Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 936e947

Contents lists available at ScienceDirect

Journal of Rock Mechanics and

Geotechnical Engineering
journal homepage: www.rockgeotech.org

Review

Geomechanical modeling of CO2 geological storage: A review

Pengzhi Pan a, b, *, Zhenhua Wu a, b, Xiating Feng a, Fei Yan a
a
State Key Laboratory of Geomechanics and Geotechnical Engineering, Institute of Rock and Soil Mechanics, Chinese Academy of Sciences, Wuhan, 430071,
China
b
University of Chinese Academy of Sciences, Beijing, 100049, China

a r t i c l e i n f o a b s t r a c t

Article history: This paper focuses on the progress in geomechanical modeling associated with carbon dioxide (CO2)
Received 17 July 2016 geological storage. The detailed review of some geomechanical aspects, including numerical methods,
Received in revised form stress analysis, ground deformation, fault reactivation, induced seismicity and crack propagation, is
8 October 2016
presented. It is indicated that although all the processes involved are not fully understood, integration of
Accepted 10 October 2016
Available online 27 November 2016
all available data, such as ground survey, geological conditions, microseismicity and ground level
deformation, has led to many new insights into the rock mechanical response to CO2 injection. The
review also shows that in geomechanical modeling, continuum modeling methods are predominant
Keywords:
Geomechanical modeling
compared with discontinuum methods. It is recommended to develop continuumediscontinuum nu-
Carbon dioxide (CO2) geological storage merical methods since they are more convenient for geomechanical modeling of CO2 geological storage,
Continuum numerical method especially for fracture propagation simulation. The MohreCoulomb criterion is widely used in prediction
Continuumediscontinuum numerical of rock mass mechanical behavior. It would be better to use a criterion considering the effect of the
method intermediate principal stress on rock mechanical behavior, especially for the stability analysis of deeply
Fault representation seated rock engineering. Some challenges related to geomechanical modeling of CO2 geological storage
Fault reactivation are also discussed.
Fracture propagation

2016 Institute of Rock and Soil Mechanics, Chinese Academy of Sciences. Production and hosting by
Induced seismicity
Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/
licenses/by-nc-nd/4.0/).

1. Introduction Injection-induced seismicity might cause structural damage and

perhaps panic among local people (Rutqvist et al., 2014). Further-
Greenhouse gas is regarded as one of the contributions to the more, CO2 injection may introduce new hydraulic fractures within
global climate change, and sound ways to reduce carbon dioxide or near the injection zone. These fractures may propagate upwards
(CO2) emissions have been extensively studied. Carbon capture and into the lower caprock and further through the upper caprock
storage (CCS) in deep geological formations has been recognized as (Ringrose et al., 2013). As a result, the shallow drinking water might
a promising option (Nordbotten and Celia, 2011). Such geological be contaminated by the CO2 leakage (Zheng et al., 2009; Apps et al.,
formations are mainly deep sedimentary formations, including oil 2010; Keating et al., 2010).
and gas reservoirs and deep saline aquifers (Bachu, 2008; Benson Concerns for geomechanical aspects associated with geological
and Cole, 2008; Bickle, 2009). carbon storage (GCS) originated in the 1990s (Holloway and Savage,
The injection of large amounts of CO2 into the deep subsurface 1993; Rutqvist, 2012). Later, a series of studies, including generic
may be associated with a number of geomechanical risks. Fig. 1 modeling and actual CO2 injection activities (e.g. In Salah CO2
illustrates the main typical observations during CO2 injection into storage project in Algeria, WASP in Canada), showed that significant
brine aquifer (Ringrose et al., 2013). The fault or fracture zone will geomechanical changes may indeed occur, depending on the in-
behave as a flow conduit for CO2 and a focal point for rock failure. jection pressure and site-specific geomechanical conditions (Li
The pressure buildup inside the storage formation might lead to et al., 2002; Rutqvist and Tsang, 2002; Streit and Hillis, 2004;
slip and dilation along these preexisting faults and fracture zones. Yamamoto and Takahashi, 2004; Hawkes et al., 2005; Keith and
Lavoie, 2009; Rutqvist et al., 2010; Goodarzi et al., 2012; Rutqvist,
2012; Ringrose et al., 2013). In geomechanical aspects, changes in
* Corresponding author. stresses and strains, ground surface deformations and potential
E-mail address: [email protected] (P. Pan). dangers such as new caprock fracture initiation and propagation or
Peer review under responsibility of Institute of Rock and Soil Mechanics, preexisting fault opening and slippage are important for large-scale
Chinese Academy of Sciences.

http://dx.doi.org/10.1016/j.jrmge.2016.10.002
1674-7755
2016 Institute of Rock and Soil Mechanics, Chinese Academy of Sciences. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-
NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
 P. Pan et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 936e947 937

Rock mechanical strain propagating to surface

HUC
Upper caprock fault
(Main Seal Unit) ~ 600m

Lower caprock Possible vertical extension

(Second of fault/fracture Elevated pressure in
Hot Shale
reservoir volume ~ 300m
Storage Unit)
C20.1
C10.3
CO2 plume C10.2
(free-phase gas) D70
~ 1 km

Fig. 1. Sketch illustration of the main geomechanical observations in CO2 storage (Ringrose et al., 2013).

and long-term CO2 storage (Rutqvist, 2012). Therefore, it is of great failure. The reactivation of large faults might induce a significant
significance to assess the geomechanical risks and stability before seismic event (e.g. magnitude >3).
commencing the operation of CO2 injection. To simulate the main physical processes, especially the me-
Geomechanical modeling is an important way to understand chanical aspects, appropriate numerical methods should be
and predict the mechanical behavior of geological media. It can be selected.
adopted to characterize, predict and optimize subsurface systems
of increasing complexity. This increase arises foremost from the
2.2. Numerical methods
technological challenges that must be addressed to ensure the
sustainability of water, energy and environmental systems.
For geomechanical problems, the most commonly used nu-
Therefore, in this paper, the progress in modeling of some
merical methods are classified into three types (Jing, 2003). The
geomechanical aspects of CO2 geological storage is summarized. A
first comprises continuum methods, such as the finite element
review on numerical methods related to CO2 geological storage is
method (FEM), finite difference method (FDM) and boundary
presented first. Modeling of fault representation, fault activation
element method (BEM). The second comprises the discontinuum
and induced seismicity, and modeling technique for fracture
methods, including the discrete (distinct) element method (DEM),
propagation are summarized. Some challenges in geomechanical
discontinuous displacement analysis (DDA) method and discrete
modeling are also discussed.
fracture network (DFN) method. The third comprises continuume
discontinuum methods, such as hybrid FEM/BEM and FEM/DEM.
2. Numerical methods for geomechanical modeling In modeling of GCS, the numerical methods are usually com-
bined with or linked to reservoir modules or software. For example,
2.1. Main physical processes in GCS in the implementation of thermo-hydro-mechanical (THM)
coupling, two schemes, i.e. fully coupled and sequentially coupled,
The operation of CO2 geological storage will induce complex are usually used. The fully coupled scheme is far more complex,
physical processes, concerning the interactions between thermo- while the sequentially coupled one is more practical and has been
hydro-mechano-chemical fields. These interactions may be direct used widely. In a typical sequentially coupled scheme, as shown in
or indirect. For example, geomechanical and geochemical effects Fig. 2, two modules (one for fluid flow and the other for mechanical
may significantly affect the aqueous phase composition, porosity behavior) are explicitly coupled (on a time step basis), i.e. once the
and permeability of the formation, which in turn influence the flow fluid flow equations have been solved for the increments of pres-
and transport (Zhang et al., 2015a). During the injection of low- sure and temperature over a time step, these two variables for each
temperature CO2, pore pressure increase and temperature reduc- grid block will be transferred to the geomechanical model to solve
tion create geomechanical deformation of both the reservoir and
surrounding rocks. The induced mechanical aspects are varied with
time and space. In a reservoir, the mechanical response is depen-
dent on temperature, hydraulic pressure and stress change. In the TOUGH2

caprock or overburden, or the area far beyond the injection site, the Multi-phase flow and heat conduction
mechanical response is mainly related to stress change. At the
beginning of injection, the hydraulic pressure is relatively small. Pressure Permeability
Temperature THM simulation Porosity
The mechanical responses of rock mass may be reversible or elastic. Capillary pressure
Saturation
However, due to the heterogeneity of rock mass and stress change,
microseismic events might be triggered, and the ground deforma-
tion may be detectable (Rutqvist, 2012). With increasing injection THM model Non-linearity THM model
pressure, irreversible mechanical changes, such as plasticity or Discontinuity
Stress and strain
damage, may occur. If the pressure is sufficiently high, the failure of
rock mass may occur. In GCS, the generally considered modes of RDCA

failure include shear or tension of intact rock, re-shearing of

cohesionless faults, and the opening of new fractures from tensile Fig. 2. Schematic of sequential scheme of coupled THM simulations (Pan et al., 2014a).
938 P. Pan et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 936e947

the solid mechanics equations for strains and to update the me- studies (Sorai et al., 2015). Actually, in geomechanical modeling of
chanical states (deformation or failure). The new mechanical states GCS, the choice of numerical methods (continuum or dis-
will then be used to calculate the mechanical variable dependent continuum) depends on problem-specific factors, such as the pur-
permeability modifiers for the fractured grid blocks to simulate pose of the study, problem scale and fracture geometry. Therefore,
fracture propagation for the next time step in the flow model. the right selection is made if the numerical method is adequate for
The continuum-based numerical simulators, such as CODE- the representation of the mechanical behavior of rock mass.
BRIGHT (FEM) (Olivella et al., 1994, 1996; Vilarrasa et al., 2010),
FEMH (Bower and Zyvoloski, 1997; Deng et al., 2011), TOUGH-FLAC
2.3. Upscaling of geomechanical properties
(FVM þ FDM) (Rutqvist and Tsang, 2002), OpenGeoSys (FEM)
(Wang and Kolditz, 2007; Goerke et al., 2011), ECLIPSE-FE-IE
The numerical modeling of geomechanical aspects of CO2
(FDM þ FEM) (Ferronato et al., 2010), ECLIPSE-VISAGE
geological storage, such as uplift, fault reactivation and induced
(FDM þ FEM) (Ouellet et al., 2011), STARS (Bissell et al., 2011),
seismicity, is usually at a large scale. In some cases, it is impractical
NUFT-SYNEF (NUFT-GEODYN-L (nonlinear FEM code) (Vorobiev,
and unnecessary to track all local behaviors of rock mass. Simpli-
2010)) (Morris et al., 2011a,b,c), DYNAFLOW (FEM) (Prevost, 1981;
fication or upscaling can be used to reduce the computational
Preisig and Prevost, 2011), TOUGH2-Code_Aster (FEM) (Rohmer
burden. In this case, appropriate model and input parameters
and Seyedi, 2010), AEEA Coupler (FEM þ FDM) (Fei et al., 2015),
should be assigned to reasonably characterize the mechanical
and THM-CO2 (integral FDM) (Huang et al., 2015), have been
behavior of rock mass. The input hydromechanical properties must
developed and used to study the THM process related to GCS. In
be based on the values measured by small-scale laboratory tests.
these numerical methods, the elastic or elastoplastic constitutive
However, the measured values are valid only at these small scales,
relation is usually used to characterize the geomechanical behavior
representing only intact rock matrix or single small fractures.
of rocks.
Upscaling methods should be used to transform the small-scale
The discontinuum or continuumediscontinuum based numer-
data to large-scale values required as inputs to numerical models
ical methods are used to simulate the hydraulic fracturing of rocks.
(Blum et al., 2005). However, upscaling of mechanical properties is
For example, Weng et al. (2011) used the displacement disconti-
regarded as a key and challenging issue in geomechanical modeling
nuity method to develop a hydraulic fracture model to simulate
(Alain and Vincent, 2004; Guvanasen and Chan, 2004; Blum et al.,
complex-fracture-network propagation in a formation with pre-
2005; Rodriguez et al., 2006; Ne zerka et al., 2014; Liu et al., 2016;
existing fractures. Fu et al. (2013) simulated hydraulic fracture
Zeng et al., 2016). The difficulty of upscaling lies in several as-
propagation in an arbitrary discrete fracture network using an
pects. Because complete three-dimensional (3D) information about
explicit coupled hydromechanical model. Torres and Castaño
mechanical properties is never obtainable, numerous methods
(2007) introduced a discrete element simulation for the hydraulic
have been developed to interpolate among data and to use
fracturing process of rocks. Dahi Taleghani (2009) presented a
geological, hydrogeological, and geophysical information to create
complex hydraulic fracture propagation model based on extended
images of formation properties (Koltermann and Gorelick, 1996).
FEM. Damjanac et al. (2010) simulated typical hydraulic fracturing
However, one of the challenges is that the interpolation of me-
operation in a naturally fractured rock using the DEM. Hamidi and
chanical parameters determined at specific locations in the rock to
Mortazavi (2014) used 3DEC for simulating the initiation and
locations throughout the rock may be significantly different.
propagation of hydraulically induced fractures in rock mass. De
Furthermore, whether a representative element volume (REV) ex-
Pater and Beugelsdijk (2005) used DEM to handle multiple frac-
ists is also a problem. Since it is dependent on the variations in the
ture propagation. Meyer and Bazan (2011) presented a DFN nu-
fracture density and geometry, the REV may or may not exist for
merical simulator formulated for a pseudo-three-dimensional
fractured rocks. In addition, the presence of minor faults in caprock
(P3D) hydraulically induced fracture system. Nasehi and Mortazavi
greatly affects reservoir performance. These faults are usually
(2013) employed a two-dimensional (2D) DEM (UDEC) to simulate
below the resolution limit of conventional 3D seismic testing.
the fully coupled hydromechanical interaction between fluid flow
Conventional methods of fracture characterization such as one-
and rock in a typical hydraulic fracturing process. Jiao et al. (2015)
dimensional (1D) (well logs or cores) or 2D (outcrop analogs)
developed a 2D coupled hydromechanical discontinuum model for
approach lack the spatial characterization that is needed to
simulating the rock hydraulic fracturing process based on DDA.
comprehensively assess their impact on reservoir performance. To
Wang et al. (2016) presented a coupled bonded particle and lattice
gain insight into the fracture characteristics of the target formation,
Boltzmann method for modeling hydraulic fracturing. For applica-
an iterative process of modeling and incorporating production and/
tion of discontinuum methods to the study of GCS, relatively few
or injection data from field experiments may be helpful
studies have been reported. For example, the NUFT (nonisothermal
(Chiaramonte et al., 2011).
unsaturatedesaturated flow and transport model) code has been
used previously in combination with the Livermore distinct
element code (Morris et al., 2006) to investigate caprock integrity 3. Modeling of ground deformation
during CO2 storage (Johnson et al., 2005). Pan et al. (2012a, 2014b,c)
developed a continuumediscontinuum numerical model (RDCA) to During the injection of CO2 around the injection zone, the
simulate the rock fracture propagation and coalescence under fluid changes (even small changes at the beginning) of reservoir pressure
pressurization induced by CO2 injection. will induce some mechanical and hydraulic responses, including
Based on the literature review, it is found that the continuum ground deformation (or uplift), stress distribution and permeability
methods are predominant in geomechanical modeling of GCS variation. In these responses, the estimation of ground deformation
compared to discontinuum or continuumediscontinuum methods. induced by CO2 injection is important for assessing the suitability of
For numerical methods, there are no absolute advantages of one the injection site. For this purpose, the analytical methods, semi-
method over another. It should be noted that the continuum-based analytical methods and numerical models are often used to
numerical approaches are also practically useful for simulation of assess the ground deformation, stress distribution, stability and
rock failure process, especially for pre-failure damage evolution. leakage of CO2. For example, Fjaer et al. (2008) proposed a simple
For example, the TOUGH-FLAC approach has been successfully used but practical analytical formula to calculate the reservoir vertical
in laboratory-scale studies (Lei et al., 2016) and natural analog expansion:
 P. Pan et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 936e947 939

The analytical and semi-analytical models for mechanical sta-

Dh ð1 þ nÞð1  2nÞ bility and leakage of CO2 storage can also be found in the literature
¼ a DP (1) (Streit and Hillis, 2004; Lucier et al., 2006; Chiaramonte et al., 2008;
h ð1  nÞE
Mathias et al., 2009; Selvadurai, 2009; Soltanzadeh and Hawkes,
where Dh is the vertical expansion of the reservoir, h is the thick- 2009; Rohmer and Bouc, 2010; Vidal-Gilbert et al., 2010; Gor
ness of the reservoir, a is the Biot’s coefficient, n is the Poisson’s et al., 2013; Sun et al., 2013). In some cases, these simplified models
ratio, E is the Young’s modulus, and DP is the change in reservoir (analytical or semi-analytical) provide an effective and quick way to
pressure. analyze the mechanical behavior of rock mass. For example,
The analytical solution to Eq. (1) was based on the assumption Chiaramonte et al. (2011) developed a simplified, geomechanically
that the 1D reservoir is thin and laterally extensive. It does not constrained 3D reservoir fluid flow simulation at Teapot Dome Oil
consider the effects of different geological formations and the re- Field, WY, USA, a site of a CO2 enhanced oil reservoir (CO2-EOR)
striction of vertical expansion by overlying rock. Furthermore, the sequestration pilot project. Nevertheless, to further understand the
uplift in reality is not dependent on well pressure but on the failure mechanism or behavior of rocks in GCS, numerical
average pressure within the injection zone. Therefore, the model modeling, especially the associated geomechanical aspects, is
may overestimate the ground deformation. Despite this, Rutqvist required. Fig. 3 shows a typical numerically simulated ground
(2012) showed that the uplift obtained by Eq. (1) is in the correct deformation using a continuum numerical method (Zhang et al.,
order of magnitude. To consider these effects on the uplift, semi- 2015b).
analytical models and numerical methods might be more useful.
For semi-analytical models, typically, Selvadurai (2009) pro-
4. Modeling technique for faults
vided an elementary model of the interaction between a surface
rock layer and a deep rock mass that is pressurized during the in-
Geological systems are heterogeneous and discontinuous me-
jection of storage fluids. The heave (i.e. w(r)) of the surficial rock
dia, which may contain faults and fractures of various sizes, i.e. from
layer due to pressures generated by injected fluids is expressed as small meter-scale fractures to kilometer-scale faults. Sequestration
of large amounts of CO2 within such media is a challenging task
ZN since the fluid pressurization may affect an area extending to a
Ui at ell=h
wðrÞ ¼ J ðla=hÞJ0 ðlr=aÞdl
3 1
ði ¼ f ; bÞ (2) radius more than 100 km (Birkholzer and Zhou, 2009; Cappa and
h 1 þ Fi l Rutqvist, 2011a). Over such a vast area, some major faults may
0
not be avoidable (Cappa and Rutqvist, 2011a). Under fluid pres-
where f and b are the frictionless and bonded interfaces, surization, the faults might be reactivated and associated seismicity
respectively. may occur. CO2 migration could occur through existing fractures

Fig. 3. Contours of final vertical displacement distributions for the model with weak faults and injection rates of (a) 1 Mt/yr, (b) 3 Mt/yr and (c) 5 Mt/yr at the end of a 20-year
injection period (Zhang et al., 2015b).
940 P. Pan et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 936e947

and fault zones. Therefore, it is important to assess the potential be readily applied to different types of fault structures, from a single
risk due to the existence of faults numerically. The fault zones slip plane to a complex fault zone.
generally contain complex structures and a single-fault hydrome- However, difficulties can occur with the treatment of large
chanical model covering all types of faults and fault stages may not slopes and separations along the fault because the continuum
be possible. According to the scale of interest and purpose, the model paints the fault on top of the Lagrangian mesh such that it
approach applied to analysis of fault geomechanical behavior may moves with the mesh (Morris et al., 2011c). Moreover, when there
be different. For example, it is possible to represent the fault het- are some faults or fracture networks in the system, using the finite-
erogeneities in detail for small-scale simulation, while a simplified thickness elemental fault representation is still a tedious task due
fault model may be represented for large-scale simulation. For to the limitation of continuum numerical methods. For example, in
example, in the work of Morris et al. (2011a,b,c), the faults were large-scale GCS, the fault may be over 1 km in length and several
treated by introducing directional weakness into the elements of meters in width. The edge length on the fault should be small
the calculation. In the finite element analysis by Li et al. (2006), a enough within the rupture area to ensure enough cells to well
classical spring model was adopted to consider the mechanical resolve the weakening process. For example, Cappa and Rutqvist
changes of the fault. Cappa and Rutqvist (2011a) analyzed different (2012) used an elemental size of 0.25 m for the fault in the
hydromechanical fault models and demonstrated how such models rupture zone. Using such a small elemental size to simulate a fault
can be applied as part of hydromechanical analysis of a CO2 injec- over 1 km in length will require many computer resources and
tion site. Using the continuum-based TOUGH-FLAC simulator, a mesh dependency may occur. To overcome these difficulties, a
fault or fault zone was represented by a mechanical interface or discontinuum numerical method or a combination of continuum
finite-thickness solid elements with isotropic or anisotropic me- and discontinuum methods may be more convenient to represent
chanical behavior (Fig. 4). Their analysis showed that a finite- the fault or fracture. This work has been performed by some re-
thickness element approach is easier to be implemented and can searchers. For example, recently, special shape functions, i.e. the
Heaviside function to simulate the crack (or fracture) surfaces, and
the asymptotic crack-tip displacement field function, have been
designed to approximate discontinuous displacement, which is
expressed as (Stazi et al., 2003; Budyn et al., 2004):

X X
mc X
uh ðxÞ ¼ Ni ðxÞui þ Nj ðxÞHjm ðxÞam
j
i˛N m ¼ 1 j˛N m
" # (3)
X
mt X X
4
p p
þ Nk ðxÞ Fak ðxÞbak
p ¼ 1 k˛N p a¼1

where N represents the total nodal number of the element; x rep-

resents the coordinate; Nm is the set of nodes of the elements
intersected completely by the crack m; Np is the set of nodes
associated with crack tips in their influence domain; Ni, Nj and Nk
are the shape functions of the associated node; ui is the nodal
displacement (standard degrees of freedom); am j is the additional
unknown for the modified step enrichment Hjm ðxÞ of crack m; bpak is
the vector of the additional nodal degree of freedom for the tip
enrichment of tip p for the modified branch function Fapk ðxÞ; and a is
the crack tip set.
By using the discontinuous displacement function shown in Eq.
(3), with a combination of the fracture tracking method (e.g. level
set method) and partition of unity, the fracture or fault geometry is
independent of the numerical grid (Pan et al., 2012a). Therefore, a
fault with great length can be represented in a uniform grid and the
mesh dependency can be overcome. Furthermore, arbitrary frac-
ture geometry (e.g. curved fracture) can be conveniently simulated.
Fig. 5 shows the case of a zigzag caprock fracture geometry, through
simply updating the discontinuity geometry (Pan et al., 2014a),
demonstrating the versatility of this method.

5. Modeling of fault activation and induced seismicity

5.1. Fault activation

The possible activation of a preexisting fault may occur when-

ever the shear stress acting on the fracture plane exceeds the
MohreCoulomb failure criterion (Wiprut and Zoback, 2000; Jaeger
et al., 2007):

Fig. 4. Representation of a fault (Cappa and Rutqvist, 2011a). (a) A zero-thickness jss j  c þ s0 n tanf (4)
interface, (b) solid elements and (c) solid elements with ubiquitous joints oriented
as weak planes along the strike of fault plane.
 P. Pan et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 936e947 941

Fig. 5. The fracture mechanical behavior during injection of CO2 (1000). (a) 12 months, (b) 15 months, (c) 18 months and (d) 36 months (Pan et al., 2014a).

where s0 n and ss are the normal and shear stresses, respectively; c is The MohreCoulomb failure criterion in Eq. (4) or its modifica-
the cohesion; f is the friction angle; and tanf represents the static tion is practically useful for describing the slip tendency. In the
friction coefficient. s0 n is positive when the fault is in compression. analysis, the stress is an important factor to be considered in the
s0 n represents the effective normal stress on the fault, which is evaluation of fault activation or slip tendency. In most studies, the
reduced by an increase in pore pressure or by a reduction in the in slip tendency is calculated based on the pre-injection principal
situ confining stress. The fault can slip when jssj is greater than or stress magnitude and orientation corresponding to the regional
equal to the right-hand side of Eq. (4) (Fig. 6). The fault may open stress field. However, studies show that in situ stress state can be
when the tensile stress is applied. changed because of poro-elastic stress development related to the
In practice, given that no direct measurements of c can be used changes in pore pressure and temperature (Kano et al., 2014).
in Eq. (4), the contribution of fault cohesion to shear strength is Therefore, it is necessary to assess the impact of stress uncertainty
usually neglected (i.e. c ¼ 0 MPa or a small value) (Ferronato et al., induced by CO2 injection on fault stability.
2010; Cappa and Rutqvist, 2011a). In most of the generic modeling In most recent studies, fault reactivation is treated as a quasi-
studies, a slip-weakening modeling is used. In this case, the friction static mechanical process (Cappa and Rutqvist, 2011a,b,c). In a
coefficient drops from a peak value to a residual value to model the dynamic simulation of fault slip induced by CO2 injection, rate- and
rupture of the fault. Although the peak friction coefficient (i.e. tanf) state-dependent friction laws should be involved. In this case,
is typically in the range of 0.6e0.85 (with the friction angle of about friction is assumed to evolve from the static value to a reference
30 e39 ) (Goodman, 1975; Byerlee, 1978; Vidal-Gilbert et al., dynamic value linearly with increasing slip displacement, until the
2009), many researchers used a lower value of peak friction coef- critical distance is reached. Thereafter, the friction coefficient de-
ficient (e.g. 0.47, corresponding to the friction angle of 25 , or even pends on sliding velocity, according to the rate- and state-
lower friction angle of 14 ) (Streit and Hillis, 2004; Cappa and dependent friction law (Urpi et al., 2016). This law can be inter-
Rutqvist, 2011c; Pereira et al., 2014; Figueiredo et al., 2015; Wei preted from laboratory experiments in the conceptual model of the
et al., 2016a). A residual friction coefficient of 0.2 (i.e. a friction rate- and state-dependent friction laws (Scholz, 1998; Niemeijer
angle of approximately 11 ) is often used (Cappa and Rutqvist, et al., 2012). In this case, the reliability of laboratory data, when
2012). It should be noted that the friction coefficients of faults are extrapolated to faults in nature, especially in terms of upscaling the
dependent on several factors, such as temperature, degree of results from laboratory to field data, should be evaluated.
saturation and the constituents of the fault content (Saffer and
Marone, 2003). Experimental studies indicate that the friction co- 5.2. Induced microseismicity
efficient of pure smectite under the vacuum-dried condition is 0.3e
0.4, while it would be 0.13e0.3 under the saturated condition In large-scale storage of CO2 in deep underground reservoirs,
(Morrow et al., 1992, 2000). The friction coefficient of smectite- concern has been focused on whether nearby faults could be
quartz mixtures is approximately 0.2e0.53, while it is much reactivated since fault reactivation might be accompanied by a
higher for illite shale (0.41e0.73) (Saffer and Marone, 2003). seismic event. Many relevant modeling studies have been con-
Therefore, these factors should be considered in the selection of the ducted (e.g. Lucier et al., 2006; Bissell et al., 2011; Lei et al., 2013;
friction coefficient. Atkinson et al., 2016) and the analyses show great uncertainties
942 P. Pan et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 936e947

Fig. 6. (a) Normal and shear stresses resolved on a fault with a given orientation from the remote principal stresses and pressure in fault by fluid injection; (b) Mohr diagram of
shear stress versus effective normal stress showing how the increasing fluid pressure may activate a fault.

in the in situ stress and the assumed strength properties. In GCS, propagation, where each source strength and radiation pattern are
tensile and shear failure occurs as a consequence of the injection determined by the pore pressure and a generalized moment-tensor
and the common criterion to decide that failure is based on a theory, respectively.
critical fluid pressure for fracturing that exceeds a given tectonic As mentioned above, if the MohreCoulomb failure criterion is
stress (Rutqvist et al., 2008; Stanchits et al., 2011). To use this cri- used, the fault will reactivate when jssj is greater than or equal to
terion, an accurate estimation of in situ stress field is essential the right-hand side of Eq. (4). The microseismicity or earthquake
(Rutqvist et al., 2008). However, Carcione et al. (2015) found that it induced by fault reactivation is mathematically modeled by a shear
is easier to establish the failure criterion on the basis of the strength displacement discontinuity (dislocation) across a surface S in an
of the rock, since this information (stiffness modulus) can be ob- elastic medium. This dislocation is equivalent to a distribution of
tained from seismic data. Langenbruch and Shapiro (2014) showed double couples on this surface whose total moment (i.e. seismic
that the elastic heterogeneity of rocks obtained from sonic and moment) M0 is (Kanamori and Anderson, 1975)
density logs along boreholes causes significant fluctuations of
fracture reactivation and opening pressures. The stiffness modulus M0 ¼ mAd (5)
can be obtained from seismic and sonic-log data or ultrasonic ex-
periments on cores. Other criteria exist to determine the emission, where m is the shear modulus of the fault media (Pa), A is the
for instance, Rozhko (2010) used the effective-stress law and the rupture area of S (m2), and d is the average dislocation or slip (m).
Coulomb yielding stress obtained from geomechanical triaxial Fig. 7 shows the typical simulation results of seismic moment by
laboratory measurements. Carcione et al. (2015) introduced a Cappa and Rutqvist (2011b). The seismological theories are used to
realistic forward modeling algorithm to simulate P- and S-wave calculate the magnitude (M) of microseismicity:

Fig. 7. Earthquake scaling relationship after Viegas et al. (2010): source dimension (radius) and seismic moment. Red and black circles correspond to the simulation results by Cappa
and Rutqvist (2011b).
 P. Pan et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 936e947 943

methods, the discontinuum method may be more convenient to

M ¼ log10 M0 =1:5  6:1 (6) simulate the propagation of a single fracture, multiple fractures and
In the modeling, the magnitude of microseismicity depends on even the fracture networks. For example, Pan et al. (2014a) devel-
the constitutive law for the fault. A brittle fault with a peak and a oped a TOUGH-RDCA simulator for modeling rock discontinuous
residual friction coefficient may produce an earthquake large mechanical behavior under multiphase fluid flow conditions. In
enough to be felt by humans. this simulator, the fracture geometry is independent of the nu-
Regarding the microseismicity propagation patterns, since merical grid so that no remeshing is needed during the propagation
plastic strain occurs when a seism occurs, the region where seismic of rock fractures, which is essentially different from continuum
events would occur can be assessed by tracking the evolution of numerical methods and therefore greatly simplifies the simulation
plastic strain, which can be obtained by using an elastoplastic of fracture propagation. The TOUGH-RDCA simulator was used to
constitutive law or elasto-viscoplastic constitutive law (Vilarrasa study the propagation of single and multiple fractures in caprock
et al., 2010, 2013). during CO2 injection into a deep brine aquifer (Pan et al., 2013,
Fault reactivation may be accompanied by seismicity as 2014b). The initiation, propagation and coalescence of fractures in
mentioned above. However, it may also be aseismic. Therefore, an caprock and the CO2 migration can be simulated explicitly by using
advanced technique should be developed to detect the aseismic this method (Figs. 8 and 9) (Pan et al., 2014b).
fault slip since it will result in leakage quietly and should be paid It should be noted that the rock mass mechanical changes
more attention. induced by CO2 injection are gradual. Before cracking, the rock can
be regarded as a continuous formation, which means that
displacement continuity is maintained. After cracking, the rock can
6. Modeling technique for fracture propagation be considered as a discontinuous medium since displacement
discontinuity occurs. Therefore, the challenge is how to find nu-
As mentioned above, the caprock may be discontinuous and merical methods to describe the rock fracturing process from
may contain heterogeneities such as faults and fractures of various continuity to discontinuity. The continuum damage mechanics and
sizes, i.e. from small meter-scale fractures to kilometer-scale faults. fracture mechanics are usually used to describe the continuity and
One of the most important concerns with respect to the safe storage discontinuity parts, respectively. However, the link between these
of CO2 is that the formation fracturing could potentially provide two approaches suffers from several difficulties: energetic equiva-
pathways for CO2 leakage through previously impermeable rocks. It lence, initial crack shape and size definition. The application of
is important to model fracture propagation for several reasons damage-based propagation in the XFEM framework is a possible
(Goodarzi et al., 2012): (1) to avoid fracture propagation through way to address this problem (Minnebo et al., 2011).
the caprock to the extent that it would create a loss of containment Although productive work has been obtained in the modeling of
(i.e. connect to other permeable zones) by controlling the injection the rock fracturing problem, the modeling of 3D fracturing is still a
rate, (2) to use the information on fracture length to correctly challenge. There are only a few papers concerned about the 3D
design the well pattern, and (3) to be able to control the injection implementation of such methods. One of the difficult issues in a 3D
rate to avoid excessive fracture lengths. setting is how to describe and track the crack surface (Rabczuk
However, because of the complexity of rock formations, analysis et al., 2010). In this case, the difficulties in the analysis of 3D frac-
of rock fracture propagation is one of the challenges in computa- ture problems by the standard FEM lie in (1) the need to construct a
tional rock mechanics. mesh that conforms to both the crack surface and the body ge-
Over the past few decades, several approaches have been pro- ometry, (2) the need to refine the mesh around the crack front, and
posed to model crack problems (Abdelaziz and Hamouine, 2008): (3) the need to remesh at each growth step in crack propagation
the method based on quarter-point finite element (Henshell and studies. By using the enrichment functions, which are realized
Shaw, 1975), enriched FEM (Benzley, 1974; Gifford and Hilton, through the partition of unity concept, the mesh does not need to
1978), the boundary collocation method (Newman, 1971), the in- conform to the crack boundaries. The 3D curved cracks can be
tegral equation method, the body force method (Nisitani, 1985), tracked by using the level set method (González-Albuixech et al.,
BEM (Cruse, 1988), the dislocation method (Vitek, 1977), and mesh- 2013).
free methods such as the element-free Galerkin method (Belytschko
et al., 1994). To avoid the remeshing step in crack modeling, several 7. Discussion and conclusions
techniques have been proposed: the incorporation of a discontin-
uous model on an element level (Oliver et al., 2002), a moving mesh In this paper, we presented a review on the geomechanical
technique (Rashid, 1998), and an enrichment technique based on a modeling of CO2 geological storage, emphasizing on numerical
partition of unity XFEM (Belytschko and Black, 1999). methods, representation of a fault, fault activation and fracture
For determination of crack propagation, two sets of criteria, i.e. propagation. Based on this work, some possible challenges and
toughness-based fracturing criterion and MohreCoulomb criterion recommendations in geomechanical modeling are listed below:
with tension cutoff, have been commonly used to simulate the
fracturing behavior of rocks. The toughness-based fracturing crite- (1) Rock mass is significantly heterogeneous. It usually contains
rion is typically employed in small-scale fracture propagation complex combinations of constituents. With the long history
(Adachi et al., 2007). For large-scale simulation (e.g. CO2 injection of formation, various geological media are ubiquitous, which
into brine aquifer), the MohreCoulomb criterion with tension cutoff makes rock masses a difficult material for mathematical
is often used to determine the condition of fracture propagation. representation via numerical modeling. For example, a
This criterion uses well-known parameters: tensile strength, friction typical CO2 storage reservoir usually contains different
angle, cohesion and dilation angle (Zhang et al., 2015b). geological groups (e.g. lower aquifer, caprock rock and upper
For the modeling of fracturing induced by CO2 injection, only a aquifer), faults and cracks, etc. In numerical modeling, a
few publications can be found in the literature. For example, Ji et al. convenient way to represent the interfaces between different
(2009) presented a fracturing model that couples a 3D finite material domains is still a challenge. Under mechanical
element geomechanical model with a conventional 3D finite dif- loading, the interface may behave strain continuity (weak
ference reservoir simulator. Compared to the continuum numerical discontinuity) initially and then displacement discontinuity
944 P. Pan et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 936e947

Fig. 8. Fracturing of F1 and connections to F2 in caprock during CO2 injection (Pan et al., 2014b).

Fig. 9. CO2 saturation evolution during injection (Pan et al., 2014b) (X and Z in m).

(strong discontinuity) may occur. One possible solution is to adjustment will lead to further failure of rock mass. There-
use an enrichment function in displacement approximation fore, it is recommended to consider the whole failure process
through the concept of partition of unity. The interfaces can of rock mass in geomechanical modeling of GCS. The frac-
be located by level set method (Sukumar et al., 2001). By turing of rock mass is actually a continuousediscontinuous
doing so, the modeling of large-scale CO2 storage in complex process, which generally experiences elastic deformation,
geological conditions would be greatly simplified. plastic deformation or damage, fracture initiation, fracture
(2) It is recognized that CO2 injection may stimulate or reactivate propagation and coalescence. A numerical method to char-
natural fractures and may introduce new hydraulic fractures acterize this continuousediscontinuous deformation failure
within or near the injection zone. These fractures may process induced by CO2 injection is needed.
propagate upwards into the lower caprock and further (3) CO2 is usually injected into subsurface media with great
through the upper caprock (Ringrose et al., 2013). For the depth, where the stress environment is much more complex
initiation of fracture or crack in caprock, Rutqvist et al. (2002) than that in near-surface rock engineering. Therefore, any of
used the pressure margin to assess the possible fracturing the three mutually perpendicular principal stress compo-
region. This is an elastic analysis and provides a quick way to nents in 3D stress space should not be neglected in the
evaluate the approximate fracturing zone. Using this modeling. Rock will exhibit different failure conditions under
method, the fracturing zone may be smaller than that using different stress paths. In previous studies, the Mohre
an elastoplastic analysis, since the plasticity-induced stress Coulomb criterion, which neglects the effect of the
 P. Pan et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 936e947 945

intermediate principal stress, is widely used to determine Seismological Research Letters 2016;87(5). http://dx.doi.org/10.1785/
0220150263.
whether the rock mass yields or not. However, many studies
Bachu S. CO2 storage in geological media: role, means, status and barriers to
have shown that both the intermediate and minimum deployment. Progress in Energy and Combustion Science 2008;34(2):254e
principal stresses greatly affect the mechanical behavior of 73.
rock mass (Mogi, 2007; Pan et al., 2012b,c). Therefore, in the Belytschko T, Black T. Elastic crack growth in finite elements with minimal
remeshing. International Journal for Numerical Methods in Engineering
assessment of rock mass stability in GCS, a yield criterion 1999;45(5):601e20.
considering both the intermediate and minimum principal Belytschko T, Lu YY, Gu L. Element free Galerkin methods. International Journal for
stresses is recommended. Numerical Methods in Engineering 1994;37(2):229e56.
Benson SM, Cole DR. CO2 sequestration in deep sedimentary formations. Elements
(4) Although the induced seismicity has been extensively stud- 2008;4(5):325e31.
ied by researchers, challenges in assessing the induced Benzley SE. Representation of singularities with isoparametric finite elements. In-
seismicity risk in CO2 storage still exist. Due to geophysical ternational Journal for Numerical Methods in Engineering 1974;8(3):537e45.
Bickle MJ. Geological carbon storage. Nature Geoscience 2009;2(12):815e8.
characterization and monitoring limitations, there are very Birkholzer JT, Zhou Q. Basin-scale hydrogeologic impacts of CO2 storage: capacity
large uncertainties in the inputs of the modeling (White and and regulatory implications. International Journal of Greenhouse Gas Control
Foxall, 2016). Therefore, novel strategies are necessary for 2009;3(6):745e56.
Bissell RC, Vasco DW, Atbi M, Hamdani M, Okwelegbe M, Goldwater MH. A full field
reducing and/or circumventing these uncertainties (Wei
simulation of the In Salah gas production and CO2 storage project using a
et al., 2016b). coupled geo-mechanical and thermal fluid flow simulator. Energy Procedia
(5) A true geomechanical model for CO2 storage requires com- 2011;4:3290e7.
Blum P, Mackay R, Riley MS, Knight JL. Performance assessment of a nuclear waste
plete knowledge of the geometrical and physical properties
repository: upscaling coupled hydro-mechanical properties for far-field trans-
and parameters of the fractured rock masses. In other words, port analysis. International Journal of Rock Mechanics and Mining Sciences
if the modeling is to incorporate the main components of the 2005;42(5e6):781e92.
rock realitydthe fractures, inhomogeneity, anisotropy and Bower KM, Zyvoloski G. A numerical model for thermo-hydro-mechanical coupling
in fractured rock. International Journal of Rock Mechanics and Mining Sciences
inelasticity, including failureda more extensive model and a 1997;34(8):1201e11.
more extensive rock mass characterization are required. Budyn E, Zi G, Moës N, Belytschko T. A method for multiple crack growth in brittle
However, the great uncertainties of geological conditions in materials without remeshing. International Journal for Numerical Methods in
Engineering 2004;61(10):1741e70.
GCS make this complete knowledge impossible. Thus, the Byerlee J. Friction of rocks. Pure and Applied Geophysics 1978;116(4e5):615e26.
challenge is to know how to develop an adequate model. This Cappa F, Rutqvist J. Impact of CO2 geological sequestration on the nucleation of
will rest on a scientific foundation and require empirical earthquakes. Geophysical Research Letters 2011b;38(17). http://dx.doi.org/
10.1029/2011GL048487.
judgments supported by accumulated experiences through Cappa F, Rutqvist J. Impact of CO2 geological sequestration on the nucleation of
long-term practices (Jing, 2003). seismic fault ruptures. In: Proceedings of the 45th US Rock Mechanics/Geo-
mechanics Symposium. American Rock Mechanics Association; 2011c.
Cappa F, Rutqvist J. Modeling of coupled deformation and permeability evolution
The above issues or challenges cannot cover all points of diffi-
during fault reactivation induced by deep underground injection of CO2. In-
culty in geomechanical modeling in GCS. Other issues, such as the ternational Journal of Greenhouse Gas Control 2011a;5(2):336e46.
large-scale computational capacity, the scale and time effects, long- Cappa F, Rutqvist J. Seismic rupture and ground accelerations induced by CO2 injection
in the shallow crust. Geophysical Journal International 2012;190(3):1784e9.
term behavior of rock mass, the evaluation of geological and en-
Carcione JM, Da Col F, Currenti G, Cantucci B. Modeling techniques to study CO2-
gineering uncertainties, and the determination of in situ stress, are injection induced micro-seismicity. International Journal of Greenhouse Gas
also the major concerns and should be further studied for the long- Control 2015;42:246e57.
term storage of CO2. Chiaramonte L, Zoback M, Friedmann J, Stamp V, Zahm C. Fracture characterization
and fluid flow simulation with geomechanical constraints for a CO2eEOR and
sequestration project Teapot Dome Oil Field, Wyoming, USA. Energy Procedia
Conflict of interest 2011;4:3973e80.
Chiaramonte L, Zoback MD, Friedmann J, Stamp V. Seal integrity and feasibility of
CO2 sequestration in the Teapot Dome EOR pilot: geomechanical site charac-
The authors wish to confirm that there are no known conflicts of terization. Environmental Geology 2008;54(8):1667e75.
interest associated with this publication and there has been no Cruse TA. Boundary element analysis in computational fracture mechanics.
Springer; 1988.
significant financial support for this work that could have influ-
Dahi Taleghani A. Analysis of hydraulic fracture propagation in fractured reservoirs:
enced its outcome. an improved model for the interaction between induced and natural fractures.
PhD Thesis. The University of Texas at Austin; 2009.
Damjanac B, Gil I, Pierce M, Sanchez M, Van As A, McLennan J. A new approach to
Acknowledgements hydraulic fracturing modeling in naturally fractured reservoirs. In: The 44th US
Rock Mechanics Symposium and 5th US-Canada Rock Mechanics Symposium.
This work was finically supported by the National Natural Sci- American Rock Mechanics Association; 2010.
De Pater CJ, Beugelsdijk LJL. Experiments and numerical simulation of hydraulic
ence Foundation of China (Grant Nos. 41272349 and 51322906) and
fracturing in naturally fractured rock. In: Alaska Rocks 2005, the 40th US
Key Research Program of Frontier Sciences, Chinese Academy of Symposium on Rock Mechanics (USRMS). American Rock Mechanics Associa-
Sciences (Grant No. QYZDB-SSW-DQC029). tion; 2005.
Deng H, Dai Z, Jiao Z, Stauffer PH, Surdam RC. Simulation of CO2 sequestration
at Rock Spring uplift, Wyoming: heterogeneity and uncertainties in storage
References capacity, injectivity and leakage. Los Alamos National Laboratory (LANL);
2011.
Abdelaziz Y, Hamouine A. A survey of the extended finite element. Computers & Fei WB, Li Q, Wei XC, Song RR, Jing M, Li XC. Interaction analysis for CO2 geological
Structures 2008;86(11e12):1141e51. storage and underground coal mining in Ordos Basin, China. Engineering Ge-
Adachi J, Siebrits E, Peirce A, Desroches J. Computer simulation of hydraulic frac- ology 2015;196:194e209.
tures. International Journal of Rock Mechanics and Mining Sciences 2007;44(5): Ferronato M, Gambolati G, Janna C, Teatini P. Geomechanical issues of anthropo-
739e57. genic CO2 sequestration in exploited gas fields. Energy Conversion and Man-
Alain T, Vincent R. Hydro-mechanical upscaling of a fractured rockmass using a 3D agement 2010;51(10):1918e28.
numerical approach. In: Stephanson O, editor. Elsevier Geo-Engineering Book Figueiredo B, Tsang CF, Rutqvist J, Bensabat J, Niemi A. Coupled hydro-mechanical
Series, vol. 2. Elsevier; 2004. p. 275e80. processes and fault reactivation induced by CO2 Injection in a three-layer
Apps JA, Zheng L, Zhang Y, Xu T, Birkholzer JT. Evaluation of potential changes in storage formation. International Journal of Greenhouse Gas Control 2015;39:
groundwater quality in response to CO2 leakage from deep geologic storage. 432e48.
Transport in Porous Media 2010;82(1):215e46. Fjaer E, Holt RM, Horsrud P, Raaen AM, Risnes R. Petroleum related rock mechanics.
Atkinson GM, Eaton DW, Ghofrani H, Walker D, Cheadle B, Schultz R, 2nd ed. Elsevier; 2008.
Shcherbakov R, Tiampo K, Gu J, Harrington RM, Liu Y, van der Baan M, Kao H. Fu P, Johnson SM, Carrigan CR. An explicitly coupled hydro-geomechanical model
Hydraulic fracturing and seismicity in the western Canada sedimentary basin. for simulating hydraulic fracturing in arbitrary discrete fracture networks.
946 P. Pan et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 936e947

International Journal for Numerical and Analytical Methods in Geomechanics Mathias SA, Hardisty PE, Trudell MR, Zimmerman RW. Screening and selection of
2013;37(14):2278e300. sites for CO2 sequestration based on pressure buildup. International Journal of
Gifford LN, Hilton P. Stress intensity factors by enriched finite elements. Engineering Greenhouse Gas Control 2009;3(5):577e85.
Fracture Mechanics 1978;10(3):485e96. Meyer BR, Bazan LW. A discrete fracture network model for hydraulically induced
Goerke UJ, Park CH, Wang W, Singh AK, Kolditz O. Numerical simulation of multi- fractures-theory, parametric and case studies. In: SPE Hydraulic Fracturing
phase hydromechanical processes induced by CO2 injection into deep saline Technology Conference. Society of Petroleum Engineers; 2011.
aquifers. Oil & Gas Science and Technology e Revue d’IFP Energies nouvelles Minnebo H, Van Hoof T, Lani F. Damage to fracture in 3D with XFEM. In: The 5th
2011;66(1):105e18. International Conference on Advanced Computational Methods in Engineering
González-Albuixech VF, Giner E, Tarancón JE, Fuenmayor FJ, Gravouil A. Domain (ACOMEN 2011). University of Liège; 2011.
integral formulation for 3D curved and non-planar cracks with the extended Mogi K. Experimental rock mechanics. London: Taylor & Francis/A.A.Balkema; 2007.
finite element method. Computer Methods in Applied Mechanics and Engi- Morris JP, Hao Y, Foxall W, McNab W. In Salah CO2 storage JIP: hydromechanical
neering 2013;264:129e44. simulations of surface uplift due to CO2 injection at In Salah. Energy Procedia
Goodarzi S, Settari A, Keith D. Geomechanical modeling for CO2 storage in Nisku 2011a;4:3269e75.
aquifer in Wabamun Lake area in Canada. International Journal of Greenhouse Morris JP, Detwiler RL, Friedmann SJ, Vorobiev OY, Hao Y. The large-scale geo-
Gas Control 2012;10:113e22. mechanical and hydrogeological effects of multiple CO2 injection sites on for-
Goodman RE. Methods of geological engineering in rock discontinuities. St. Paul: mation stability. International Journal of Greenhouse Gas Control 2011b;5(1):
West Publishers; 1975. 69e74.
Gor GY, Stone HA, Prévost JH. Fracture propagation driven by fluid outflow from a Morris JP, Hao Y, Foxall W, McNab W. A study of injection-induced mechanical
low-permeability aquifer. Transport in Porous Media 2013;100(1):69e82. deformation at the In Salah CO2 storage project. International Journal of
Guvanasen V, Chan T. Upscaling the thermohydromechanical properties of a frac- Greenhouse Gas Control 2011c;5(2):270e80.
tured rock mass using a modified crack tensor theory. In: Stephanson O, editor. Morris JP, Rubin MB, Block GI, Bonner MP. Simulations of fracture and fragmenta-
Elsevier Geo-Engineering Book Series, vol. 2. Elsevier; 2004. p. 251e6. tion of geologic materials using combined FEM/DEM analysis. International
Hamidi F, Mortazavi A. A new three dimensional approach to numerically model Journal of Impact Engineering 2006;33:463e73.
hydraulic fracturing process. Journal of Petroleum Science and Engineering Morrow CA, Moore DE, Lockner DA. The effect of mineral bond strength and
2014;124:451e67. adsorbed water on fault gouge frictional strength. Geophysical Research Letters
Hawkes CD, McLellan P, Bachu S. Geomechanical factors affecting geological storage 2000;27(6):815e8.
of CO2 in depleted oil and gas reservoirs. Journal of Canadian Petroleum Morrow CA, Radney B, Byerlee J. Frictional strength and the effective pressure law of
Technology 2005;44(10):52e61. montmorillonite and illite clays. International Geophysics 1992;51:69e88.
Henshell RD, Shaw KG. Crack tip finite elements are unnecessary. International Nasehi MJ, Mortazavi A. Effects of in-situ stress regime and intact rock strength
Journal for Numerical Methods in Engineering 1975;9(3):495e507. parameters on the hydraulic fracturing. Journal of Petroleum Science and En-
Holloway S, Savage D. The potential for aquifer disposal of carbon dioxide in the UK. gineering 2013;108:211e21.
Energy Conversion and Management 1993;34(9e11):925e32. Newman JC. An improved method of collocation for the stress analysis of cracked
Huang ZQ, Winterfeld PH, Xiong Y, Wu YS, Yao J. Parallel simulation of fully-coupled plates with various shaped boundaries. Washington, D.C.: National Aeronautics
thermal-hydro-mechanical processes in CO2 leakage through fluid-driven and Space Administration; 1971.
fracture zones. International Journal of Greenhouse Gas Control 2015;34:39e51. Nezerka V, Slí
zková Z, Tesárek P, Plachý T, Frankeová D, Petrán ová V. Comprehen-
Jaeger JC, Cook NGW, Zimmerman RW. Fundamentals of rock mechanics. 4th ed. sive study on mechanical properties of lime-based pastes with additions of
Oxford, UK: Wiley-Blackwell; 2007. metakaolin and brick dust. Cement and Concrete Research 2014;64:17e29.
Ji L, Settari A, Sullivan RB. A novel hydraulic fracturing model fully coupled with Niemeijer A, Di Toro G, Griffith WA, Bistacchi A, Smith SAF, Nielsen S. Inferring
geomechanics and reservoir simulation. SPE Journal 2009;14(3):423e30. earthquake physics and chemistry using an integrated field and laboratory
Jiao YY, Zhang HQ, Zhang XL, Li HB, Jiang QH. A two-dimensional coupled hydro- approach. Journal of Structural Geology 2012;39:2e36.
mechanical discontinuum model for simulating rock hydraulic fracturing. In- Nisitani H. Body force method for determination of stress intensity factors. Journal
ternational Journal for Numerical and Analytical Methods in Geomechanics of Aerospace Sciences and Technologies 1985;37:21e41.
2015;39(5):457e81. Nordbotten JM, Celia MA. Geological storage of CO2: modeling approaches for large-
Jing L. A review of techniques, advances and outstanding issues in numerical scale simulation. John Wiley & Sons; 2011.
modelling for rock mechanics and rock engineering. International Journal of Olivella S, Carrera J, Gens A, Alonso EE. Nonisothermal multiphase flow of brine and
Rock Mechanics and Mining Sciences 2003;40(3):283e353. gas through saline media. Transport in Porous Media 1994;15(3):271e93.
Johnson JW, Nitao JJ, Morris JP. Reactive transport modeling of cap-rock integrity Olivella S, Gens A, Carrera J, Alonso EE. Numerical formulation for a simulator
during natural and engineered CO2 storage. In: Thomas DC, editor. Carbon di- (CODE_BRIGHT) for the coupled analysis of saline media. Engineering Compu-
oxide capture for storage in deep geologic formations, vol. 2. Amsterdam: tations 1996;13(7):87e112.
Elsevier; 2005. p. 787e813. Oliver J, Huespe AE, Pulido MDG, Chaves E. From continuum mechanics to fracture
Kanamori H, Anderson DL. Theoretical basis of some empirical relations in seis- mechanics: the strong discontinuity approach. Engineering Fracture Mechanics
mology. Bulletin of the Seismological Society of America 1975;65(5):1073e95. 2002;69(2):113e36.
Kano Y, Funatsu T, Nakao S, Kusunose K, Ishido T, Lei X, Tosha T. Analysis of changes Ouellet A, Bérard T, Desroches J, Frykman P, Welsh P, Minton J, Pamukcu Y, Hurter S,
in stress state and fault stability related to planned CO2 injection at the Tom- Schmidt-Hattenberger C. Reservoir geomechanics for assessing containment in
akomai offshore site. Energy Procedia 2014;63:2870e8. CO2 storage: a case study at Ketzin, Germany. Energy Procedia 2011;4:3298e
Keating EH, Fessenden J, Kanjorski N, Koning DJ, Pawar R. The impact of CO2 on 305.
shallow groundwater chemistry: observations at a natural analog site and im- Pan PZ, Feng XT, Zhou H. Development and applications of the elasto-plastic cellular
plications for carbon sequestration. Environmental Earth Sciences 2010;60(3): automaton. Acta Mechanica Solida Sinica 2012b;25(2):126e43.
521e36. Pan PZ, Yan F, Feng XT. Modeling the cracking process of rocks from continuity to
Keith D, Lavoie R. An overview of the Wabamun area CO2 sequestration project discontinuity using a cellular automaton. Computers & Geosciences 2012a;42:
(WASP). Energy Procedia 2009;1(1):2817e24. 87e99.
Koltermann CE, Gorelick SM. Heterogeneity in sedimentary deposits: a review of Pan PZ, Feng XT, Hudson JA. The influence of the intermediate principal stress on rock
structure-imitating, process-imitating, and descriptive approaches. Water Re- failure behaviour: a numerical study. Engineering Geology 2012c;124:109e18.
sources Research 1996;32(9):2617e58. Pan PZ, Rutqvist J, Feng XT, Yan F. Modeling of caprock discontinuous fracturing
Langenbruch C, Shapiro S. Gutenberg-Richter relation originates from Coulomb during CO2 injection into a deep brine aquifer. International Journal of Green-
stress fluctuations caused by elastic rock heterogeneity. Journal of Geophysical house Gas Control 2013;19:559e75.
Research: Solid Earth 2014;119(2):1220e34. Pan PZ, Rutqvist J, Feng XT, Yan F. An approach for modeling rock discontinuous
Lei X, Funatsu T, Ma S, Liu L. A laboratory acoustic emission experiment and nu- mechanical behavior under multiphase fluid flow conditions. Rock Mechanics
merical simulation of rock fracture driven by a high-pressure fluid source. and Rock Engineering 2014a;47(2):589e603.
Journal of Rock Mechanics and Geotechnical Engineering 2016;8(1):27e34. Pan PZ, Rutqvist J, Feng XT, Yan F. TOUGH-RDCA modeling of multiple fracture in-
Lei X, Ma S, Chen W, Pang C, Zeng J, Jiang B. A detailed view of the injection-induced teractions in caprock during CO2 injection into a deep brine aquifer. Computers
seismicity in a natural gas reservoir in Zigong, southwestern Sichuan Basin, & Geosciences 2014b;65:24e36.
China. Journal of Geophysical Research: Solid Earth 2013;118(8):4296e311. Pan PZ, Rutqvist J, Feng XT, Yan F, Jiang Q. A discontinuous cellular automaton
Li Q, Wu Z, Bai Y, Yin X, Li X. Thermo-hydro-mechanical modeling of CO2 seques- method for modeling rock fracture propagation and coalescence under fluid
tration system around fault environment. Pure and Applied Geophysics pressurization without remeshing. Rock Mechanics and Rock Engineering
2006;163(11):2585e93. 2014c;47(6):2183e98.
Li Q, Wu Z, Li X, Ohsumi T, Koide H. Numerical simulation on crust deformation due Pereira FLG, Roehl D, Paulo Laquini J, Oliveira MFF, Costa AM. Fault reactivation case
to CO2 sequestration in deep aquifers. Journal of Applied Mechanics 2002;5: study for probabilistic assessment of carbon dioxide sequestration. Interna-
591e600. tional Journal of Rock Mechanics and Mining Sciences 2014;71:310e9.
Liu J, Pereira GG, Liu Q, Regenauer-Lieb K. Computational challenges in the analyses Preisig M, Prevost JH. Coupled multi-phase thermo-poromechanical effects. Case
of petrophysics using microtomography and upscaling: a review. Computers & study: CO2 injection at In Salah, Algeria. International Journal of Greenhouse
Geosciences 2016;89:107e17. Gas Control 2011;5(4):1055e64.
Lucier A, Zoback M, Gupta N, Ramakrishnan T. Geomechanical aspects of CO2 Prevost JH. Dynaflow: a nonlinear transient finite element analysis program.
sequestration in a deep saline reservoir in the Ohio River Valley region. Envi- Princeton: Department of Civil Engineering, School of Engineering and Applied
ronmental Geosciences 2006;13(2):85e103. Science, Princeton University; 1981.
 P. Pan et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 936e947 947

Rabczuk T, Bordas S, Zi G. On three-dimensional modelling of crack growth using CO2 injection and storage. International Journal of Greenhouse Gas Control
partition of unity methods. Computers & Structures 2010;88(23e24):1391e411. 2010;4(5):827e39.
Rashid MM. The arbitrary local mesh replacement method: an alternative to Viegas G, Abercrombie RE, Kim WY. The 2002 M5 Au Sable Forks, NY, earthquake
remeshing for crack propagation analysis. Computer Methods in Applied Me- sequence: source scaling relationships and energy budget. Journal of
chanics and Engineering 1998;154(1e2):133e50. Geophysical Research: Solid Earth 2010;115(B7). http://dx.doi.org/10.1029/
Ringrose PS, Mathieson AS, Wright IW, Selama F, Hansen O, Bissell R, Saoula N, 2009JB006799.
Midgley J. The In Salah CO2 storage project: lessons learned and knowledge Vilarrasa V, Bolster D, Olivella S, Carrera J. Coupled hydromechanical modeling of
transfer. Energy Procedia 2013;37:6226e36. CO2 sequestration in deep saline aquifers. International Journal of Greenhouse
Rodriguez AA, Klie H, Sun S, Gai X, Wheeler MF, Florez H. Porous media upscaling of Gas Control 2010;4(6):910e9.
hydraulic properties: full permeability tensor and continuum scale simulations. Vilarrasa V, Carrera J, Olivella S. Hydromechanical characterization of CO2 injection
In: SPE/DOE Symposium on Improved Oil Recovery. Society of Petroleum En- sites. International Journal of Greenhouse Gas Control 2013;19:665e77.
gineers; 2006. Vitek V. Plane strain stress intensity factors for branched cracks. International
Rohmer J, Bouc O. A response surface methodology to address uncertainties in cap Journal of Fracture 1977;13(4):481e501.
rock failure assessment for CO2 geological storage in deep aquifers. Interna- Vorobiev O. Discrete and continuum methods for numerical simulations of non-
tional Journal of Greenhouse Gas Control 2010;4(2):198e208. linear wave propagation in discontinuous media. International Journal for
Rohmer J, Seyedi DM. Coupled large scale hydromechanical modelling for caprock Numerical Methods in Engineering 2010;83(4):482e507.
failure risk assessment of CO2 storage in deep saline aquifers. Oil & Gas Science Wang M, Feng YT, Wang CY. Numerical investigation of initiation and propagation
and Technology e Revue d’IFP Energies nouvelles 2010;65(3):503e17. of hydraulic fracture using the coupled Bonded Particle-Lattice Boltzmann
Rozhko AY. Role of seepage forces on seismicity triggering. Journal of Geophysical Method. Computers & Structures 2016. http://dx.doi.org/10.1016/
Research: Solid Earth 2010;115(B11). http://dx.doi.org/10.1029/2009JB007182. j.compstruc.2016.02.014.
Rutqvist J. The geomechanics of CO2 storage in deep sedimentary formations. Wang W, Kolditz O. Object-oriented finite element analysis of thermo-hydro- me-
Geotechnical and Geological Engineering 2012;30(3):525e51. chanical (THM) problems in porous media. International Journal for Numerical
Rutqvist J, Birkholzer JT, Tsang CF. Coupled reservoir-geomechanical analysis of the Methods in Engineering 2007;69(1):162e201.
potential for tensile and shear failure associated with CO2 injection in multi- Wei X, Li Q, Li X, Niu Z. Modeling the hydromechanical responses of sandwich
layered reservoir-caprock systems. International Journal of Rock Mechanics and structure faults during underground fluid injection. Environmental Earth Sci-
Mining Sciences 2008;45(2):132e43. ences 2016a;75(16):1155. http://dx.doi.org/10.1007/s12665-016-5975-9.
Rutqvist J, Cappa F, Rinaldi AP, Godano M. Modeling of induced seismicity and Wei X, Li Q, Li X, Sun Y. Impact indicators for caprock integrity and induced seis-
ground vibrations associated with geologic CO2 storage, and assessing their micity in CO2 geosequestration: insights from uncertainty analyses. Natural
effects on surface structures and human perception. International Journal of Hazards 2016b;81(1):1e21.
Greenhouse Gas Control 2014;24:64e77. Weng X, Kresse O, Cohen CE, Wu R, Gu H. Modeling of hydraulic-fracture-network
Rutqvist J, Tsang CF. A study of caprock hydromechanical changes associated with propagation in a naturally fractured formation. SPE Production & Operations
CO2-injection into a brine formation. Environmental Geology 2002;42(2):296e 2011;26(4):368e80.
305. White JA, Foxall W. Assessing induced seismicity risk at CO2 storage projects: recent
Rutqvist J, Vasco DW, Myer L. Coupled reservoir-geomechanical analysis of CO2 progress and remaining challenges. International Journal of Greenhouse Gas
injection and ground deformations at In Salah, Algeria. International Journal of Control 2016;49:413e24.
Greenhouse Gas Control 2010;4(2):225e30. Wiprut D, Zoback MD. Fault reactivation and fluid flow along a previously dormant
Rutqvist J, Wu YS, Tsang CF, Bodvarsson G. A modeling approach for analysis of normal fault in the northern North Sea. Geology 2000;28(7):595e8.
coupled multiphase fluid flow, heat transfer, and deformation in fractured Yamamoto K, Takahashi K. Importance of the geomechanics for the safety of CO2
porous rock. International Journal of Rock Mechanics and Mining Sciences geologic sequestration. In: Proceedings of the 3rd Asian Rock Mechanics
2002;39(4):429e42. Symposium. Kyoto, Japan. Rotterdam: Millpress; 2004.
Saffer DM, Marone C. Comparison of smectite- and illite-rich gouge frictional Zeng J, Li H, Zhang D. Numerical simulation of proppant transport in hydraulic
properties: application to the updip limit of the seismogenic zone along sub- fracture with the upscaling CFD-DEM method. Journal of Natural Gas Science
duction megathrusts. Earth and Planetary Science Letters 2003;215(1e2):219e and Engineering 2016;33:264e77.
35. Zhang R, Winterfeld PH, Yin X, Xiong Y, Wu YS. Sequentially coupled THMC model
Scholz CH. Earthquakes and friction laws. Nature 1998;391(6662):37e42. for CO2 geological sequestration into a 2D heterogeneous saline aquifer. Journal
Selvadurai APS. Heave of a surficial rock layer due to pressures generated by of Natural Gas Science and Engineering 2015a;27:579e615.
injected fluids. Geophysical Research Letters 2009;36(14). http://dx.doi.org/ Zhang Y, Langhi L, Schaubs PM, Piane CD, Dewhurst DN, Stalker L, Michael K.
10.1029/2009GL038187. Geomechanical stability of CO2 containment at the South West Hub Western
Soltanzadeh H, Hawkes CD. Assessing fault reactivation tendency within and sur- Australia: a coupled geomechanical-fluid flow modelling approach. Interna-
rounding porous reservoirs during fluid production or injection. International tional Journal of Greenhouse Gas Control 2015b;37:12e23.
Journal of Rock Mechanics and Mining Sciences 2009;46(1):1e7. Zheng L, Apps JA, Zhang Y, Xu T, Birkholzer JT. On mobilization of lead and arsenic in
Sorai M, Lei X, Nishi Y, Ishido T, Nakao S. CO2 geological storage. In: Handbook of groundwater in response to CO2 leakage from deep geological storage. Chemical
climate change mitigation and adaptation. Springer; 2015. p. 1e54. Geology 2009;268(3e4):281e97.
Stanchits S, Mayr S, Shapiro S, Dresen G. Fracturing of porous rock induced by fluid
injection. Tectonophysics 2011;503(1e2):129e45.
Stazi FL, Budyn E, Chessa J, Belytschko T. An extended finite element method with
higher-order elements for curved cracks. Computational Mechanics 2003;31(1): Dr. Pengzhi Pan obtained his B.S. and M.S. degrees in
38e48. Engineering Mechanics and Solid Mechanics from Wuhan
Streit JE, Hillis RR. Estimating fault stability and sustainable fluid pressures for University of Technology, Ph.D. degree in Rock Engineering
underground storage of CO2 in porous rock. Energy 2004;29(9e10):1445e56. from the Institute of Rock and Soil Mechanics (IRSM),
Sukumar N, Chopp DL, Moës N, Belytschko T. Modeling holes and inclusions by level Chinese Academy of Sciences (CAS) in 2006. Then he
sets in the extended finite-element method. Computer Methods in Applied worked at IRSM as an Assistant Professor, and was pro-
Mechanics and Engineering 2001;190(46e47):6183e200. moted to Associate Professor in 2009, and Professor in
Sun AY, Zeidouni M, Nicot JP, Lu Z, Zhang D. Assessing leakage detectability at 2013. In 2011e2012, he worked at Lawrence Berkeley
geologic CO2 sequestration sites using the probabilistic collocation method. National Laboratory (LBNL) as a Visiting Scholar in the
Advances in Water Resources 2013;56:49e60. modeling of coupled thermo-hydro-mechano-chemical
Torres SAG, Castaño JDM. Simulation of the hydraulic fracture process in two di- (THMC) processes in geological media. His research is
mensions using a discrete element method. Physical Review E 2007;75(6): currently focused on continuumediscontinuum numerical
066109. methods to simulate rock nonlinear fracturing process
Urpi L, Rinaldi AP, Rutqvist J, Cappa F, Spiers CJ. Dynamic simulation of CO2-injec- with and without consideration of coupled THMC pro-
tion-induced fault rupture with slip-rate dependent friction coefficient. Geo- cesses in geological media. He developed a series of comprehensive successive numer-
mechanics for Energy and the Environment 2016;7:47e65. ical codes (e.g. EPCA , EPCA3D, RDCA, TOUGH-RDCA) with a combination of multi-
2D

Vidal-Gilbert S, Nauroy JF, Brosse E. 3D geomechanical modelling for CO2 geologic discipline and theories. The codes have been applied to a wide range of geomechanics
storage in the Dogger carbonates of the Paris Basin. International Journal of and geotechnical engineering, including the stability analysis of subsurface rock engi-
Greenhouse Gas Control 2009;3(3):288e99. neering, geological disposal of high-level nuclear waste and geological sequestration of
Vidal-Gilbert S, Tenthorey E, Dewhurst D, Ennis-King J, Van Ruth P, Hillis R. Geo- CO2, etc., to understand the underlying failure mechanism and coupling process in
mechanical analysis of the Naylor field, Otway Basin, Australia: implications for complex geological systems.