the degree of variation includes under-burden, reservoir, seal, and overburden. The vertical
from one data series to another, even if the means are drastically cross-section of the whole geological complex is shown in Fig. 1, which
different from one another. The aforementioned models do not exhibit was constructed through the analysis and integration of borehole im­
the level of heterogeneities that could be expected in a real field case. ages, well logs, and well test data (Le Gallo and de Dios, 2018). The
Therefore, in this simulation study of the Hontomín deep saline aquifer, overburden comprises of the Dogger, Purbeck, and Weald Formations of
the coefficient of variation was gradually increased to greater than one. late Jurassic to lower Cretaceous age and is placed unconformably on
The Hontomín site is an onshore deep saline aquifer located close to top of the middle Jurassic marine rocks (Alcalde et al., 2014). The for­
Burgos in the north of Spain and is operated by Fundación Ciudad de la mations are mainly composed of clays, sandstone, and carbonate rocks,
Energía (CIUDEN, Cubillos del Sil, Spain). It was recognized by the and the depth of these formations changes along with the structure. The
European Parliament (Davies, 2013) as a key test facility for Carbon caprock is mainly composed of highly carbonated marls (close to 50 % of
Capture and Storage (CCS) technology development. The scCO2 is being its composition) and consists of the Pozazal and Marly Lias Formations
stored in a naturally fractured limestone and dolomite reservoir of lower of middle Jurassic age which are approximately 163 m thick at the well
Jurassic age (Alcalde et al., 2014). The geological model of the site was location. The lower Jurassic Sopeña formation is the reservoir and it is
built by Geogreen and updated by the integration of the latest data, such composed of fractured limestone in the upper part and dolomites in the
as the image log interpretations from newly-drilled injection and lower part. The reservoir limestone is located at a depth of 1435 m in the
observation wells (Le Gallo and de Dios, 2018). injection well (HI) and is 79 m thick, whereas the dolomite was found at
The focus of this article is on the effect of matrix heterogeneity in a a depth of 1514 m in HI and is 41 m thick. Therefore, the total reservoir
fractured system. This is because the interaction between the reservoir thickness is 120 m at the injection well location. The reservoir has an
storage capacity and the extent of plume migration in a tight reservoir increasing dip towards the north and north-east from the injection well,
depends upon the matrix heterogeneity rather than the fractures, which with its apex around the H2 legacy well (Le Gallo and de Dios, 2018).
act as conduits for fluid flow. Therefore, investigation of the fracture The reservoir was modeled with 39 layers, and the thicknesses of layers
density and distribution was not part of this study, although the flow vary from 1 to 10 m. The matrix permeability of the reservoir is
through the fractures dominates. One particular area of interest was approximately a half milli-Darcy in the geological model and the
whether matrix heterogeneity accentuates or mitigates bypassing of the porosity varies from 3 to 13 %. The Keuper formation of the Triassic age
matrix by the plume flow in fractures for a tight reservoir. The plume is the under-seal, located at a depth of 1555 m in the HI well, and is

2
M.A. Sohal et al. International Journal of Greenhouse Gas Control 108 (2021) 103306

Fig. 1. Vertical cross-section of the geological model of the Hontomín storage complex; The HA and HI newly-drilled wells are the observation & injection wells,
respectively, while well H1 to H4 are legacy wells (Le Gallo and de Dios, 2018). The legacy wells were used to confirm the geological sections & reservoir geometry.

mainly composed of anhydrite. A total of five geological facies (lime­ Lemonnier and Bourbiaux, 2010b). The dual-porosity model consists of
stone, dolomite, marls, shale, and anhydrite) was included in the two regions with distinct porosities and permeabilities, representing the
geological model. The permeability and porosity of different facies were matrix and fractures within the formation. The matrix constitutes the
assumed to be constant in the model and available from the hydraulic region with negligible flow capacity but significant pore volume that is
characterization tests (Le Gallo and de Dios, 2018; de Dios et al., 2017). providing the primary porosity to the reservoir system. The fracture
The permeability of marls, shale, and anhydrite is almost zero, whereas system provides the path for fluid flow from the formation to the well
limestone and dolomite matrix permeabilities are in the range of ~0.5 and does not contribute to storage. Several studies pointed out that the
mD (Le Gallo and de Dios, 2018). These values were determined during dual-porosity model may not be appropriate for interpreting the well
the hydraulic characterization campaigns conducted on-site (de Dios tests data from all fractured reservoirs (Egya et al., 2019). The Hontomín
et al., 2017). homogeneous model was initially history matched with
There are eight normal faults inferred at the top of the reservoir from water-alternating-gas (WAG) well test data and then used to simulate
the 3-D seismic interpretations, and are assumed to be vertical within long-term CO2 injection. During the long-term CO2 injection, the effect
the reservoir. Two main faults cross the storage complex from the of geological heterogeneities, vertical to horizontal matrix permeability
reservoir to the overburden, which limits the south-eastward extension ratio (kv/kh), and fault transmissibilities were explored. The dimensions
of the reservoir. These are the Ubierna fault, located at the southern part, of the Sopeña formation in the Hontomín geological model are 5.7 km
and the East fault located at the eastern part of the Hontomín site, as by 5.6 km with a thickness of 120 m. The horizontal matrix permeability
shown in Fig. 2. The boundary faults (Ubierna & East faults) are sup­ (kh) is 0.5 mD, the vertical matrix permeability (kv) was set to 0.05 mD,
posed to be sealing faults and the transmissivity of all other faults in the and the porosity varies from 3 to 13 % in the calibrated homogeneous
geological model is unknown. There are four legacy wells; H1, H2, H3, model. The model was considered initially in hydrostatic equilibrium,
and H4, located in the study area; these wells were commonly used to with an initial reservoir pressure of 141 bar at a True Vertical Depth
confirm the structure, geometry, and properties of the whole geological (TVD) of 1459 m. The initial water saturation (Sw) was 100 % and the
complex. However, the main source to gather data for the geological salinity of the formation water was 40,000 ppm. There was no dissolved
model was the newly drilled observation (HA) and injection (HI) wells. gas in the aqueous phase at the beginning of the simulation. The Hon­
Petrophysical properties are known only around the well drainage area. tomín geological model was uniformly gridded in the horizontal direc­
However, their properties are uncertain in the rest of the reservoir and tion with 57 cells in the X-direction, 56 in the Y-direction and 658 cells
could be varied to match the observed data in the history matching in the Z-direction with different thickness. The reservoir model details
process. are summarized in Table 1.

2. Modeling approach 2.1. Relative permeability and capillary pressure data

In this study, the effects of geological heterogeneities on reservoir In the literature, different CO2 brine relative permeability curves for
storage capacity and the migration of the CO2 plume were investigated carbonate rocks have been proposed, but the following set of drainage
during long term injection in the Sopeña Formation. Since the Sopeña relative permeability curves, measured in the Wabamun Limestone
Formation is highly fractured, it was simulated by using the dual Formation (Bennion and Bachu, 2008) were initially used in this simu­
permeability option within E300. The dual-permeability model assumes lation study. However, the relative permeabilities for matrix and frac­
that the matrix is permeable and allows the flow between matrix blocks, tures were further adjusted with the help of Corey’s model (Dandekar,
in addition to matrix-fracture and, a fracture-fracture flow which is 2013) for gas water system as given in Eqs. (1) and (2):
observed in a dual-porosity model (Lemonnier and Bourbiaux, 2010a;

3
M.A. Sohal et al. International Journal of Greenhouse Gas Control 108 (2021) 103306

Fig. 2. Location of wells and the extensions of all the eight faults within the Hontomín geological model in Petrel. F11 (South Fault) is located in the south, while the
F9 (East Fault) is located on the east side and controls the structural extension of the reservoir in the south-east direction.

[ ]ng
Sg − Sgr pressure (capillary pressure) for the majority of the rock pores is greater
krg = krg(wir) (1)
1 − Sgr − Swir than the CO2 injection pressure. Consequently, CO2 will not invade a
significant part of the reservoir rock during the injection period and
[ ]nw
Sw − Swir principally migrate through the fractures. The capillary pressure curve
krw = krw(gr) (2)
1 − Sgr − Swir was calculated by the Brooks and Corey model (Skjaeveland et al., 2000)
in Petrel by using the input parameters given in Table 2 and depicted in
where Swir is the irreducible water saturation, Sgr is residual CO2 satu­ Fig. 4.
ration, and Nw and Ng are Corey’s functions for water and CO2
respectively. The values of parameters to generate the Wabamun and the
current model drainage relative permeability curves are given in 2.2. Fluid properties
Table 2. In the initial model validation process, the same relative
permeability curves were used to model CO2 displacing water and water The Eclipse compositional (E300) software calculates the density of
displacing CO2 (i.e., no hysteresis in relative permeability). The impact CO2 by using a cubic equation of state which is accurately tuned to give
of relative permeability hysteresis on geological CO2 storage can be the density of the compressed gas phase by following the method given
found elsewhere (Skjaeveland et al., 2000). The final relative perme­ in Spycher and Pruess (Spycher and Pruess, 2010). A modified Peng
ability curves for matrix and fractures are shown in Fig. 3. Robinson equation of state is used, where the attraction parameter is
It was found, from the petrophysical lab results, that 38%–78% of the made temperature-dependent. The gas viscosity is calculated from the
pore throats within the Sopeña Formation are in the micropore scale (< method of Fenghour and Wakeham (Fenghour et al., 1998). The mean
2 nm) range according to the International Union of Pure and Applied pore brine density without any dissolved CO2 was 1002 kg/m3, based on
Chemistry (IUPAC) classification of pore size. Therefore, the pore entry a salt mass fraction of 0.038. The viscosity of brine was calculated using
the viscosity of water given in the International Association for the

4
M.A. Sohal et al. International Journal of Greenhouse Gas Control 108 (2021) 103306

Table 1
The details of the homogeneous simulation model of Hontomín deep saline
aquifer.
Parameters Values Units

Total grid cells 2,103,528

Reservoir length 5.7 km
Reservoir width 5.6 km
Reservoir thickness 0.5 m
Matrix horizontal permeability 202 mD
Matrix vertical permeability 0.05 mD
Matrix porosity 3− 13 %
Fracture permeability in x-direction 240 mD
Fracture permeability in y-direction 360 mD
Fracture permeability in z-direction 360 mD
Fracture porosity 0.01
Net to gross ratio 1
Sigma 1
Reservoir depth (TVD) 1459 m
Reservoir pressure 141 Bar Fig. 4. A capillary pressure curve that was used in the validated base case.
o
Reservoir temperature 42 C
Initial water saturation (Swi) 1 %
Residual water saturation 10 % μw = 4.2844 × 10− 5 + (0.157(t + 64.993)2 − 91.296)− 1
(5)
Water endpoint relperm 1
Residual gas saturation 0.05 % The mutual solubility of CO2 and H2O is calculated in E300 to match
Gas end point relperm 0.9 the experimental data for typical CO2 storage conditions for a temper­
Pore brine salinity 40,000 ppm
ature range between 12− 250 ◦ C and pressure up to 600 bars. They were
Components H2O, CO2
Pore brine density 1.002 g/cm3 calculated following the procedure given by Spycher and Pruess
FVF 1.01 Rm3/Sm3 (Spycher et al., 2003; Spycher and Pruess, 2010), based on fugacity
Pore brine viscosity 0.66 cP equilibration between water and a CO2 phase. Water fugacity was ob­
tained by Henry’s law, while CO2 fugacity was calculated using a
modified Redlich-Kwong equation of state.
Table 2
The details of the parameters which were used to generate the drainage relative 2.3. History matching and model calibration
permeability curves for Wabamun and the Hontomín Limestone.
Parameters Wabamun (drainage) Hontomín Hontomín is a low permeability and a highly fractured carbonate
Sgr 0.005 0.05
reservoir. Therefore, it was simulated by a dual permeability option in
Ng 5.60 2 E300. The dual permeability option in a reservoir simulator involves a
Krg@Swir 0.5289 0.9 highly interconnected set of fractures which are uniformly distributed
Swc/Swir 0.5950 0.1 throughout the reservoir. The simulator divides the whole reservoir into
Nw 1.40 1
an equal number of matrix and fracture grid blocks, unlike a real
Krw@Sgr 0.970 0.945
reservoir system. This model of an idealized fracture system was initially
introduced by Warren & Root (Warren and Root, 1963) and has been
commonly used since then. The Hontomín model was calibrated against
the well test brine and WAG data, and constructed as a homogeneous
model. The model validation process commenced with an assisted his­
tory matching method utilising a Petrel uncertainty and optimization
workflow, and, later on, the best match was manually tuned. In the
validation process, capillary pressure and local grid refinement (LGR)
around the well were tuned otherwise it was not possible to get a good
match to the observed data. In the history matching process, more than
850 realizations were simulated with a reasonable range of uncertain
parameters. Initially, calculation of the fracture permeability and
porosity for fractured carbonate reservoirs was achieved by employing
the method given by Nelson (2001). However, due to the many un­
known input parameters, each was then treated as a free-fitting
parameter in the history matching process. Similarly, fluid compress­
ibility and formation volume factor were calculated by the method
described by Ahmed (2018). The impact of each uncertain reservoir
parameter on the bottom-hole pressure was thoroughly investigated in
Fig. 3. The relative permeability curves for matrix and fractures that were used the process of getting a good final match to the history data. The match
in the history matching process. of reservoir bottom-hole pressure achieved by the simulated data for all
the well test campaigns is depicted in Fig. 5.
Properties of Water and Steam (IAPWS), 2008 (Cooper and Dooley,
2008) with the help of Eqs. 3 and 4 as described by Sharqawy et al. 3. Results and discussion
(Sharqawy et al., 2010):
( ) 3.1. Simulation results for homogeneous model
μsw = μsw 1 + AS + BS2
A = 1.541 + 1.998 × 10 t − 9.52 × 10− 5 t2
− 2 (4) In the homogeneous Hontomín simulation model, the best match
B = 7.974 − 7.561 × 10− 2 t + 4.724 × 10− 4 t2 during the history matching process to the observed well test WAG data

5
M.A. Sohal et al. International Journal of Greenhouse Gas Control 108 (2021) 103306

Fig. 5. History matching of the simulated bottom-hole pressure for the Hontomín model with brine and WAG well test observed data respectively for 2017.

was achieved using a matrix permeability (km) of 3mD for limestone and gradually declined to 15,700 kPa until April 2019, and then, again,
dolomite facies and with km almost constant throughout the Sopeña steadily increased to 15,900 kPa over the next 30 years injection time.
reservoir. Therefore, all the layers of the Sopeña reservoir were equally This pressure behavior observed shows the greater ease of the gas flow
likely for CO2 migration. Fracture permeabilities within the Sopeña with time in a homogeneous reservoir, which could be associated with
reservoir were set 1.5 times higher in the E–W direction compared to the increasing gas relative permeability due to an increase in gas saturation.
N–S direction. It has been identified, by Le Gallo and de Dios (2018) The historical data for the WAG process indicated it started with water
from the Tele-viewer log interpretation, that the number of fractures in injection possibly to make sure the gas would flow away from the well
the E–W direction (278) is greater than the number of fractures in the into the reservoir. The validated homogeneous simulation model was
N–S direction (48). In the well test historical data, slugs of brine and CO2 used as a reference case (denoted below as the ‘base case’). against
were injected into the Sopeña Formation, and the average injected which all other simulated cases were compared to determine the impact
volume of brine and gas in each slug was approximately 21 and 28,333 of the changes in a sensitivity study.
sm3 for brine and CO2 respectively. Slugs of brine and CO2 were
continued for two days in WAG injection fashion and ceased for three 3.2. Simulation results of heterogeneous models
days before the start of the next slug. The plot of water and CO2
WAG-type slug injection pressure for the initial three years is shown in Several heterogeneous models were simulated by altering the matrix
Fig. 6 and a small-scale plot of the pressure versus injection rate for gas permeability and porosity in order to investigate the impact of these
slugs is depicted in Fig. 7 to show the slug injection pattern. The changes on CO2 plume migration, reservoir storage capacity, and pres­
reservoir shut-in pressure rose from 14,300 to 14,900 kPa during the sure. The heterogeneous models were created using a sequential
brine and gas slug injections over 30 years. However, the injection Gaussian simulation method (Journel and Huijbregts, 1978) and ranked
pressure rose to 16,300 kPa at the beginning (July 2017) and then based on the coefficients of variation of the porosity and permeability.

Fig. 6. Build-up and shut-in reservoir pressures during water alternating gas slug injections in the homogeneous base case. The boxed part of the plot is shown on a
smaller scale in Fig. 7.

6
M.A. Sohal et al. International Journal of Greenhouse Gas Control 108 (2021) 103306

Fig. 7. Small scale display of CO2 slug injection rate and pressure during the WAG process in the homogeneous base case.

As mentioned above, the base model for these heterogeneous cases is the produce the heterogeneous cases. This variation is within the range of
validated homogeneous model, known as the base case. It served as a the permeability distribution which was observed in the Turonian –
reference against which to compare the changes resulting from the Campanian Kometan Formation (Rashid et al., 2015), which is a similar
various heterogeneous models. There are six reservoir zones in the tight limestone reservoir located in the north of Iraq. The heterogeneous
geological model and each reservoir zone has a different minimum, cases thus produced are in a realistic range of permeability heteroge­
maximum, mean, and standard deviation value for matrix porosity and neity and tested to observe the minimum and maximum reservoir
permeability. The permeability was log-normally distributed, while the response which could be anticipated during and after the injection
porosity was normally distributed, amongst the various reservoir zones. campaigns. It will be seen the heterogeneity affected the migration of
In all the heterogeneous cases the mean values were close to the base CO2 plume reservoir storage and pressure differently.
case, but the standard deviation was different. The level of heterogeneity
amongst the models was gradually enhanced by increasing the standard 3.2.1.1. Reservoir pressure and heterogeneity. It was observed that the
deviation of the given parameter, which meant the spread of the prop­ reservoir shut-in pressure increased as the level of heterogeneity
erties around the mean value increased. It is noted that, in Petrel, it is not increased. However, the rise in pressure with respect to the homoge­
possible to produce heterogeneous models with identical mean values by neous base-case was small at the given injection rate over 50 years. It
changing only the standard deviation, and minimum and maximum for can be seen in Fig. 9 that the shut-in pressure continuously increased
the range of the properties. Therefore, the mean value of heterogeneous with time as long as gas was being injected. However, the pressure rise,
cases slightly varies from the base case, as well as from each other. in the beginning, was quite sharp but declined with time. The sharp
pressure gradient, in the beginning, could be due to the slow mobility of
3.2.1. Heterogeneity based on permeability CO2 which was improved as the gas saturation increased with time. The
A comparison of the matrix permeability for all the six zones of a mobility of the gas phase increased as its relative permeability increased
typical heterogeneous case (base_13) with those of the homogeneous which depended on gas saturation in the porous media. The pressure
base case is given in Table 3 and displayed in Fig. 8. The degree of response depended on the matrix permeability distribution in each layer
heterogeneity is characterized by the coefficient of variation. Therefore, of a reservoir zone and it increased as the permeability heterogeneity
the degree of heterogeneity of geological models, based on the coeffi­ increased. The build-up pressure trend in the cell containing the injec­
cient of variation for matrix permeability, relative to the base case is tion well is depicted in Fig. 10 and it illustrates that the pressure is
given in Table 4. The typical reservoir matrix permeability is 0.5mD, inversely related to the permeability distribution in the same grid cell.
and, therefore, it has been varied in a range between 0.01 to <10mD to While the change in pressure response is relatively low in absolute terms
in this dual permeability case because the permeability difference
among the layers is small, nonetheless it still gives an insight into a
Table 3
pressure variation trend that may well be non-trivial in a single
The permeability distribution range, standard deviation, and coefficient of
permeability case depending on permeability variation among the
variation for each zone of the reservoir for the given two cases.
layers.
Cases Reservoir Zones kmin kmax kmean Stdv CV

Upper calcareous 0.0 3.0 2.9 0.4710 0.16 3.2.1.2. Reservoir storage capacity and heterogeneity. The impact of
Upper baffle calcareous 0.0 3.0 2.0 1.4171 0.71
geological heterogeneities on reservoir storage capacity was investi­
Middle calcareous 3.0 3.6 3.1 0.2548 0.08
BASE CASE
Lower baffle calcareous 3.0 3.6 3.3 0.2961 0.09 gated by injecting CO2 at a constant rate instead of constant pressure.
Lower calcareous 3.0 3.6 3.0 0.1213 0.04 Therefore, the reservoir storage capacity was derived from a change in
Dolomitic 0.0 0.6 0.6 0.0622 0.10 reservoir pressure by injecting a similar volume of gas over the same
Upper calcareous 0.04 9.96 2.4 2.0606 0.85 period of time into different heterogeneous models. The reservoir shut-
Upper baffle calcareous 0.02 9.99 2.0 2.6706 1.34
Middle calcareous 0.08 9.86 2.3 1.9921 0.89
in pressure versus cumulative CO2 injection for different simulated
BASE_13 cases, as compared to the base case, is shown in Fig. 11. It is obvious,
Lower baffle calcareous 0.20 9.22 2.0 1.4631 0.72
Lower calcareous 0.9 9.97 3.0 2.4277 0.81 from the comparison of these various cases with the base case, that, as
Dolomitic 0.01 10.0 0.4 1.0018 2.45 the reservoir matrix permeability heterogeneity increased, the storage

7
M.A. Sohal et al. International Journal of Greenhouse Gas Control 108 (2021) 103306

Fig. 8. Comparison of matrix permeability distribution in X-direction for the homogeneous base case with that for the heterogeneous case (base_13).

Table 4
The coefficient of variation of matrix permeability for the set of simulated heterogeneous cases.
Cases Base case Base_1 Base_2 Base_3 Base_4 Base_5 Base_6 Base_7 Base_8

CV<1 0.46 0.77 0.81 0.85 0.88 0.91 0.94 0.97 0.99
Cases Base_9 Base_10 Base_11 Base_12 Base_13 Base_14 Base_15 Base_16 Base_17
CV>1 1.02 1.04 1.07 1.09 1.11 1.19 1.26 1.32 1.38

Fig. 9. The variation of reservoir shut-in pressure, over 50 years of gas injection, with increased heterogeneity in matrix permeability in the horizontal direction.

capacity markedly decreased. The comparison of base case with base_17 migration, in the horizontal and vertical directions away from the in­
at reservoir shut-in pressure of 14,826 kPa showed that the reservoir jection well, for base case and base_16, is portrayed in Fig. 12. The
storage capacity decreased approximately 2.8 times for this highly spread of the CO2 plume in the topmost zone of the reservoir slightly
heterogeneous case. In the dual permeability simulated model of Hon­ varied between these two cases although the heterogeneity difference
tomín, heterogeneity based on matrix permeability adversely affected based on matrix permeability was quite significant. This could be a
the reservoir storage capacity. In carbonate reservoirs, an increase in consequence of dual permeability modelling in which flow took place
permeability is associated with pore connectivity as observed by He between fracture and matrix as well as between matrix and matrix.
et al. (2014). Therefore, the plume relatively migrated more in the south-west direc­
tion in base_16 compared to the base case. The vertical gas migration in
3.2.1.3. Migration of CO2 plume and heterogeneity. The CO2 plume a homogeneous case is relatively larger compared to the heterogeneous

8
M.A. Sohal et al. International Journal of Greenhouse Gas Control 108 (2021) 103306

Fig. 10. The relationship between permeability and pressure build-up for each layer of the reservoir as it was recorded in the grid cells that were penetrated by the
injection well. The pressure build-up is inversely proportional to the matrix permeability in this case.

Fig. 11. Reservoir shut-in pressure versus the cumulative injected mass of CO2 for a similar time period. The heterogeneity of the simulated models based on the
variation in matrix permeability.

case and gas is accumulating underneath the caprock which is known as porosity in transport modelling, since it could induce irregular plume
hydrodynamic trapping. For the heterogeneous case, gas saturation in distributions, delay and spread solute breakthrough curves, and increase
rock layers from top to bottom is relatively high compared to the base plume second moments as observed previously (Tian et al., 2014; Hu
case for which most of the gas is stored in the topmost layer. The hori­ et al., 2009; Pan et al., 2009; Pan et al., 2011). Therefore, porosity
zontal plume migration among the layers also depends on the part of the heterogeneity must be incorporated into reactive transport simulations
reservoir perforated for the injection well for gas injection, which was to more accurately evaluate CO2–brine–rock interactions.
the entire reservoir zone in this case. Thus, each zone of the reservoir in The degree of heterogeneity of geological models based upon matrix
all heterogeneous cases was equally open for gas injection like the base porosity relative to the base case is described by the coefficient of
case. variation and values are given in Table 5. The method to produce the
heterogeneous models based on porosity heterogeneity is similar to that
3.2.2. Heterogeneity based on porosity used for permeability. The heterogeneity based on porosity affected the
In most previous studies (Chen et al., 2013; Rapaka et al., 2008; Xu migration of the CO2 plume and reservoir storage capacity differently
et al., 2006; Farajzadeh et al., 2011; Lengler et al., 2010; Tian et al., compared to the heterogeneity based on matrix permeability.
2014), the assigned porosity is taken as uniform since it is a common The plume migration away from the injection well was significantly
assumption that the heterogeneity for flow simulations may be man­ different between the homogeneous base case and the heterogeneous
ifested entirely in the hydraulic conductivity field, and that variations in case (base_72) as can be seen in Fig. 13. An equal amount of CO2 was
porosity have negligible effects on fluid flow. However, we contend that injected in both cases over 50 years, and at a similar injection rate. The
it is also very important to characterize and incorporate heterogeneity of heterogeneous case base_72 increased the reservoir storage capacity by a

9
M.A. Sohal et al. International Journal of Greenhouse Gas Control 108 (2021) 103306

Fig. 12. Part (a) shows the migration of CO2 plume in base_16 and part (b) shows the migration of CO2 plume in the homogeneous base case over 50 years of
injection. The same volume of CO2 was injected in both cases but the migration of plume was different in the heterogeneous case because of the variation in matrix
permeability distribution.

relatively more convenient when fracture permeability was increased

Table 5
compared to the matrix at the same rate in this case. It was observed that
The coefficient of variation in matrix porosity for the set of simulated cases.
the reservoir shut-in pressure gradually increased by 16 kPa more in the
Cases ɸmin ɸmax ɸmean Stdv CV case when the matrix Y-direction permeability was doubled, compared
Base case 3% 13 % 7% 0.0094 0.13 to the corresponding case of increased fracture Y-direction permeability,
Base_72 1% 20 % 9% 0.0481 0.51 and stayed constant over 50 years of injection. The degree of kmy and kfy
Base_73 1% 20 % 10% 0.00542 0.52
change is identical but the shift in pressure is higher in the matrix case
Base_74 1% 20 % 9% 0.0497 0.55
Base_75 1% 20 % 10% 0.0569 0.58 which verifies that the fluid prefers to flow through fractures instead of
matrix. However, by doubling the matrix permeability of base case
reservoir storage capacity increased 3.85 times more than the original
factor of 1.5 times more than the base case at the shut-in pressure of capacity at a pressure of 14,850 kPa as can be seen in Fig. 16. While the
14,806 kPa as shown in Fig. 14. It was found that plume spread is pressure of the base case decreased 68 kPa on doubling the km over 50
inversely related to available pore volume as arising from the variation years of injection.
in the heterogeneous porosity distribution. Reservoir shut-in pressure
was 42.5 kPa higher in the base case compared to the heterogeneous
case over 50 years of injection. Initially, reservoir shut-in pressure was 3.4. Effect of kv/kh ratio
identical in both cases but it gradually diverged as more gas was injected
and the pressure difference was continuously increasing with time. The An increase in the matrix kv/kh ratio increased the rate of vertical
variation in reservoir storage capacity and plume migration just movement of gas during the injection phase. There might be a risk of gas
amongst the heterogeneous cases themselves is very small. This is leakage through the aquifer boundaries if kv/kh ratio is decreased
because the increase in coefficient of variation for matrix porosity is because the horizontal migration of CO2 may dominate in that condi­
small among the heterogeneous cases. tion. Fig. 17 shows the horizontal and vertical migration of the CO2
plume in the base case when vertical to horizontal permeability ratios
were decreased from 1 to 0.0001 for a similar period of time at the same
3.3. Effect of kf and km injection rate. However, when the kv/kh ratio was equal to 1 the gas
plume was mainly migrating towards the top of the structure and
When the degree of matrix and fracture permeability variation in the accumulating under the caprock. The reservoir shut-in pressure did not
horizontal direction was similar, the CO2 plume migrated the same significantly change when the matrix kv/kh ratio varied from 0.0001 to
distance from the injection well in both cases. However, the change in 1. The difference in reservoir pressure between the extreme cases was
gas saturation at the outer periphery of the plume was slightly different less than 15 kPa after 50 years of gas injection as illustrated in Fig. 18.
between the two cases. It was observed that the CO2 plume migration in The small pressure difference, as a result of a significant change in
each of the horizontal and vertical directions was almost identical when vertical permeability of matrix, could be attributed to flow through the
either matrix or fracture permeability of the base case was doubled fractures which counterbalanced the pressure build-up by increasing the
separately. The plume migration in the base case is compared in Fig. 15 fracture flow. The reservoir storage capacity was enhanced by 16 %
to cases when kmy and kfy were each individually doubled. more when kv/kh was increased from 0.0001 to 1 in the base case at a
The reservoir pressure response was more substantially different, pressure of 15,722 kPa. It could be speculated that an increase in kv/kh,
compared to the changes in plume migration, when the matrix and from 0.0001 to 1, comparatively enhanced the mobility of the CO2
fracture Y-direction permeabilities for the base case were doubled. The plume in both the horizontal and vertical directions, which improved
reservoir pressure depends on the ease of plume flow which seemed the pore fluid and rock surface exposure to CO2. There was no influence

10
M.A. Sohal et al. International Journal of Greenhouse Gas Control 108 (2021) 103306

Fig. 13. Part (a) shows the migration of CO2 plume in the base case and part (b) shows migration in case base_72 over 50 years of gas injection. The same volume of
CO2 was injected but the migration of plume was different in both cases as porosity changes.

Fig. 14. Reservoir shut-in pressure versus the cumulative injected mass of CO2 for a similar time period. The heterogeneity of the simulated models based on the
variation in matrix porosity.

of faults transmissibility on CO2 migration in this case. 3.5. Effect of faults transmissibility

There are eight faults inferred at the top of the reservoir from the 3-D
seismic interpretations and are assumed to be vertical within the

11
M.A. Sohal et al. International Journal of Greenhouse Gas Control 108 (2021) 103306

Fig. 15. Comparison of CO2 plume migration in all directions from the injection well when a change in the matrix or fracture permeability in the y-direction is twice
higher compared to the base case in part (a) and (b) respectively. The CO2 plume spread in both cases is almost identical but there is slightly higher gas saturation in
the matrix case at the outer periphery of the plume and pointed out with the arrows.

Fig. 16. Comparison of reservoir shut-in pressures over 50 years of gas injection for the base case when increasing either matrix or fracture permeability in the Y-
direction by a factor of two. The red dotted line shows the cumulative mass injection of CO2 for fifty years.

reservoir. Two main faults cross the storage complex from the reservoir years of gas injection in a fully non-sealing scenario of the base case, are
to the overburden, which limits the south-eastward extension of the shown in Fig. 19.
reservoir. These faults are the F11 (Ubierna fault), located at the The transmissibilities of all the faults were varied from 0 (complete
southern part, and the F9 (East fault) located at the eastern part of the barrier) to 1 (fully transmissive) in a sensitivity study. The fault trans­
Hontomín site. The location of all the eight faults, and the migration missibility multiplier values together with the simulation cases that
distance of the CO2 plume from the injection well (HI_G) following 200 were used in the sensitivity study are given in Table 6. After 200 years of

12
M.A. Sohal et al. International Journal of Greenhouse Gas Control 108 (2021) 103306

Fig. 17. Movement of CO2 plume in the horizontal and vertical direction when kv/kh increased from 0.0001 to 1 (a) and (b) respectively during gas injection for 50
years in the base case. The E-W & N-S segments of the reservoir are taken from the part in which the injection well is located. The black rectangles shown are to
highlight the key differences between cases.

Fig. 18. Simulated pressure build-up over 50 years of gas injection when kv/kh of the matrix was gradually decreased from 1 to 0.0001 for the base case. The plot
shows pressure build-up and cumulative mass injection during the prediction phase.

simulated time, the CO2 plume could still not reach the boundaries of the and the pressure difference was increasing as long as gas was being
reservoir at the injection rate which was used in the historical data. The injected. In contrast, the reservoir shut-in pressure slightly decreased
gas and water injection rates in the predictive simulation study followed when fault R4+5 was made fully transmissive relative to the sealing
the in-situ injection strategy as used in the historical data. The influence conditions. The influence of both faults on the pressures can be seen in
of the transmissibility of most of the faults on the migration of the CO2 Fig. 20 when both faults are fully transmissive, as compared to a
plume was not apparent over 200 years of gas injection. Only the in­ completely sealing state. Reservoir shut-in pressure increased from
fluence of the F11 and R4+5 faults was observed, which were the faults 14,585 kPa to 15,482 kPa in 200 years of gas injection at the historic
located within the range of CO2 plume spread. The influence of the injection rate when all the faults were completely sealing faults. The
transmissibility of fault F11 (Ubierna fault) on the migration of the CO2 reservoir storage capacity slightly increased under fully transmissive
plume was greater than for the fault R4+5 in the given scenario. It was conditions. Therefore, the influence of fault transmissibility on reservoir
observed that the shut-in reservoir pressure gradually decreased when storage capacity was minor like pressure.
the Ubierna fault was fully transmissive, compared to the sealing state,

13
M.A. Sohal et al. International Journal of Greenhouse Gas Control 108 (2021) 103306

Fig. 19. Location of faults in Sopeña Formation and migration distance of CO2 plume from the injection well in 200 years of gas injection in a non-sealing scenario.

Table 6
Faults names as defined in the Geological model and transmissibility multiplier in respective simulated cases in the sensitivity study.
Cases TM_F11 TM_F9 TM_FJ1 TM_FJ2 TM_FJ3 TM_R3 TM_R6 TM_R4+5

Base_62 0 0 0 0 0 0 0 0
Base_63 1 0 0 0 0 0 0 0
Base_64 0 1 0 0 0 0 0 0
Base_65 0 0 1 0 0 0 0 0
Base_66 0 0 0 1 0 0 0 0
Base_67 0 0 0 0 1 0 0 0
Base_68 0 0 0 0 0 1 0 0
Base_69 0 0 0 0 0 0 1 0
Base_70 0 0 0 0 0 0 0 1
Base_71 1 1 1 1 1 1 1 1

4. Conclusion cases, gas saturation in rock layers from top to bottom is relatively high
compared to the homogeneous base case for which most of the gas is
In this work simulations of long-term CO2 storage in the Hontomín stored in the topmost layer.
deep saline aquifer have been presented, with emphasis on heteroge­ Plume migration, or spread, from the injection well was almost
neity effects on reservoir storage capacity, and plume migration. The identical in both cases when either matrix or fracture permeabilities
fractured carbonate reservoir was simulated using the dual permeability were individually multiplied by the same factor, but gas saturation in the
option in E300 and validated with well test data in order to investigate plume was different between the two cases. The reservoir pressure
the impact of heterogeneities in porosity and permeability, permeability increased a bit more when matrix permeability was changed by the same
ratio, and fault transmissibility on long term CO2 injection. The CV factor as for fractures.
parameter characterizing matrix heterogeneity was gradually increased The heterogeneity based on porosity variation significantly increased
to greater than one (CV > 1) and the impact of matrix heterogeneity the reservoir storage capacity as expected and decreased the shut-in
variation was more prominent on reservoir storage capacity compared pressure compared to the homogeneous base case. The CO2 plume
to the plume migration in a fractured system. spread also reduced on increasing the matrix porosity heterogeneity.
The reservoir shut-in and injection pressures gradually increased When kv/kh ratio was equal to 1 the gas plume mainly migrated
with the level of heterogeneity based on matrix permeability, but towards the top of the structure and accumulated under the caprock.
reservoir storage capacity significantly decreased. It was concluded that However, reservoir shut-in pressure did not significantly change when
the fracture flow counterbalanced the matrix flow as the matrix het­ the matrix kv/kh ratio varied from 0.0001 to 1. This minor pressure
erogeneity increased. However, fractures do not contribute to storage difference as a result of a significant change in matrix permeability ratio
therefore reservoir storage capacity drastically decreased as heteroge­ was associated with fracture flow which counterbalanced the reduced
neity based on matrix permeability increased. It was also observed that matrix flow in a vertical direction.
the vertical gas migration in a homogeneous case was relatively larger The effect of faults F11 (Ubierna fault) and R4+5 on the migration of
compared to the heterogeneous cases and gas accumulated underneath CO2 plume and reservoir pressure were observed, out of eight normal
the caprock in the process of hydrodynamic trapping. In heterogeneous faults, which were located in the range of CO2 plume spread. The

14
M.A. Sohal et al. International Journal of Greenhouse Gas Control 108 (2021) 103306

Fig. 20. Influence of transmissibility of faults F11 and R4+5 on reservoir and injection pressure compared to the fully sealing case over 200 years of gas injection.
The cases denoted base_63 and base_70 represent when either the fault F11 or fault R4+5, respectively, were fully transmissive as compared to base_62 which
corresponds to a fully sealing state for both faults.

reservoir pressure increased by 42 kPa more when all the reservoir faults Cooper, J., Dooley, R., 2008. The International Association for the Properties of Water
and Steam. Release on the IAPWS Formulation for the Thermodynamic Properties of
were sealing, compared to the fully transmissive scenario, over 200
Seawater, p. 19.
years of gas injection. The influence of the faults on reservoir storage Dai, Z., et al., 2014. Pre-site characterization risk analysis for commercial-scale carbon
capacity was thus only minor. sequestration. Environ. Sci. Technol. 48 (7), 3908–3915.
Dandekar, A.Y., 2013. Petroleum Reservoir Rock and Fluid Properties. CRC press.
Davies, C., 2013. European Parliament Resolution. Implementation Report 2013:
Author statement Developing and Applying Carbon Capture and Storage Technology in Europe (2013/
2079(INI)).
de Dios, J.C., et al., 2017. Hydraulic characterization of fractured carbonates for CO2
M. Adeel Sohal: Conceptualization, Methodology, Investigation, geological storage: Experiences and lessons learned in Hontomín Technology
Software, Validation, Writing – Original draft preparation. Development Plant. Int. J. Greenh. Gas Control. 58, 185–200.
Yann Le Gallo: Geological Model Doughty, C., 2010. Investigation of CO2 plume behavior for a large-scale pilot test of
geologic carbon storage in a saline formation. Transp. Porous Media 82 (1), 49–76.
J.Carlos de Dios: Provided Field data
Egya, D.O., et al., 2019. Analysing the limitations of the dual-porosity response during
Pascal Audigane: Reviewing well tests in naturally fractured reservoirs. Pet. Geosci. 25 (1), 30–49.
Sean P. Rigby: Supervision, Reviewing and Editing Farajzadeh, R., et al., 2011. The effect of heterogeneity on the character of density-driven
natural convection of CO2 overlying a brine layer. Adv. Water Resour. 34 (3),
327–339.
Fenghour, A., Wakeham, W.A., Vesovic, V., 1998. The viscosity of carbon dioxide.
Declaration of Competing Interest J. Phys. Chem. Ref. Data 27 (1), 31–44.
Flett, M., Gurton, R., Weir, G., 2007. Heterogeneous saline formations for carbon dioxide
The authors report no declarations of interest. disposal: impact of varying heterogeneity on containment and trapping. J. Pet. Sci.
Eng. 57 (1–2), 106–118.
Gaus, I., 2010. Role and impact of CO2-rock interactions during CO2 storage in
Acknowledgements sedimentary rocks. Int. J. Greenh. Gas Control. 4 (1), 73–89.
Gershenzon, N.I., et al., 2015. Influence of small-scale fluvial architecture on CO2
trapping processes in deep brine reservoirs. Water Resour. Res. 51 (10), 8240–8256.
This research was funded by the European Union’s Horizon 2020 Ghanbari, S., et al., 2006. Simulation of CO2 storage in saline aquifers. Chem. Eng. Res.
research and innovation program under Grant Agreement No. 653718- Des. 84 (A9), 764–775.
ENOS. The authors acknowledge the role of funding entities, project Hassanzadeh, H., Pooladi-Darvish, M., Keith, D.W., 2009. Accelerating CO2 dissolution
in saline aquifers for geological storage - mechanistic and sensitivity studies. Energy
partners, and collaborators without which this project would not have Fuels 23 (5–6), 3328–3336.
been completed successfully. We thank Schlumberger for the donation He, L., et al., 2014. Complex relationship between porosity and permeability of
of the simulation software. carbonate reservoirs and its controlling factors: a case study of platform facies in Pre-
Caspian Basin. Pet. Explor. Dev. 41 (2), 225–234.
Hesse, M.A., Orr, F.M., Tchelepi, H.A., 2008. Gravity currents with residual trapping.
References J. Fluid Mech. 611, 35–60.
Hu, B.X., et al., 2009. Examining the influence of heterogeneous porosity fields on
conservative solute transport. J. Contam. Hydrol. 108 (3–4), 77–88.
Ahmed, T., 2018. Reservoir Engineering Handbook. Gulf professional publishing.
Iglauer, S., et al., 2011. Residual CO2 imaged with X-ray micro-tomography. Geophys.
Alcalde, J., et al., 2014. 3D geological characterization of the Hontomín CO2 storage site,
Res. Lett. 38.
Spain: Multidisciplinary approach from seismic, well-log and regional data.
Iglauer, S., et al., 2015. CO2 wettability of caprocks: Implications for structural storage
Tectonophysics 627, 6–25.
capacity and containment security. Geophys. Res. Lett. 42 (21), 9279–9284.
Al-Khdheeawi, E.A., et al., 2017. Influence of injection well configuration and rock
Journel, A.G., Huijbregts, C.J., 1978. Mining Geostatistics, 600. Academic press, London.
wettability on CO2 plume behaviour and CO2 trapping capacity in heterogeneous
Kumar, A., et al., 2005. Reservoir simulation of CO2 storage in deep saline aquifers. Spe
reservoirs. J. Nat. Gas Sci. Eng. 43, 190–206.
J. 10 (3), 336–348.
Bachu, S., Gunter, W., Perkins, E., 1994. Aquifer disposal of CO2: hydrodynamic and
Lackner, K.S., 2003. Climate change. A guide to CO2 sequestration. Science 300 (5626),
mineral trapping. Energy Convers. Manage. 35 (4), 269–279.
1677–1678.
Bennion, B., Bachu, S., 2008. Drainage and imbibition relative permeability relationships
Le Gallo, Y., de Dios, J., 2018. Geological model of a storage complex for a CO2 storage
for supercritical CO2/brine and H2S/brine systems in intergranular sandstone,
operation in a naturally-fractured carbonate formation. Geosciences 8 (9), 354.
carbonate, shale, and anhydrite rocks. Spe Reserv. Eval. Eng. 11 (03), 487–496.
Lemonnier, P., Bourbiaux, B., 2010a. Simulation of naturally fractured reservoirs. state of
Chen, C., Zeng, L., Shi, L., 2013. Continuum-scale convective mixing in geological CO2
the art-part 1–physical mechanisms and simulator formulation. Oil Gas Sci.
sequestration in anisotropic and heterogeneous saline aquifers. Adv. Water Resour.
Technol.–Revue de l’Institut Français du Pétrole 65 (2), 239–262.
53, 175–187.

15
M.A. Sohal et al. International Journal of Greenhouse Gas Control 108 (2021) 103306

Lemonnier, P., Bourbiaux, B., 2010b. Simulation of naturally fractured reservoirs. state Rapaka, S., et al., 2008. Non-modal growth of perturbations in density-driven convection
of the art-Part 2–matrix-fracture transfers and typical features of numerical studies. in porous media. J. Fluid Mech. 609, p. 285.
Oil Gas Sci. Technol.–Revue de l’Institut Français du Pétrole 65 (2), 263–286. Rashid, F., et al., 2015. Porosity and permeability of tight carbonate reservoir rocks in
Lengler, U., De Lucia, M., Kühn, M., 2010. The impact of heterogeneity on the the north of Iraq. J. Pet. Sci. Eng. 133, 147–161.
distribution of CO2: Numerical simulation of CO2 storage at Ketzin. Int. J. Greenh. Sharqawy, M.H., Lienhard, J.H., Zubair, S.M., 2010. Thermophysical properties of
Gas Control. 4 (6), 1016–1025. seawater: a review of existing correlations and data. Desalin. Water Treat. 16 (1–3),
Lindeberg, E., WesselBerg, D., 1997. Vertical convection in an aquifer column under a 354–380.
gas cap of CO2. Energy Convers. Manage. 38, S229–S234. Skjaeveland, S.M., et al., 2000. Capillary pressure correlation for mixed-wet reservoirs.
Metz, B., et al., 2005. Special Report on Carbon Dioxide Capture and Storage Prepared by Spe Reserv. Eval. Eng. 3 (1), 60–67.
the Working Group III of the Intergovernmental Panel on Climate Change. Spycher, N., Pruess, K., 2010. A Phase-Partitioning Model for CO2-Brine Mixtures at
Cambridge University Press, Cambridge. Elevated Temperatures and Pressures: Application to CO2-Enhanced Geothermal
Naylor, M., Wilkinson, M., Haszeldine, R., 2011. Calculation of CO2 column heights in Systems. Transp. Porous Media 82 (1), 173–196.
depleted gas fields from known pre-production gas column heights. Mar. Pet. Geol. Spycher, N., Pruess, K., Ennis-King, J., 2003. CO2-H2O mixtures in the geological
28 (5), 1083–1093. sequestration of CO2. I. Assessment and calculation of mutual solubilities from 12 to
Nelson, R., 2001. Geologic Analysis of Naturally Fractured Reservoirs. Elsevier. Elsevier, 100 degrees C and up to 600 bar. Geochim. Cosmochim. Acta 67 (16), 3015–3031.
2001. Tian, H., et al., 2014. Impacts of hydrological heterogeneities on caprock mineral
Pan, F., et al., 2009. Incorporating layer-and local-scale heterogeneities in numerical alteration and containment of CO2 in geological storage sites. Int. J. Greenh. Gas
simulation of unsaturated flow and tracer transport. J. Contam. Hydrol. 103 (3–4), Control. 24, 30–42.
194–205. Warren, J., Root, P., 1963. The behavior of naturally fractured reservoirs. SPEJ 3 (3),
Pan, F., et al., 2011. Sensitivity analysis of unsaturated flow and contaminant transport 245–255, 228. Trans., AIME.
with correlated parameters. J. Hydrol. 397 (3–4), 238–249. Xu, T.F., Apps, J.A., Pruess, K., 2004. Numerical simulation of CO2 disposal by mineral
Pentland, C.H., et al., 2011. Measurements of the capillary trapping of super-critical trapping in deep aquifers. Appl. Geochem. 19 (6), 917–936.
carbon dioxide in Berea sandstone. Geophys. Res. Lett. 38. Xu, X., Chen, S., Zhang, D., 2006. Convective stability analysis of the long-term storage of
carbon dioxide in deep saline aquifers. Adv. Water Resour. 29 (3), 397–407.

16