Tectonophysics 690 (2016) 63–75

Contents lists available at ScienceDirect

Tectonophysics

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

The impact of in-situ stress and outcrop-based fracture geometry on

hydraulic aperture and upscaled permeability in fractured reservoirs
Kevin Bisdom a,⁎, Giovanni Bertotti a, Hamidreza M. Nick a,b
a
Department of Geoscience & Engineering, Delft University, Delft, Netherlands
b
Technical University of Denmark, The Danish Hydrocarbon Research and Technology Centre, Copenhagen, Denmark

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

Article history: Aperture has a controlling impact on porosity and permeability and is a source of uncertainty in modeling of nat-
Received 24 October 2015 urally fractured reservoirs. This uncertainty results from difficulties in accurately quantifying aperture in the sub-
Received in revised form 25 March 2016 surface and from a limited fundamental understanding of the mechanical and diagenetic processes that control
Accepted 3 April 2016
aperture. In the absence of cement bridges and high pore pressure, fractures in the subsurface are generally con-
Available online 8 April 2016
sidered to be closed. However, experimental work, outcrop analyses and subsurface data show that some frac-
Keywords:
tures remain open, and that aperture varies even along a single fracture. However, most fracture flow models
Aperture consider constant apertures for fractures. We create a stress-dependent heterogeneous aperture by combining
Naturally fractured reservoirs Finite Element modeling of discrete fracture networks with an empirical aperture model. Using a modeling ap-
Equivalent permeability proach that considers fractures explicitly, we quantify equivalent permeability, i.e. combined matrix and
Fracture geometry stress-dependent fracture flow. Fracture networks extracted from a large outcropping pavement form the
Discrete fracture networks basis of these models. The results show that the angle between fracture strike and σ1 has a controlling impact
on aperture and permeability, where hydraulic opening is maximum for an angle of 15°. At this angle, the fracture
experiences a minor amount of shear displacement that allows the fracture to remain open even when fluid
pressure is lower than the local normal stress. Averaging the heterogeneous aperture to scale up permeability
probably results in an underestimation of flow, indicating the need to incorporate full aperture distributions
rather than simplified aperture models in reservoir-scale flow models.
© 2016 Elsevier B.V. All rights reserved.

1. Introduction
Naturally fractured reservoirs are thought to have a large potential
for increased hydrocarbon recovery (Nelson, 2001). This potential is,
however, largely untouched partly because of the difficulty of predicting
flow along complex fracture networks such as those occurring in nature
(Berkowitz, 2002). The difficulty mainly lies in the heterogeneous sub-
seismic-scale characteristics of fractures, which can only be partially
sampled by core data or image logs (National Research Council, 1996;
Wu and Pollard, 2002; Laubach, 2003). Therefore, outcrops are often
used to better characterize the fracture spatial distribution, including
length, height, orientation, spacing and aperture (Bonnet et al., 2001;
Chesnaux et al., 2009; Agosta et al., 2010; Guerriero et al., 2010;
Wilson et al., 2011; Hooker et al., 2013, 2014).
Out of these parameters, the fracture aperture distribution is one of
the main factors controlling flow, as aperture defines fracture porosity
and permeability (National Research Council, 1996; Guerriero et al.,
2013). A wide range of studies, mostly based on outcrops where the
⁎ Corresponding author at: Department of Geoscience & Engineering, Delft University of
Technology, Stevinweg 1, 2628CN Delft, Netherlands.
E-mail address: k.bisdom@tudelft.nl (K. Bisdom).
aperture distribution can be studied in full, have shown that aperture
size varies within fracture orientation sets and along the length or
height of individual fractures (Laubach and Ward, 2006; Hooker et al.,
2012, 2013, 2014; Iñigo et al., 2012). The most common observation
in these studies is that the aperture-size distribution is best described
by power-law scaling (Hooker et al., 2014). These outcrop observations
have also been observed in subsurface datasets (Laubach, 2003; Hooker
et al., 2009; Becker et al., 2010).
Stress, in the form of relatively high fluid pressures, can drive gener-
ation and propagation of cracks in the subsurface (Atkinson, 1987), but
the propagation of fractures in the absence of high fluid pressures is
most likely driven by a coupled process of stress and cement precipita-
tion (Laubach et al., 2004a; Alzayer et al., 2015). Moreover, partial
cementation, and particularly the occurrence of cement bridges, is cru-
cial in ensuring that fractures remain hydraulically open, even if fluid
pressure is low (Laubach et al., 2004b). By modeling cement precipita-
tion, the rate of fracture growth or propagation can be quantified, to
better understand how fracture networks grow (Philip et al., 2005;
Lander and Laubach, 2015).
Although models and outcrop descriptions of heterogeneous
apertures along single fractures exist, they are only rarely included in
Discrete Fracture Network (DFN) models, which typically consider a

http://dx.doi.org/10.1016/j.tecto.2016.04.006
0040-1951/© 2016 Elsevier B.V. All rights reserved.
64 K. Bisdom et al. / Tectonophysics 690 (2016) 63–75

constant aperture per fracture or even per orientation set, with some describing how the aperture model is implemented into the Finite Ele-
notable exceptions (Philip et al., 2005; Olson et al., 2009; Nick et al., ment (FE) modeling workflow.
2011; de Dreuzy et al., 2012; Lei et al., 2014). Using a heterogeneous
aperture distribution derived from the local stress acting on a fracture
surface, we aim to illustrate the impact of a heterogeneous versus ho- 2.1. Barton–Bandis aperture model
mogeneous aperture distribution on permeability. The relation between
stress and aperture is quantified using the Barton–Bandis method, The Barton–Bandis method is based on an initial aperture, which is a
which has been shown to produce highly heterogeneous aperture dis- function of fracture roughness (JRC) and strength (JCS) (Barton and
tributions (Lei et al., 2014). Bandis, 1980). An overview of all variables and their meaning is present-
The Barton–Bandis model is an empirical approach that quantifies ed in Table 1. The initial aperture is defined by E0 = JRC/50.
the aperture that remains when irregular mismatching fracture walls By applying in-situ compression, the initial aperture decreases,
are partially closed under compression (Barton, 1982; Bandis et al., whereby part of the fracture may remain open by poorly interlocking ir-
1983). Whereas the critical stress model used for faults and fractures re- regular fracture walls (Fig. 1). The resulting physical or mechanical ap-
quires high pore pressures, such that the stress within the fracture is erture is a function of normal, i.e. perpendicular to two sub-parallel
close to the least principle horizontal stress (Barton et al., 1995; fracture walls, and shear-related opening. In this study, we focus on nor-
Rogers, 2003; Zoback, 2007), the Barton–Bandis model predicts that mal aperture En and the shear-related hydraulic aperture e, which is the
fractures can be hydraulically open in the absence of high fluid pres- opening that effectively contributes to fluid flow under given in-situ
sures (Olsson and Barton, 2001). Barton–Bandis furthermore takes stress conditions, controlled by the amount of shear-induced dilation.
into account horizontal stress anisotropy during production and the Normal and hydraulic apertures are defined by (Bisdom et al., in press):
subsequent heterogeneous flow behavior along fracture walls, where
local stress conditions may prevent flow along some fractures (Olsson

and Barton, 2001; Matsuki et al., 2008). This model does not consider 1 K −1
En ¼ E 0 − þ ni ð1Þ
the impact of diagenesis, and hence it may not be representative for vm σ n
chemically reactive rocks such as carbonates, but it has been successful-
ly applied to model permeability in shales (Barton, 2014). 8
The aperture distribution and subsequent fluid flow through frac- >
> En 2 us
< for ≤0:75
tures predicted by the Barton–Bandis method has been studied before, e¼ JRC 2:5 upeak ð2Þ
> p ffiffiffiffiffi u
but mainly in synthetic fracture networks with simplified geometries >
: En JRC mob for s ≥1
or outcrop-based models with small dimensions (Nemoto et al., 2009; upeak
Tao et al., 2009; Lei et al., 2014). We apply this aperture method to
models of fractured rocks under in-situ stress conditions in 2-D horizon-
tal cross-sections of natural fracture networks affecting bodies of up to The us/upeak term quantifies the shear displacement as a function of a
360 m across. These fracture networks are digitized from outcropping peak shear displacement, which depends mainly on the block size of a
fracture pavements in central Tunisia which display well resolved fracture. The block size L is the spacing between fractures that intersect
fracture geometries. the fracture of interest (Barton, 1982). The domain in between 0.75≤us/
The impact of hydraulic aperture on fluid flow is modeled using a upeak ≤1.0 is interpolated linearly (Olsson and Barton, 2001). The normal
hybrid Finite-Element Finite-Volume (FEFV) approach that models stress σn and shear displacement u required for calculating aperture in
single-phase incompressible flow through explicit fractures, as well the above equations are quantified using FE models.
as the flow exchange between fractures and matrix (Matthäi and
Belayneh, 2004; Matthäi et al., 2007; Paluszny et al., 2007). Using this Table 1
integrated workflow, we quantify the impact on aperture and perme- Overview of parameters.
ability of: i) different fracture geometrical parameters, ii) variations in Symbol Definition Units Constant
the magnitude and direction of horizontal principle stresses and iii)
JRC Joint Roughness Coefficient – 15.0
rock properties. JCS Joint Compressive Strength MPa 120
The last section of this paper focuses on upscaling. The Barton– E0 Initial ‘unstressed’ aperture mm 0.3
Bandis method produces heterogeneous aperture distributions even En Normal aperture mm
along single fractures, while reservoir-scale fracture-flow models us Shear displacement mm
upeak Peak shear displacement mm
generally assume a constant aperture per fracture or sometimes even L Block size m
per fracture set. Models with single apertures per fracture found that a e and e Hydraulic aperture (and averaged mm
single equivalent aperture can be defined, that yields the same result hydraulic aperture)
as a heterogeneous aperture distribution (Nick et al., 2011). We analyze vm Maximum closure mm
Kni Initial stiffness MPa/mm
whether such an aperture can still be derived for an aperture distribu-
E Young's Modulus GPa
tion that varies even along a single fracture. Secondly, we study whether ν Poisson's ratio –
permeability calculated for small-scale (i.e. single reservoir grid cells) σn, σ1 and σ3 Normal, maximum and minimum MPa
models accurately predicts the permeability of a larger-scale model horizontal stress
with heterogeneous apertures. δij Kronecker delta –
ε Strain –
β Angle between fracture strike and σ1 °
τ Shear stress MPa
2. Stress-induced Barton–Bandis aperture modeling c Cohesion MPa
ϕ Internal friction angle °
q Darcy flow m/s
The Barton–Bandis method defines aperture based on fracture me-
k, keq, km, kf Permeability (equivalent, matrix, fracture) m2 or mD
chanical properties and the local shear and normal stresses (Barton s 2D fracture spacing m
and Choubey, 1977; Barton and Bandis, 1980; Bandis et al., 1983; A Fractured rock area m2
Barton et al., 1985). The full set of equations used to translate stress T Total length of fractures m
into aperture is discussed in Bisdom et al. (in press). Here, we provide d0 Dimension parallel to main fracture trend –
le Element length m
a brief recap of the main functions defining aperture, followed by
 K. Bisdom et al. / Tectonophysics 690 (2016) 63–75
Fig. 1. Overview of the parameters used to calculate the normal and hydraulic apertures: a) the initial aperture which is laterally consistent and a function of fracture roughness and
strength; b) following the application of in-situ stress, the fracture experiences limited shear displacement and partial closure.
2.2. Governing equations
Numerical models such as FE or Finite-Element Discrete-Element
(FDEM) have proven to be efficient for modeling different aspects of
fracture behavior, including growth and opening (e.g. Latham et al.,
2013; Lei et al., 2014). Many solutions exist in both the commercial
and academic domain for FE modeling of fractured media (Paluszny
and Zimmerman, 2011). In this study we use the commercial ABAQUS
code, which has been proven effective in modeling of geological discon-
tinuities at a small (i.e. fracture) scale (e.g. Smart et al., 2014).
We consider plane-strain models with a minimum and maximum
horizontal stress component applied to the boundaries of the system.
The local stress state inside the body is solved taking into account
non-linear behavior resulting from the fracture interactions. The
stress–strain relationship of a multi-dimensional linear elastic material
is written as (Dassault Systemes, 2013):
E Eν
σ ij ¼ εij þ δij εkk 
1þν ð1 þ νÞð1−2νÞ
The normal stress acting on each fracture is calculated from its bor-
dering elements, where for a single fracture in a fully elastic medium
normal stress is defined as (e.g. Zoback, 2007):
σ n ¼ 0:5ðσ 1 þ σ 3 Þ þ 0:5ðσ 1 −σ 3 Þ cos 2 β
We consider elastic and elastoplastic models. For elastoplastic mate-
rials a linear Mohr–Coulomb criterion is used (Dassault Systemes,
2013):
τ ¼ c−σ n tan ϕ
The transition from elastic to plastic deformation is defined by the
cohesion yield stress. Non-linear Mohr–Coulomb behavior may be con-
sidered as more representative of porous reservoir rocks that experi-
ence significant changes in stress during production (Barton, 1976,
2014), but the change in aperture during production is not considered
in our models. Furthermore, it is assumed that no fracture growth or ini-
tiation of new fractures occurs, as the main focus is to analyze the
change in opening within a fracture as a function of stress.
2.3. Model setup
The FE models consist of an unstructured mesh with triangular
quadratic elements. The average element size is 0.01% of the smallest
dimension of the model, resulting in approximately 60,000 elements
65
for a 50 × 50 m model, depending on fracture intensity. In all models,
we introduce a buffer zone around the fractured mesh to avoid bound-
ary effects. The validity of this mesh size is tested by looking at the
discretization error as a function of the resulting hydraulic aperture.
Fractures are represented by splitting the mesh along fracture lines, cre-
ating double nodes along the fracture surface, resulting in seams in the
mesh, that are able to open or close (Fig. 2). For each of the fracture
nodes, the normal stress and shear displacement are calculated in
ABAQUS.
For the rock matrix, we obtain solutions for elastic and elastoplastic
Mohr–Coulomb failure behavior. The elastic properties are measured
from samples of the analyzed outcrops and include bulk rock density,
dry porosity, and seismic velocities, which are used to calculate the
Young's Modulus and Poisson's ratio. To define the linear Mohr–
Coulomb failure criterion, typical values from the literature representa-
tive of the studied rock are used (Smart et al., 2010).
ð3Þ
2.4. Single-fracture model behavior
To illustrate the change in aperture modeled by this FE Barton–
Bandis approach, a tectonic stress is applied to the 2-D model in one di-
rection. As the body is limited and no-displacement conditions are ap-
plied, representing a body of rock in the subsurface confined within a
ð4Þ larger unit, the tectonic stress (i.e. σ1) applied in the y-direction results
in a Poisson's stress (σ3) in the x-direction (Fig. 3). Throughout this
study, we assume rough fractures with a JRC of 15, resulting in an initial
aperture of 0.3 mm.
For this example model, with a σ1 of 30 MPa and a σ3 of 10 MPa, the
normal aperture of a single fracture decreases with increasing normal
ð5Þ stress, whereas hydraulic aperture increases (Fig. 4a,b). The aperture
distribution is sensitive to model orientation, as a fracture oblique to
σ1 experiences more shear, coinciding with a decrease in normal stress
(Fig. 4c). This has a negative impact on normal aperture but positive im-
pact on hydraulic aperture.
3. Modeling aperture and permeability in natural fracture networks
We apply the described workflow for heterogeneous aperture
modeling to a large and complex fracture system, analyzing the rela-
tions between fracture geometry, different mechanical properties and
boundary conditions, and hydraulic aperture and permeability. For ge-
ometry, we consider commonly used parameters such as spacing and
size, as well as connectivity (Berkowitz and Balberg, 1993; Berkowitz,
2002).
66 K. Bisdom et al. / Tectonophysics 690 (2016) 63–75
Fig. 2. Modeling fractures as seams in a Finite Element mesh: a) triangular mesh with 2 intersecting fracture seams (thick black line elements). Nodes along these line elements are
duplicated to create seams; b) resulting deformed mesh after (exaggerated) opening of the two seams, showing normal opening (En), shear displacement (Δus) and local normal
stress (σn), as well as how often each fracture node is duplicated to create the seams.
3.1. Regional setting and geometry of the fracture network
We extract the fracture network used for modeling from an outcrop-
ping carbonate pavement that is part of the Alima fold (Gafsa basin, cen-
tral Tunisia; Fig. 5a) (Riley et al., 2010; Saïd et al., 2011). This basin
contains several E–W trending folds with excellent exposures of frac-
tures. The Alima fold is an E–W striking anticline with a length of
20 km and an outcropping width of 7 km (Fig. 5b). The anticline formed
during N–S regional shortening starting in the Eocene (Fig. 5c) (Riley
et al., 2010). The southern flank has been steepened to 60–70°, resulting
in a large well-preserved pavement of carbonates of the Eocene Kef
Eddour Fm. The width of the pavement is 360 m, the maximum height
(perpendicular to the fold axis) is 70 m at the center, decreasing to 30 m
at the sides (Fig. 6a).
Fractures in the pavement are digitized on georeferenced outcrop
images (Hardebol and Bertotti, 2013), resulting in 345 fractures inclu-
sive of their length, strike, and spacing (Table 2; Bisdom, 2015). The da-
tabase is supplemented with observations made at the outcrop, which
indicate that fractures are generally bed-perpendicular.
The minimum fracture length captured is 1 m. Smaller fractures are
present, but their distribution cannot be fully defined as they are below
the image resolution, resulting in a truncated distribution (Ortega et al.,
2006). Maximum measured fracture length is equal to the smallest di-
mension of the outcrop, resulting in censoring of the true maximum
length (Ortega et al., 2006). However, the Barton–Bandis method de-
fines aperture as a function of block size rather than total fracture
Fig. 3. Set-up of the numerical mechanical models with a single fracture. Maximum
horizontal compression σ1 is applied in the y-direction, which exerts a Poisson's stress
σ3 in the x-direction. The rock matrix is fully elastic with a Young's Modulus of 50 GPa
and a Poisson's ratio of 0.3.
length, so this method suffers less from censoring artifacts. The average
2-D fracture spacing s in meters is calculated using (Wu and Pollard,
2002):
A
s¼ ð6Þ
T þ d0
The resulting spacing varies between 3.3 and 4.9 m in different parts
of the outcrop, with an average of 3.8 m.
Low-angle stylolites attributed to early folding are crosscutting bed-
perpendicular fractures, indicating that fracturing occurred when layers
were horizontal. Therefore, the pavement and its fractures are rotated
back to the horizontal pre-folding situation (Fig. 6b-c). The resulting
dataset consists of two bed-perpendicular orientation families, striking
on average NNW–SSE (Set 1) and ENE–WSW (Set 2), with a scatter of
40° within each set (Fig. 6c). The geometry per set is summarized in
Table 2.
From the outcrop model, we extract four 50 × 50 m sub-models
(Alima1–4; Fig. 6d) which are used for the analysis of the impact of frac-
ture geometry and stress conditions on aperture. All sub-models contain
comparable N–S and E–W striking fractures, but the number of fractures
varies (Table 2). There are no nearby faults, nor any changes in the li-
thology along the pavement, so it is assumed that the variations in frac-
ture intensity are representative of the natural variability of a fracture
network (Bisdom et al., 2014). Aperture and permeability are calculated
for all fractures, i.e. considering both sets.
3.2. Model set-up
3.2.1. Mechanical FE model
We model aperture in the natural fracture networks using the same
fracture and stress conditions as for the basic model presented in Fig. 3,
with an initial aperture of 0.3 mm for all fractures. Model set-up, rock
properties and boundary conditions are also the same as those in
Fig. 3. When modeling the 50 × 50 m sub-models, we add a buffer
zone extending the models to 100 × 100 m to avoid boundary effects in-
terfering with the fractures.
To quantify the uncertainty in aperture modeling, we consider a
range of stresses and rock properties as input. Although the regional
strike of σ1 during fracturing was N–S (Bouaziz et al., 2002), we consider
also the impact of σ1 on aperture and permeability in the E–W direction.
The x-axis of the models is set perpendicular to the N–S direction. For
both σ1 directions, the maximum compressional stress varies between
1 and 80 MPa.
 K. Bisdom et al. / Tectonophysics 690 (2016) 63–75 67

Fig. 4. Relation between normal opening (En) and hydraulic aperture (e) for a fracture that is striking parallel to the direction of σ1, compared to that of a fracture striking at 30°: a) normal
aperture; b) hydraulic aperture; c) average normal and hydraulic aperture as a function of orientation.

3.2.2. Coupled fracture-matrix permeability model
To quantify the impact of hydraulic aperture on flow, equivalent
permeability is modeled, which is a function of combined matrix and
fracture permeability (Matthäi and Nick, 2009; Bisdom et al., in press).
Equivalent permeability describes the ability of the fractured formations
to permit the flow of fluid through both the porous rock matrix and
open fractures in each direction from one boundary to the opposite
one, assuming no-flow boundary conditions at the other two bound-
aries parallel to the far field pressure gradient (e.g. Matthäi and
Belayneh, 2004). It is calculated using the steady state continuity equa-
tion ∇ ⋅ q = 0, which is solved for pressure using Darcy flow in a hybrid
FEFV approach implemented in the Complex System Modeling Platform
(CSMP++; Matthäi et al., 2007). A far field pressure gradient of 10kPa
is applied in either the N–S or E–W direction of the model, after which
the steady state equation is solved (Paluszny and Matthäi, 2010; Nick
and Matthäi, 2011).
Fracture permeability is calculated from hydraulic aperture at each
node in the mesh using the cubic law (Snow, 1969; Fig. 1 in Nick
et al., 2011). The potential contribution from disconnected fractures is
captured by including a matrix permeability of 10 mD. To limit the
number of variables in the models, matrix permeability km is assumed
constant. The contrast between matrix and fracture permeability can
impact equivalent permeability, but this is addressed elsewhere
(Matthäi and Belayneh, 2004; Bisdom et al., in press). Equivalent
permeability for both the N–S and E–W directions is calculated by
integrating the fluid flux across the model boundaries (Matthäi
and Nick, 2009).
4. Results
4.1. The impact of fracture network geometry
We compare the relation between equivalent permeability and frac-
ture attributes including intensity, size, orientation and connectivity
(Table 2). The geometrical analysis is done for elastic models with a
constant compression of 30 MPa in N–S (Fig. 7a) and E–W directions
(Fig. 7b). Under these constant stress conditions, there are large differ-
ences in permeability between the different sub-models, related to
geometry.

Fig. 5. Regional setting of the Alima anticline in the Gafsa basin in central Tunisia: a) Structural setting of the Gafsa basin, adapted from Bouaziz et al. (2002); b) The E–W striking Alima fold
(backdrop from Bing maps, September 2014); c) Stratigraphic chart of the Late Cretaceous to present-day compiled from Bouaziz et al. (2002) and Zouaghi et al. (2009).
68 K. Bisdom et al. / Tectonophysics 690 (2016) 63–75

Fig. 6. Original outcrop model with interpreted fractures, from which the 4 Alima sub-models are extracted: a) digitized outcrop model of the steeply dipping pavement in the Alima
anticline with the position of the four outcrop windows; b) lower hemisphere stereoplot of the 345 fracture orientations (red) and bedding (blue); c) stereoplot with back-rotated
fracture orientations; and d) the four fracture models.

4.1.1. Impact of spacing 4.1.2. Impact of connectivity

Spacing has no strong impact on equivalent permeability, as for a N–
S oriented σ1, the outcrops with the smallest spacing (i.e. highest inten-
sity) have the lowest fracture network permeability (Alima1), whereas
the model with the largest spacing (Alima3) has the highest equivalent
permeability (Fig. 7a,c). The spacing distributions are similar to each
other in the different sub-models, with relatively small variations in av-
erage aperture (except for Alima3), although variations in permeability
are large.
In terms of spacing per orientation set, the N–S system has a larger
spacing than the E–W system (Table 2), but the higher-intensity E–W
system has a lower permeability, even when compression is in E–W di-
rection (Fig. 7b).
Fracture connectivity does not provide a representative description
of equivalent permeability, as the model with the lowest connectivity,
expressed as the number of fracture intersections, has the highest
permeability for a N–S oriented σ1 (Alima3 in Fig. 7a). Alima1, which
has the highest connectivity, has the lowest permeability when σ1 is
in N–S direction. Models Alima2 and Alima4 do have high permeabilities
as well as a high connectivity, but it is still lower than that of Alima3,
which has 50% fewer intersections compared to the other models.
With the exception of Alima4, three of the four models have perco-
lating pathways of connected fractures (Fig. 6c and Table 2). However,
in terms of hydraulic aperture, there is no fully connected pathway in
any of the models, so the connected fractures in Alima4 do not form a
percolating pathway in terms of fluid flow.

Table 2
Summary of fracture geometry. 4.1.3. Impact of fracture length
The large contrast in N–S versus E–W equivalent permeability is
ALIMA1 ALIMA2 ALIMA3 ALIMA4
mainly related to the maximum and average fracture lengths (Table 2
# fset 1 19 20 9 22 and Fig. 7d). Average fracture length of the N–S trending system is
spacing fset 1 [m] 6.8 604 9.2 9.2
length fset 1 [m] 13.2 15.6 18.5 14.8
more than 50% larger than that of the E–W fracture system (Table 2).
# fset 2 50 38 34 52 The longest fractures in each model are part of the N–S system. More-
spacing fset 2 [m] 5.6 7.2 8.8 5.1 over, when ranking the outcrops from large to small average length of
length fset 2 [m] 7.8 9.0 8.8 5.1 N–S sets, the ranking is identical to the ranking of N–S equivalent
# total 69 58 43 74
permeability (Fig. 7a).
spacing [m] 3.3 3.6 4.9 3.5
average length [m] 9.5 11.6 10.1 9.0 Average and maximum length better describe the resulting equiva-
maximum length [m] 42.2 48.5 37.7 28.5 lent permeability trends, compared to spacing, even though spacing is
# intersections 48 41 16 43 also defined as a function of length through Eq. (6). However, this equa-
N–S pathways 1 1 1 0 tion does not distinguish between many small fractures and a few very
E–W pathways 1 0 0 0
large fractures.
 K. Bisdom et al. / Tectonophysics 690 (2016) 63–75 69

Fig. 7. Equivalent permeability in N–S (black bars) and E–W directions (gray bars) for each outcrop model as a function of maximum horizontal stress oriented in: a) N–S direction; b) E–W
direction. The kf value indicated for each plot is the average fracture permeability calculated using the cubic law; c) fracture spacing distribution with the average for each sub-model;
d) fracture length and blocksize distributions for each sub-model, with the average length indicated in each plot.

4.1.4. Impact of orientation Alima3. Fracture orientation, however, does have a positive impact on
Although length correlates to some extent with equivalent perme- fluid flow in this model. Specifically, the fracture strike with respect to
ability, it does not explain the relatively high N–S permeability in the direction of σ1 is key for hydraulic aperture and subsequent

Fig. 8. Fracture orientation per model, highlighting in black all fractures that strike at an angle of 10 to 30° with respect to the direction of maximum horizontal stress. The corresponding
hydraulic aperture distribution is plotted under each model, with the percentage of hydraulically open vs. closed fracture segments.
70 K. Bisdom et al. / Tectonophysics 690 (2016) 63–75
permeability. The average strike in Alima3 is 17°, which corresponds ap-
proximately to the ideal angle between shortening direction and strike
as found in Fig. 4c. The other three Alima models contain more fractures,
but a smaller sub-set of these fractures is oriented ideally with respect
to the shortening direction (Fig. 8). As a result of this orientation distri-
bution, and to a lesser extent block size, most of these fractures are
hydraulically closed, especially in Alima1 and Alima2.
4.2. Impact of stress
We quantify the change in aperture as a function of increasing
regional stress σ1 and changing stress orientation (N–S vs. E–W). In
the single fracture model it was shown that an increase in stress may
lead to an increase in permeability, depending on whether the fracture
experiences shear stress (Fig. 4). Fractures in the Alima models have
heterogeneous aperture and permeability distributions, but clear trends
can still be distinguished (Fig. 9).
From the curves in Fig. 9 we can identify three specific equivalent
permeability domains for N–S compression. These trends are similar
for E–W compression, but as the contribution of fractures to flow is
small, the domains are less easily observed. Focusing on N–S compres-
sion, the first domain (up to 5 MPa differential stress) is characterized
by a lack of change in permeability. Between 5 and 20 MPa, permeability
increases, and flattens again for differential stresses above 20 MPa. The
flattening results from the definition of hydraulic aperture, which
cannot exceed normal aperture.
A second factor strongly affecting aperture and permeability is the
direction of σ1. When σ1 is applied in the N–S direction, most E–W strik-
ing fractures experience maximum closure, resulting in lower E–W
equivalent permeability (Fig. 9a). However, when σ1 is in the E–W di-
rection, N–S permeability in three of the four models remains larger
than permeability in the E–W direction, indicating that the direction
of highest permeability is not only a function of stress direction, but
also of other factors, such as geometry (Fig. 9b).
The irregular increase in permeability, most notably in the domain
between 5 and 15 MPa differential stress, is primarily the result of the
relation between shear displacement and peak shear displacement.
The resulting shear ratio has a major impact on hydraulic aperture, as
a small change in the shear ratio may result in changing from one
hydraulic aperture domain to another, as determined by Eq. (2).
Fig. 9. Equivalent permeability in N–S direction (black curves) and E–W direction (gray curves) for each sub-model as a function of increasing stress: a) maximum horizontal stress in N–S
direction; b) maximum horizontal stress in E–W direction.
4.3. Impact of rock properties
Compressive stress and the subsequent aperture and permeability
distributions are functions of the mechanical behavior of the rock ma-
trix. We analyze a range of values for the Young's modulus and Poisson's
ratio, and compare the elastic model with an elastoplastic Mohr–
Coulomb model. In all models, the applied σ1 is constant (30 MPa).
We do not consider the effect of deformation on matrix porosity and
permeability, and assume that the fracture properties remain constant.
The JCS is also kept constant, although it is probably related to mechan-
ical rock properties such as the Young's Modulus, but the exact depen-
dency of JCS on other rock properties is unclear.
For an elastic rheology, equivalent permeability decreases with in-
creasing Young's modulus and Poisson's ratio, irrespective of the direc-
tion of σ1 (Fig. 10). A higher Young's Modulus results in a decrease in
shear displacement and shear-induced hydraulic aperture and flow.
The impact of the Poisson's ratio is the result of the fact that the lateral
boundaries of the model are fixed. A high Poisson's ratio in combination
with this boundary condition increases σ3, and decreases normal and
subsequent hydraulic opening. This effect is most clearly visible in
Fig. 10d, where for N–S shortening without the possibility of E–W ex-
tension, the E–W trending fractures are almost immediately closed
when the Poisson's ratio starts to increase, resulting in a large contrast
in permeability in the E–W direction versus the N–S direction.
Mohr–Coulomb plasticity lowers the shear displacement on fracture
surfaces and subsequently lowers the hydraulic aperture (Fig. 10c,f).
The relative impact of fracture permeability on equivalent permeability
decreases, resulting in smaller differences in permeability between the
sub-models. A detailed sensitivity analysis of the plastic parameters is
outside the scope of this study but our results suggest that models
with lower friction and dilation angles do not significantly change per-
meability. For the regional stress conditions in the studied models, the
elastic properties have a stronger impact on permeability than the
plastic properties. For a higher differential stress, the plastic domain
may become the controlling factor, and its properties may have a larger
impact relative to the elastic properties.
5. Towards reservoir-scale aperture and permeability models
The aperture distributions of the 50 × 50 m models give an indica-
tion of small-scale permeability, i.e. within a single reservoir simulation
 K. Bisdom et al. / Tectonophysics 690 (2016) 63–75
Fig. 10. Equivalent permeability in N–S direction (black curves) and E–W direction (gray curves) as a function of a changing Young's modulus (a–c) and a changing Poisson's ratio (plots d–
f). Figures a, c–d and f are modeled with σ1 oriented N–S, while (b) and (e) are modeled with σ1 oriented E–W. Models corresponding to a-b and d-e are fully elastic, with their base case
scenario properties listed in the legend, and c and f are modeled using a elastoplastic rock.
grid cell. Using explicit fracture modeling to construct a reservoir-scale
3-D fracture model with mechanically controlled heterogeneous aper-
ture distributions is a complex task (Geiger and Matthäi, 2012).
Alternatively, the high-resolution aperture and permeability, calcu-
lated per fracture node, of explicit fracture permeability models can be
upscaled through arithmetic or harmonic averaging (Jonoud and
Jackson, 2008; Cottereau et al., 2010). We test these methods in two
ways using our explicit fracture aperture and permeability models:
i) Within each 50 × 50 m model, we quantify whether permeability
can be characterized by a single average aperture instead of a heteroge-
neous distribution, and ii) we compare the average permeability of the
50 × 50 m models with the complete Alima model, to quantify whether
these small models, representing grid cells, can capture the permeabil-
ity of the complete model.
5.1. Heterogeneous vs. averaged aperture in 50 × 50 m models
For each of the four models, we calculate the weighted average
aperture:
X
e  le
e¼ X : ð7Þ
le
The average is a function of both open and closed fracture parts,
including those that are practically closed by the first part of Eq. (2),
and is assigned as a constant aperture to the entire fracture network.
The equivalent permeability resulting from the averaged aperture
71
distributions is modeled for elastic (N–S and E–W oriented σ1) and
elastoplastic (N–S oriented σ1) models (Fig. 11).
The average aperture yields an upward trend in permeability for in-
creasing differential stress that is similar to that of the heterogeneous
distributions, with Alima3 having the highest equivalent permeability
in N–S direction (Fig. 11a,b). The permeability in the E–W direction,
however, has changed significantly, and is similar to the N–S permeabil-
ity, as all fractures in all directions have the same aperture. Permeability
resulting from average aperture in an elastoplastic model is lower, sim-
ilar to that which was observed for heterogeneous apertures.
Compared to the heterogeneous aperture distributions, the averaged
distributions result in a decrease in N–S permeability versus a relatively
large increase in E–W permeability (Fig. 11). Although the maximum of
the averaged permeability is small compared to the heterogeneous
distribution, the average of N–S and E–W flow of the averaged aperture
is comparable to that of the heterogeneous aperture.
5.2. Upscaling equivalent permeability derived from heterogeneous aper-
ture distributions
The 50 × 50 m models are representative of the grid cell size of con-
ventional reservoir simulation grids. Existing upscaling approaches
commonly assume that the equivalent permeability of the sub-models
(i.e. grid cells) combined produces an accurate representation of the
flow in the entire model (Matthäi and Nick, 2009).
We compare the heterogeneous and averaged aperture and perme-
ability distributions of the four sub-models with the complete Alima
model, which contains 345 fractures in an area of 328 × 54 m
72 K. Bisdom et al. / Tectonophysics 690 (2016) 63–75
Fig. 11. N–S (black curves) and E–W (gray curves) equivalent permeability for all four Alima models, calculated using an average aperture. For comparison with heterogeneous aperture,
the thick black and gray curves show the N–S and E–W heterogeneous aperture averaged for all sub-models; a) N–S compression; b) E–W compression; c) elastoplastic rheology under N–
S compression.
(Fig. 12a). In terms of geometry, the sub-models and the complete
model have similar distributions, with an average spacing of 3.9 m in
the complete model versus 3.8 m in the sub-models.
As with the sub-models, we model the aperture distribution as a
function of stress in the complete model, and calculate fluid pressure
in N–S and E–W directions (Fig. 12b,c). The large contrast between N–
S and E–W fluid pressures observed in the sub-models is also seen in
the complete Alima model (Fig. 12b,c). The increased permeability in
N–S direction is mainly the result of a few flow pathways along hydrau-
lically open fractures (Fig. 12c), similar to what was observed in the sub-
models.
The sub-models underestimate the permeability of the complete
Alima model for both heterogeneous and averaged apertures with an
average error of 10% (Fig. 12d-f). The permeability values in these fig-
ures are the average of N–S and E–W permeability. The underestimation
is largest for the models with the highest permeability (i.e. N–S
compression), while for E–W compression and elastoplastic models,
the underestimation is small. Under N–S compression, the increase in
permeability is controlled by a limited number of flow pathways.
These pathways are partly outside the sub-models, resulting in an
underestimation.
6. Discussion
6.1. Fracture aperture versus geometry
The amount of shear displacement and the resulting hydraulic aper-
ture distributions in the presented models are partially driven by the
orientation of the fracture with respect to the direction of σ1. For maxi-
mum hydraulic aperture, Barton–Bandis predicts that fractures should
ideally strike at a small angle (i.e. close to 15°) with respect to σ1
(Fig. 4c). As a result, model Alima3, which has the largest spacing and
smallest fracture intersection count, but ideally oriented fractures for
large hydraulic apertures, has a larger equivalent permeability than
the other models. Figures similar to Fig. 4c may be useful for defining
the relation between fracture orientation and hydraulic aperture.
When in-situ stress data are available, and the mechanical rock proper-
ties are relatively homogeneous, such that fracture attributes are only
influenced by stress, the impact of fracture orientation can be imple-
mented into DFN modeling workflows to define a hydraulic aperture
distribution without the need of geomechanical modeling of aperture.
In the studied models, the fracture orientation and the associated
hydraulic aperture distribution have a stronger impact on equivalent
permeability than length or spacing. Spacing does not significantly im-
pact equivalent permeability, even though we have defined spacing as
a function of length using Eq. (6). We do find that for constant spacing,
equivalent permeability is higher for a fracture system consisting of few
large fractures rather than many small fractures. This distinction is how-
ever not made by Eq. (6).
In terms of fracture network connectivity, the number of pathways
in both horizontal directions has no direct impact on permeability, as
these pathways are not necessarily hydraulically open under the given
stress conditions. Percolation studies often assume a non-permeable
matrix, i.e. disconnected fractures do not contribute to flow. This yields
a binary response in permeability, where breaking up a fracture net-
work by removing a single fracture may yield a large change in perme-
ability (i.e. the percolation threshold) (Long and Witherspoon, 1985;
Berkowitz and Balberg, 1993). We do not find this effect in our perme-
ability analysis, where 10 mD matrix flow is included and fracture aper-
ture is relatively small. Other studies that take into account coupled
fracture-matrix flow find that even with matrix permeability, connec-
tivity has a larger impact than aperture (Philip et al., 2005; Olson
et al., 2009). However, the contrast between matrix and fracture perme-
ability is large in most of these models, resulting in an equivalent per-
meability that is effectively only controlled by fracture permeability
(Matthäi and Belayneh, 2004). For low matrix permeability (i.e. less
than 10 mD), fracture network connectivity is controlling for fluid
flow, while for a more permeable matrix, absolute apertures define
flow (Bisdom et al., in press). Furthermore, since the Barton–Bandis
model also includes segments with zero hydraulic aperture, it also
impacts connectivity.
6.2. Upscaling aperture and permeability
In contrast to other studies (Paluszny and Matthäi, 2010; Nick et al.,
2011), the permeability of our models with average weighted aperture
is smaller than those with the heterogeneous aperture distribution. This
difference is likely related to the type of aperture modeling, as we con-
sider aperture under compression with a low fluid pressure, yielding
values smaller than 0.3 mm. These apertures do result in a fracture
permeability as defined by the cubic law that ranges up to 800 Darcy
 K. Bisdom et al. / Tectonophysics 690 (2016) 63–75 73

Fig. 12. Permeability of the Alima sub-models compared to the full model: a) hydraulic aperture distribution in the Alima model, with the areas of the sub-models indicated for reference;
b) E–W fluid pressure for Alima, white lines indicate pressure contours; c) N–S fluid pressure. Arrows indicate fluid velocity (small gray arrows indicate low velocity, large black arrows
areas with high velocity); d–f) permeability comparison between the average permeability of the sub-models and the full model, calculated using the heterogeneous and constant
averaged aperture distributions.

for the base case models in Fig. 7, but the total change in equivalent
permeability of the entire network is relatively small.
Another possible cause for the underestimation is that we apply the
averaging method for the entire hydraulic aperture distribution ranging
from 3 × 10−4 m to 4 × 10−8 m, whereas earlier work only contains
hydraulically open fractures with a hydraulic aperture larger than
1 × 10−6 m (Nick et al., 2011).
6.3. Predictive value of 2-D flow models for 3-D fractured reservoirs
The studied models are limited to fracture lengths in 2-D, representing
a horizontal section through a bed-perpendicular fracture network, cap-
turing the horizontal equivalent permeabilities, without considering ver-
tical permeability. The studied models can be used to define horizontal
permeability in layers with bed-confined fractures or highly persistent
fractures, where the 2-D horizontal slice is representative for the entire
vertical section. Gravity and overburden stress is not taken into account,
but in thin sub-horizontal layers, gravity has no impact when assuming
fully saturated single-phase flow.
In 3-D bed-perpendicular fracture systems, the stress-induced vari-
ations in aperture predicted by Barton–Bandis may become relatively
insignificant compared to mesoscopic rotation of fractured rock seg-
ments resulting from slip along bedding planes (e.g. Lei et al., 2015).
Slip along bedding planes may also result in increased opening of frac-
tures that is not described by the Barton–Bandis model. This process is
observed through microseismic for hydraulic fractures, but may also af-
fect natural fractures (Gu et al., 2008).
In natural fracture networks, where fractures commonly have
scattered orientations resulting in highly-connected networks, the
consideration of 3-D connectivity significantly increases permeability
magnitude and anisotropy (de Dreuzy et al., 2012). At the fracture
scale, these 3-D models show that heterogeneous aperture along single
fractures results in flow bottlenecks that decrease permeability, which
is also observed in our 2-D models.
74 K. Bisdom et al. / Tectonophysics 690 (2016) 63–75
7. Conclusions
We have presented an integrated workflow that applies the Barton–
Bandis method to model stress-sensitive normal and hydraulic aperture
in fracture networks to generate heterogeneous aperture distributions
even along single fractures. The impact of heterogeneous aperture on
flow is quantified in terms of equivalent permeability, modeling the
fractures with heterogeneous apertures explicitly in a permeable
matrix.
We show that this workflow can be used to study equivalent per-
meability in complex fracture networks, using deterministic fracture
patterns from outcrops to obtain realistic geometries. The studied
networks have relatively simple geometries compared to what is
commonly observed in outcrops, but this permits for studying in
detail the impact of each individual geometrical attribute on the
aperture distribution along single fractures and on permeability.
However, the presented workflow can be applied to larger multi-
scale discrete fracture networks (Bisdom et al., in press).
Our main findings are:
i. The Barton–Bandis model predicts that the hydraulic aperture of
fractures with minor amounts of shear under compression is highly
heterogeneous, even within single fractures. This results in a discon-
nected fracture network where a small fraction of the network con-
tributes significantly to the equivalent permeability. In our models,
the angle between σ1 and the fracture strike is the controlling factor
for hydraulic aperture and subsequent fracture network permeabil-
ity. Fractures that strike slightly oblique to σ1 experience both
dilation and a small amount of shear displacement, creating poorly
interlocking fractures with a relatively large hydraulic aperture.
ii. In the studied models, fracture block size and spacing have a rela-
tively small impact on equivalent permeability. When considering
both fracture and matrix flow, the role of percolating clusters of frac-
tures is negligible, as is the impact of fracture connectivity. From the
four sub-models studied, the model with the lowest fracture density
and lowest connectivity, showed the highest equivalent permeabil-
ity as most fractures were oriented ideally with respect to σ1.
iii. Upscaling of the heterogeneous aperture distribution to a single av-
eraged aperture may provide the same equivalent permeability as
the heterogeneous model when there is no large contrast between
flow in different directions. Averaging and upscaling equivalent
permeability of small-scale fracture models can provide an accurate
description of permeability when flow is relatively homogeneous,
i.e. most fractures contribute to flow. Otherwise, the small-scale
models may underestimate equivalent permeability.
The stress-based aperture and permeability modeling workflow,
with heterogeneous apertures along single fractures, may provide
more representative heterogeneous aperture distributions for describ-
ing small-scale permeability in non-reactive reservoir rocks (i.e. within
a grid cell), from which the resulting permeability may be assigned to
grid cells in reservoir-scale fracture flow models.
Acknowledgments
This work is part of the PhD research of the first author, sponsored by
TOTAL S.A.. The fieldwork data in this study has been acquired in collab-
oration with S. Bouaziz, A. Hammami, and S. Makni (ENIS Sfax), R.
Vaughan (VU University) and T. Thorgilsdottir (TU Delft). Data has
been digitized and processed using DigiFract (N. Hardebol, TU Delft)
and SKUA/Gocad (Paradigm), supported with research plugins (Gocad
Research Group). The fracture dataset for this paper is available in
CAD format through dx.doi.org/10.4121/uuid:6d096e88-14e0-4d6e-
aeca-a0b9f7b3ea56. Petrophysical analysis of rock samples has been
done by S. Toby (VU University). The authors would like to thank S.
Geiger and P. Corbett (Heriot Watt University) for providing inspiration
for the manuscript, and N. Hardebol (Delft University of Technology),
W. van der Zee and M. Holland (Baker Hughes) for their advice regard-
ing geomechanical modeling. We thank the guest editor, O. Lacombe,
and the reviewers (anonymous and P. Gillespie) for their thorough
and detailed reviews, which significantly improved this manuscript.
References
Agosta, F., Alessandroni, M., Antonellini, M., Tondi, E., Giorgioni, M., 2010. From fractures
to flow: a field-based quantitative analysis of an outcropping carbonate reservoir.
Tectonophysics 490 (3–4), 197–213. http://dx.doi.org/10.1016/j.tecto.2010.05.005.
Alzayer, Y., Eichhubl, P., Laubach, S.E., 2015. Non-linear growth Kinematics of opening-
mode fractures. J. Struct. Geol. 74, 31–44. http://dx.doi.org/10.1016/j.jsg.2015.02.003.
Atkinson, B.A., 1987. Fracture Mechanics of Rock: 548 P. http://dx.doi.org/10.1007/978-
94-007-2595-9.
Bandis, S.C., Lumsden, a.C., Barton, N.R., 1983. Fundamentals of rock joint deformation. Int.
J. Rock Mech. Min. Sci. Geomech. Abstr. 20 (6), 249–268. http://dx.doi.org/10.1016/
0148-9062(83)90595-8.
Barton, N., 1976. The shear strength of rock and rock joints. Int. J. Rock Mech. Min. Sci.
Geomech. Abstr. 13 (9), 255–279. http://dx.doi.org/10.1016/0148-9062(76)90003-6.
Barton, N., 1982. Modelling Rock Joint Behaviour from in Situ Block Tests. Implications for
Nuclear Waste Repository Design (Columbus, OH). (96 pp.).
Barton, N., 2014. Non-linear behaviour for naturally fractured carbonates and frac-
stimulated gas-shales. First Break 32 (2031), 51–66. http://dx.doi.org/10.3997/
1365-2397.2014011.
Barton, N., Bandis, S., 1980. Some effects of scale on the shear strength of joints. Int.
J. Rock Mech. Min. Sci. Geomech. Abstr. 69–73 http://dx.doi.org/10.1016/0148-
9062(80)90009-1.
Barton, N., Choubey, V., 1977. The shear strength of rock joints in theory and practice.
Rock Mech. 10 (1–2), 1–54. http://dx.doi.org/10.1007/BF01261801.
Barton, N., Bandis, S., Bakhtar, K., 1985. Strength, deformation and conductivity coupling
of rock joints. Int. J. Rock Mech. Min. Sci. Geomech. Abstr. 22 (3), 121–140. http://
dx.doi.org/10.1016/0148-9062(85)93227-9.
Barton, C.A., Zoback, M.D., Moos, D., 1995. Fluid flow along potentially active faults in crystal-
line rock. Geology 23 (8), 683. http://dx.doi.org/10.1130/0091-7613(1995)023b0683:
FFAPAFN2.3.CO;2.
Becker, S.P., Eichhubl, P., Laubach, S.E., Reed, R.M., Lander, R.H., Bodnar, R.J., 2010. A
48 m.y. history of fracture opening, temperature, and fluid pressure: Cretaceous
Travis peak Formation, East Texas basin. Geol. Soc. Am. Bull. 122 (7–8), 1081–1093.
http://dx.doi.org/10.1130/B30067.1.
Berkowitz, B., 2002. Characterizing flow and transport in fractured geological media: a re-
view. Adv. Water Resour. 25 (8–12), 861–884. http://dx.doi.org/10.1016/s0309-
1708(02)00042-8.
Berkowitz, B., Balberg, I., 1993. Percolation theory and its application to groundwater hy-
drology. Water Resour. Res. 29 (4), 775–794. http://dx.doi.org/10.1029/92wr02707.
Bisdom, K., 2015. Natural Fracture Patterns in Carbonate Rocks. TU Delft http://dx.doi.org/
10.4121/uuid:6d096e88-14e0-4d6e-aeca-a0b9f7b3ea56.
Bisdom, K., Gauthier, B.D.M., Bertotti, G., Hardebol, N.J., 2014. Calibrating discrete fracture-
network models with a carbonate three-dimensional outcrop fracture network: im-
plications for naturally fractured reservoir modeling. AAPG Bull. 98 (7), 1351–1376.
http://dx.doi.org/10.1306/02031413060.
Bisdom, K., Bertotti, G., Nick, H.M., 2016. A geometrically-based method for predicting
stress-induced fracture aperture and flow in discrete fracture networks. AAPG Bull.
http://dx.doi.org/10.1306/02111615127 (in press).
Bonnet, E., Bour, O., Odling, N.E., Davy, P., Main, I., Cowie, P., Berkowitz, B., 2001. Scaling of
fracture systems in geological media. Rev. Geophys. 39 (3), 347. http://dx.doi.org/10.
1029/1999RG000074.
Bouaziz, S., Barrier, E., Soussi, M., Turki, M.M., Zouari, H., 2002. Tectonic evolution of the
northern African margin in Tunisia from paleostress data and sedimentary record.
Tectonophysics 357 (1–4), 227–253. http://dx.doi.org/10.1016/s0040-1951(02)00370-
0.
Chesnaux, R., Allen, D.M., Jenni, S., 2009. Regional fracture network permeability using
outcrop scale measurements. Eng. Geol. 108 (3–4), 259–271. http://dx.doi.org/10.
1016/j.enggeo.2009.06.024.
Cottereau, N., Garcia, M.H., Gosselin, O.R., Vigier, L., 2010. Effective Fracture Network Per-
meability. Comparative Study of Calculation Methods. Society of Petroleum Engi-
neers, Barcelona, Spain http://dx.doi.org/10.2118/131126-ms.
Dassault Systemes, 2013. ABAQUS v6.13 Documentation.
de Dreuzy, J.-R., Méheust, Y., Pichot, G., 2012. Influence of fracture scale heterogeneity on
the flow properties of three-dimensional discrete fracture networks (DFN).
J. Geophys. Res. 117 (B11), B11207. http://dx.doi.org/10.1029/2012JB009461.
Geiger, S., Matthäi, S., 2012. What can we learn from high-resolution numerical simula-
tions of single- and multi-phase fluid flow in fractured outcrop analogues? Geol.
Soc. Lond. Spec. Publ. 374 (1), 125–144. http://dx.doi.org/10.1144/SP374.8.
Gu, H., Siebrits, E., Sabourov, A., 2008. Hydraulic Fracture Modeling with Bedding Plane
Interfacial Slip. SPE Eastern Regional/AAPG Eastern Section Joint Meeting. Society of
Petroleum Engineers http://dx.doi.org/10.2118/117445-MS.
Guerriero, V., Iannace, A., Mazzoli, S., Parente, M., Vitale, S., Giorgioni, M., 2010. Quantify-
ing uncertainties in multi-scale studies of fractured reservoir analogues: implement-
ed statistical analysis of scan line data from carbonate rocks. J. Struct. Geol. 32 (9),
1271–1278. http://dx.doi.org/10.1016/j.jsg.2009.04.016.
 K. Bisdom et al. / Tectonophysics 690 (2016) 63–75
Guerriero, V., Mazzoli, S., Iannace, A., Vitale, S., Carravetta, A., Strauss, C., 2013. A perme-
ability model for naturally fractured carbonate reservoirs. Mar. Pet. Geol. 40 (0),
115–134. http://dx.doi.org/10.1016/j.marpetgeo.2012.11.002.
Hardebol, N.J., Bertotti, G., 2013. DigiFract: a software and data model implementation for
flexible acquisition and processing of fracture data from outcrops. Comput. Geosci.
54, 326–336. http://dx.doi.org/10.1016/j.cageo.2012.10.021.
Hooker, J.N., Gale, J.F.W., Gomez, L.A., Laubach, S.E., Marrett, R., Reed, R.M., 2009.
Aperture-size scaling variations in a low-strain opening-mode fracture set, Cozzette
Sandstone, Colorado. J. Struct. Geol. 31 (7), 707–718. http://dx.doi.org/10.1016/j.jsg.
2009.04.001.
Hooker, J.N., Gomez, L.a., Laubach, S.E., Gale, J.F.W., Marrett, R., 2012. Effects of diagenesis
(cement precipitation) during fracture opening on fracture aperture-size scaling in
carbonate rocks. Geol. Soc. Lond. Spec. Publ. 370 (1), 187–206. http://dx.doi.org/10.
1144/SP370.9.
Hooker, J.N., Laubach, S.E., Marrett, R., 2013. Fracture-aperture size—frequency, spatial
distribution, and growth processes in strata-bounded and non-strata-bounded frac-
tures, Cambrian Mesón Group, NW Argentina. J. Struct. Geol. 54, 54–71. http://dx.
doi.org/10.1016/j.jsg.2013.06.011.
Hooker, J.N., Laubach, S.E., Marrett, R., 2014. A universal power-law scaling exponent for
fracture apertures in sandstones. Geol. Soc. Am. Bull. (9) http://dx.doi.org/10.1130/
B30945.1.
Iñigo, J.F., Laubach, S.E., Hooker, J.N., 2012. Fracture abundance and patterns in the
Subandean fold and thrust belt, Devonian Huamampampa Formation petroleum res-
ervoirs and outcrops, Argentina and Bolivia. Mar. Pet. Geol. 35 (1), 201–218. http://
dx.doi.org/10.1016/j.marpetgeo.2012.01.010.
Jonoud, S., Jackson, M., 2008. Validity of Steady-State Upscaling Techniques. pp. 12–15
http://dx.doi.org/10.2118/100293-PA.
Lander, R.H., Laubach, S.E., 2015. Insights into rates of fracture growth and sealing from a
model for quartz cementation in fractured sandstones. Bull. Geol. Soc. Am. 127 (3–4),
516–538. http://dx.doi.org/10.1130/B31092.1.
Latham, J.-P., Xiang, J., Belayneh, M., Nick, H.M., Tsang, C.-F., Blunt, M.J., 2013. Modelling
stress-dependent permeability in fractured rock including effects of propagating
and bending fractures. Int. J. Rock Mech. Min. Sci. 57, 100–112. http://dx.doi.org/10.
1016/j.ijrmms.2012.08.002.
Laubach, S.E., 2003. Practical approaches to identifying sealed and open fractures. AAPG
Bull. 87 (4), 561–579. http://dx.doi.org/10.1306/11060201106.
Laubach, S.E., Ward, M.E., 2006. Diagenesis in porosity evolution of opening-mode frac-
tures, Middle Triassic to Lower Jurassic La Boca Formation, NE Mexico.
Tectonophysics 419 (1–4), 75–97. http://dx.doi.org/10.1016/j.tecto.2006.03.020.
Laubach, S.E., Olson, J.E., Gale, J.F., 2004a. Are open fractures necessarily aligned with max-
imum horizontal stress? Earth Planet. Sci. Lett. 222 (1), 191–195. http://dx.doi.org/
10.1016/j.epsl.2004.02.019.
Laubach, S.E., Reed, R.M., Olson, J.E., Lander, R.H., Bonnell, L.M., 2004b. Coevolution of
crack-seal texture and fracture porosity in sedimentary rocks: cathodoluminescence
observations of regional fractures. J. Struct. Geol. 26 (5), 967–982. http://dx.doi.org/
10.1016/j.jsg.2003.08.019.
Lei, Q., Latham, J.-P., Xiang, J., Tsang, C.-F., Lang, P., Guo, L., 2014. Effects of geomechanical
changes on the validity of a discrete fracture network representation of a realistic
two-dimensional fractured rock. Int. J. Rock Mech. Min. Sci. 70, 507–523. http://dx.
doi.org/10.1016/j.ijrmms.2014.06.001.
Lei, Q., Latham, J., Xiang, J., Tsang, C., 2015. Polyaxial stress-induced variable aperture
model for persistent 3D fracture networks. Geomech. Energy Environ. 1, 34–47.
http://dx.doi.org/10.1016/j.gete.2015.03.003.
Long, J.C.S., Witherspoon, P.a., 1985. The relationship of the degree of interconnection to
permeability in fracture networks. J. Geophys. Res. 90 (B4), 3087. http://dx.doi.org/
10.1029/JB090iB04p03087.
Matsuki, K., Wang, E.Q., Giwelli, a.a., Sakaguchi, K., 2008. Estimation of closure of a frac-
ture under normal stress based on aperture data. Int. J. Rock Mech. Min. Sci. 45 (2),
194–209. http://dx.doi.org/10.1016/j.ijrmms.2007.04.009.
Matthäi, S.K., Belayneh, M., 2004. Fluid flow partitioning between fractures and a perme-
able rock matrix. Geophys. Res. Lett. 31 (7). http://dx.doi.org/10.1029/2003GL019027
(p. n/a–n/a).
Matthäi, S.K., Nick, H.M., 2009. Upscaling two-phase flow in naturally fractured reservoirs.
AAPG Bull. 93 (11), 1621–1632. http://dx.doi.org/10.1306/08030909085.
Matthäi, S.K., et al., 2007. Numerical simulation of multi-phase fluid flow in structurally
complex reservoirs. Geol. Soc. Lond. Spec. Publ. 292 (1), 405–429. http://dx.doi.org/
10.1144/SP292.22.
National Research Council, 1996. Rock Fractures and Fluid Flow. National Academies
Press, Washington, D.C. http://dx.doi.org/10.17226/2309 (568 pp.).
75
Nelson, R.A., 2001. Geologic Analysis of naturally Fractured Reservoirs. Gulf Professional
Publishing, Woburn (332 pp.).
Nemoto, K., Watanabe, N., Hirano, N., Tsuchiya, N., 2009. Direct measurement of contact
area and stress dependence of anisotropic flow through rock fracture with heteroge-
neous aperture distribution. Earth Planet. Sci. Lett. 281 (1–2), 81–87. http://dx.doi.
org/10.1016/j.epsl.2009.02.005.
Nick, H.M., Matthäi, S.K., 2011. Comparison of three FE-FV numerical Schemes for single-
and two-phase flow simulation of fractured porous media. Transp. Porous Media 90
(2), 421–444. http://dx.doi.org/10.1007/s11242-011-9793-y.
Nick, H.M., Paluszny, A., Blunt, M.J., Matthäi, S.K., 2011. Role of geomechanically grown
fractures on dispersive transport in heterogeneous geological formations. Phys. Rev.
E 84 (5), 056301. http://dx.doi.org/10.1103/PhysRevE.84.056301.
Olson, J.E., Laubach, S.E., Lander, R.H., 2009. Natural fracture characterization in tight gas
sandstones: integrating mechanics and diagenesis. AAPG Bull. 93 (11), 1535–1549.
http://dx.doi.org/10.1306/08110909100.
Olsson, R., Barton, N., 2001. An improved model for hydromechanical coupling during
shearing of rock joints. Int. J. Rock Mech. Min. Sci. 38 (3), 317–329. http://dx.doi.
org/10.1016/S1365-1609(00)00079-4.
Ortega, O.J., Marrett, R.A., Laubach, S.E., 2006. A scale-independent approach to fracture
intensity and average spacing measurement. AAPG Bull. 90 (2), 193–208. http://dx.
doi.org/10.1306/08250505059.
Paluszny, A., Matthäi, S.K., 2010. Impact of fracture development on the effective perme-
ability of porous rocks as determined by 2-D discrete fracture growth modeling.
J. Geophys. Res. 115 (B2), B02203. http://dx.doi.org/10.1029/2008JB006236.
Paluszny, A., Zimmerman, R.W., 2011. Numerical simulation of multiple 3D fracture prop-
agation using arbitrary meshes. Comput. Methods Appl. Mech. Eng. 200 (9–12),
953–966. http://dx.doi.org/10.1016/j.cma.2010.11.013.
Paluszny, A., Matthäi, S.K., Hohmeyer, M., 2007. Hybrid finite element-finite volume
discretization of complex geologic structures and a new simulation workflow dem-
onstrated on fractured rocks. Geofluids 7 (2), 186–208. http://dx.doi.org/10.1111/j.
1468-8123.2007.00180.x.
Philip, Z.G., Jennings, J.W., Olson, J.E., Laubach, S.E., Holder, J., 2005. Modeling coupled frac-
ture-matrix fluid flow in geomechanically Simulated fracture networks. SPE Reserv.
Eval. Eng. 8 (04), 300–309. http://dx.doi.org/10.2118/77340-PA.
Riley, P., Gordon, C., Simo, J.a., Tikoff, B., Soussi, M., 2010. Structure of the Alima and asso-
ciated anticlines in the foreland basin of the southern Atlas Mountains, Tunisia. Lith-
osphere 3 (1), 76–91. http://dx.doi.org/10.1130/L119.1.
Rogers, S.F., 2003. Critical stress-related permeability in fractured rocks. Geol. Soc. Lond.
Spec. Publ. 209 (1), 7–16. http://dx.doi.org/10.1144/GSL.SP.2003.209.01.02.
Saïd, A., Baby, P., Chardon, D., Ouali, J., 2011. Structure, paleogeographic inheritance, and
deformation history of the southern Atlas foreland fold and thrust belt of Tunisia. Tec-
tonics 30 (6). http://dx.doi.org/10.1029/2011TC002862 p. n/a–n/a.
Smart, K., Ferrill, D., Morris, A.P., McGinnis, R.N., 2010. Geomechanical Modeling of a Res-
ervoir-Scale Fault-Related Fold: the Bargy Anticline, France. 44th US Rock Mechanics
Symposium.
Smart, K.J., Ofoegbu, G.I., Morris, A.P., Mcginnis, R.N., Ferrill, D.A., 2014. Geomechanical
modeling of hydraulic fracturing : why mechanical stratigraphy, stress state, and
pre-existing structure matter. AAPG Bull. 11 (11), 2237–2261. http://dx.doi.org/10.
1306/07071413118.
Snow, D.T., 1969. Anisotropic Permeability of Fractured Media. p. 1273 http://dx.doi.org/
10.1029/WR005i006p01273.
Tao, Q., Ehlig-Economides, C.A., Ghassemi, A., 2009. Investigation of Stress-Dependent
Permeability in Naturally Fractured Reservoirs Using a Fully Coupled Poroelastic Dis-
placement Discontinuity Model. SPE Annual Technical Conference and Exhibition: So-
ciety of Petroleum Engineers http://dx.doi.org/10.2118/124745-MS.
Wilson, C.E., Aydin, A., Karimi-Fard, M., Durlofsky, L.J., Amir, S., Brodsky, E.E., Kreylos, O.,
Kellogg, L.H., 2011. From outcrop to flow simulation: constructing discrete fracture
models from a LIDAR survey. AAPG Bull. 95 (11), 1883–1905. http://dx.doi.org/10.
1306/03241108148.
Wu, H.Q., Pollard, D.D., 2002. Imaging 3-D fracture networks around boreholes. AAPG Bull. 86
(4), 593–604. http://dx.doi.org/10.1306/61EEDB52-173E-11D7-8645000102C1865D.
Zoback, M.D., 2007. Reservoir Geomechanics: Cambridge, Cambridge University Press,
449 P. http://dx.doi.org/10.1017/CBO9780511586477.
Zouaghi, T., Bédir, M., Abdallah, H., Inoubli, M.H., 2009. Seismic sequence stratigraphy,
basin structuring, and hydrocarbon implications of Cretaceous deposits (Albian–
Maastrichtian) in Central Tunisia. Cretac. Res. 30 (1), 1–21. http://dx.doi.org/10.
1016/j.cretres.2008.02.005.