Computers and Geotechnics 85 (2017) 151–176

Contents lists available at ScienceDirect

Computers and Geotechnics

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

Review

The use of discrete fracture networks for modelling coupled

geomechanical and hydrological behaviour of fractured rocks
Qinghua Lei a,⇑, John-Paul Latham a, Chin-Fu Tsang b,c
a
Department of Earth Science and Engineering, Imperial College London, London, UK
b
Department of Earth Sciences, Uppsala University, Uppsala, Sweden
c
Earth Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA, USA

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

Article history: We present a discussion of the state-of-the-art on the use of discrete fracture networks (DFNs) for mod-
Received 24 June 2016 elling geometrical characteristics, geomechanical evolution and hydromechanical (HM) behaviour of nat-
Received in revised form 3 November 2016 ural fracture networks in rock. The DFN models considered include those based on geological mapping,
Accepted 24 December 2016
stochastic generation and geomechanical simulation. Different types of continuum, discontinuum and
Available online 2 January 2017
hybrid geomechanical models that integrate DFN information are summarised. Numerical studies aiming
at investigating geomechanical effects on fluid flow in DFNs are reviewed. The paper finally provides rec-
Keywords:
ommendations for advancing the modelling of coupled HM processes in fractured rocks through more
Fractured rock
DFN
physically-based DFN generation and geomechanical simulation.
Geomechanical modelling Ó 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://
Fluid flow creativecommons.org/licenses/by/4.0/).
HM coupling

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
2. Representation of natural fracture networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
2.1. Geologically-mapped fracture networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
2.2. Stochastically-generated fracture networks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
2.3. Geomechanically-grown fracture networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
2.4. Summary with a view towards HM modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
3. Modelling of the geomechanical behaviour of fracture networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
3.1. Continuum/extended-continuum models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
3.2. Block-type discontinuum models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
3.2.1. Distinct element method (DEM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
3.2.2. Discontinuous deformation analysis (DDA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
3.3. Particle-based discontinuum models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
3.4. Hybrid finite-discrete element models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
3.4.1. ELFEN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
3.4.2. Y-code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
3.5. Summary with a view towards HM modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
4. Modelling of the hydrological behaviour of fracture networks under geomechanical effects. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
4.1. Fluid pathways . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
4.2. Permeability. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
4.3. Transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
4.4. Comments on the use of DFNs for HM modelling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
5. Outstanding issues and concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170

⇑ Corresponding author.
E-mail addresses: q.lei12@imperial.ac.uk (Q. Lei), j.p.latham@imperial.ac.uk (J.-P. Latham), cftsang@lbl.gov (C.-F. Tsang).

http://dx.doi.org/10.1016/j.compgeo.2016.12.024
0266-352X/Ó 2017 The Authors. Published by Elsevier Ltd.
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
152 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176
1. Introduction
Fractures such as joints, faults, veins and bedding planes are
ubiquitous in crustal rocks. These naturally occurring discontinu-
ities often comprise complex networks and dominate the geome-
chanical and hydrological behaviour of subsurface rocks [1].
Understanding of the nontrivial effect of fractures is a challenging
issue which is relevant to many engineering applications such as
underground construction, enhanced geothermal systems, uncon-
ventional shale gas production (fracking), groundwater manage-
ment and radioactive waste disposal [2–4]. The importance of
the presence of natural fractures, which can result in heteroge-
neous stress fields [5] and channelized fluid flow pathways [6] in
highly disordered geological formations, has promoted the devel-
opment of robust fracture network models for numerical simula-
tion of fractured rocks [7].
The first difficulty in modelling fractured rocks is the geometri-
cal representation of complex three-dimensional (3D) discontinu-
ity systems. Natural fractures form under certain mechanically
self-organised dynamics, where breakage and fragmentation can
occur at all scales [8]. They are subject to in-situ stress fields at
depth and can form intricate topologies, such as cross-cutting,
abutting, branching, termination, bends, spacing and clustering
[9,10]. However, direct observation of the detailed 3D structure
of fracture networks deep in the crust is impossible. Field data
are usually collected from lower dimensional limited exposures,
e.g. one-dimensional (1D) borehole logging and two-dimensional
(2D) outcrop mapping [11]. Seismological surveys may be able to
locate 3D large-scale structures but the current technology can
hardly detect widely-spreading medium and small fractures due
to the resolution limit. The description of natural fracture geome-
tries, therefore, has to largely rely on extrapolations, from 1D/2D
to 3D and from small samples to the whole study domain. Hence,
the question of how to create realistic fracture networks remains
an unresolved issue. Often simplifications have to be made by
ignoring some details of less importance for the particular ques-
tions at hand.
The second fundamental issue in modelling fractured rocks is to
solve the fracture and solid mechanics problems of the discontinu-
ous geological media under complex boundary conditions, i.e. the
fractured rock mass mechanical response to boundary stresses or
displacements. When the spacing of natural fractures is compara-
ble to the scale of interest of the problem to be modelled, the con-
ventional continuum approach may not be adequate to capture
some important mechanical behaviour of fractured rocks [12], such
as fracturing of intact rocks [13], interaction of multiple fractures
[5], and opening, shearing and dilatancy of rough fracture walls
[14]. Thus, many computational schemes based on extended con-
tinuum, discontinuum, or hybrid continuum-discontinuum meth-
ods have been developed to solve a numerical system with
fracture geometries explicitly represented.
Another important question is about the hydrological behaviour
of fractured rocks under geomechanical conditions. Fractured rocks
may deform in response to geological and/or man-made perturba-
tions (e.g. tectonic events, underground excavations), resulting in
changes of bulk permeability and fluid migration [2]. The intricacy
of such coupled hydromechanical (HM) processes is increased if
the presence of natural fractures associated with topological com-
plexities is to be considered [1,15]. The quest for a means of quan-
tifying the influence of in-situ stresses on the permeability of
fractured reservoirs has been driven largely by the motivation from
petroleum engineering [16]. The understanding of contaminant
migration through tectonically strained fractured formations is
also crucial for the groundwater community [17] and nuclear
waste management [3,4].
The objective of this paper is to review the current state-of-the-
art of fracture network models and to provide some discussions
and recommendations for HM modelling of fractured rocks. In this
context, the issues arising in the aforementioned three key subject
areas (i.e. geometry, geomechanics and hydromechanics) will be
examined progressively. The paper will present a summary of var-
ious approaches used in developing fracture network models that
represent natural fracture geometries often with different degrees
of simplification, and different numerical frameworks that inte-
grate discrete fracture representations for modelling geomechani-
cal and HM behaviour of fractured rocks. One important focus of
this work is to discuss the features or characteristics of a fracture
network model needed to properly simulate coupled HM processes
in fractured rocks. The rest of the paper is organised as follows.
Section 2 reviews the methods of representing natural fracture
geometries by geological mapping, stochastic generation or
geomechanical simulation. Section 3 provides a brief overview of
continuum and discontinuum models that integrate explicit frac-
ture geometries for geomechanical modelling of fractured rocks.
A short summary is presented in each of Sections 2 and 3 with a
view towards HM modelling. Section 4 summarises the numerical
studies of geomechanical effects on fluid flow in fractured rocks
and further provides some suggestions on choosing appropriate
DFNs and geomechanical models for HM modelling. Finally, out-
standing issues are discussed and some concluding remarks are
made. It has to be mentioned that this review is focused on HM
behaviour of fractured rocks and is not exhaustive. Only limited
references are cited for brevity. For more extensive descriptions
of fracture network models, the reader is referred to the reviews
by Dershowitz and Einstein [18] and Liu et al. [19], and the text-
books by Adler and Thovert [20] and Adler et al. [21]. Extensive
reviews of different numerical methods developed in the field of
rock mechanics can be found in Jing and Hudson [22], Jing [12],
Yuan and Harrison [23], Bobet et al. [24], and Lisjak and Grasselli
[25]. More in-depth discussions about stress effects on fluid flow
in fractured rocks can be found in Zhang and Sanderson [26], and
Rutqvist and Stephansson [2].
2. Representation of natural fracture networks
A ‘‘discrete fracture network” (DFN) refers to a computational
model that explicitly represents the geometrical properties of each
individual fracture (e.g. orientation, size, position, shape and aper-
ture), and the topological relationships between individual frac-
tures and fracture sets. Unlike the conventional definition of
DFNs that corresponds to stochastic fracture networks, the term
DFN here represents a much broader concept of any explicit frac-
ture network model. A DFN can be generated from geological map-
ping, stochastic realisation or geomechanical simulation to
represent different types of rock fractures including joints, faults,
veins and bedding planes.
2.1. Geologically-mapped fracture networks
Fracture patterns can be mapped from the exposure of rock out-
crops or man-made excavations (e.g. borehole, quarry, tunnel and
roadcut). These geologically-mapped fracture networks were
widely used to understand the process of fracture formation
[5,27], interpret the history of tectonic stresses [28–30], and derive
the statistics and scaling of fracture populations [8,31,32]. How-
ever, digitised outcrop analogues (Fig. 1) can also be used to build
DFNs for numerical simulations. For example, a series of discrete
fracture patterns were mapped from limestone outcrops at the
south margin of the Bristol Channel Basin, UK [33]. The traced
DFNs were used to study the connectivity [34], multiphase flow
 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176
Fig. 1. Geologically-mapped DFN patterns based on (a) a limestone outcrop at the south margin of the Bristol Channel Basin, UK [33], (b) sandstone exposures in the
Dounreay area, Scotland [43], and (c) fault zone structures in the Valley of Fire State Park of southern Nevada, USA [53].
[35–37], solute transport [38] and HM behaviour [15,39–41] of
natural fracture systems. Similar outcrop-based DFNs have also
been constructed by many other researchers for modelling natural
fracture systems [42–48]. It is noteworthy that the terrestrial
remote sensing techniques (e.g. LIDAR and photogrammetry) have
recently been introduced to better capture large-scale trace fea-
tures [49–52]. Fracture apertures may be determined from a
detailed field mapping [53,54] and further calibrated by comparing
flow simulation results with in-situ measurements [55]. However,
apertures were more commonly assumed to be constant or to fol-
low an a priori statistical distribution (sometimes correlated with
trace lengths).
Advantages of such an outcrop-based DFN model include
preservation of natural fracture features (e.g. curvature and seg-
mentation) and unbiased characterisation of complex topologies
(e.g. intersection, truncation, arrest, spacing, clustering and hierar-
chy). However, it is often constrained to 2D analysis and may not
be applicable for general 3D problems involving obliquely dipping
fractures. The LIDAR survey combined with assumptions of frac-
ture shape/persistence [56,57] may provide an estimation of 3D
structural variations in near surface, but it is not applicable to bur-
ied geological complexities. Extrapolation from borehole imaging
can provide an estimation of 3D fracture distributions but confi-
dence can only be guaranteed for the areas close to boreholes
[58]. Seismic data can be used to build 3D maps of large-scale geo-
logical structures, for which, however, the limited resolution often
obscures detailed features such as the segmentation of faults and
impedes the detection of small cracks widely spreading in subsur-
face rocks [59].
2.2. Stochastically-generated fracture networks
Due to the difficulty of performing a complete measurement of
3D natural fracture systems, stochastic approaches using statistics
from limited sampling have been developed and widely used [18].
The stochastic DFN method emerged in the 1980s with the aims to
study the percolation of finite-sized fracture populations [60–62]
and fluid flow in complex fracture networks [63–67].
153
The general stochastic DFN approach assumes fractures to be
straight lines (in 2D) or planar discs/polygons (in 3D), and treats
the other geometrical properties (e.g. position, frequency, size, ori-
entation, aperture) as independent random variables obeying cer-
tain probability distributions [68] derived from field
measurements such as scanline [69] or window sampling [70] of
outcrop traces as well as borehole imaging [71]. The orientation
data can be processed using a rosette or stereogram so that frac-
tures can be grouped into different sets with their orientations
characterised by e.g. a uniform, normal or Fisher distribution
[11]. Fracture sizes may exhibit a negative exponential, lognormal,
gamma or power law distribution [72,73]. Fracture frequency can
be described in terms of fracture density or fracture intensity
and denoted using the Pij system, where i gives the dimension of
the sample and j the dimension of the measurement [74]. Fracture
density measures the number of fractures per unit volume (P30),
area (P20) or length (P10), while fracture intensity represents the
total fracture persistence per unit volume (P32), area (P21) or length
(P10). Fracture spacing is related to fracture frequency (P10) [75],
which is both a density and intensity measure. Fracture spacing
may follow a negative exponential, lognormal or normal distribu-
tion depending on the degree of fracture saturation in the network
[76]. Fracture apertures usually obey a lognormal [77] or power
law distribution [78,79], and may be related to fracture sizes by
a power law [73] with a linear [5] or sublinear [80] scaling relation-
ship. The measurement data (e.g. those based on outcrop sam-
pling) may be biased under the truncation and censoring effects
and requires to be amended to determine the underlying statistical
distributions [81–83]. The statistics of 3D fracture systems (e.g.
size, spacing, density distributions) may be extrapolated from
1D/2D sampling data based on stereological analysis [84,85]. In
the stochastic simulation, fractures can be assumed randomly
located (represented by their barycentres), while the geometrical
attributes can be sampled from the corresponding probability den-
sity functions [18]. Such a random fracture network modelling
approach, termed the conventional Poisson DFN model (or Baecher
model) (Fig. 2), has been implemented within the commercial soft-
ware FracMan [86] and also adopted by many research codes to
study the connectivity, fragmentation, deformability, permeability
154 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176
Fig. 2. The conventional Poisson DFN models: (a) a 2D random fracture pattern conditioned by field data from the Sellafield site, Cumbria, UK [95] and (b) a conceptual model
(no scale specified) of 3D random fracture network with three orthogonal sets of disc-shaped fractures [64].
and transport properties of fractured rocks in the past three dec-
ades [87–107] (only a few among many others).
However, the conventional Poisson DFN model tends to have
large uncertainties due to its assumption of a homogeneous spa-
tial distribution, simplification of fracture shape using linear/pla-
nar geometries, and negligence of the correlations between
different geometrical properties as well as disregard of the diverse
topological relations (e.g. ‘‘T” type intersections). Several
researchers have examined the conventional Poisson DFN model
by comparing it with an original natural fracture network with
respect to geometrical, geomechanical and hydrological proper-
ties, and significant discrepancies were observed [36,39,108–
110]. Several improvements on the Poisson DFN model have been
developed and include considerations of: (i) the inhomogeneity of
fracture spatial distribution based on a geostatistically-derived
density field [67,111] or a spatial clustering process [112–114],
(ii) the correlation between fracture attributes (i.e. length, orien-
tation and position) based on an elastic energy criterion
[115,116], (iii) the unbroken areas inside individual fracture
planes based on a Poisson line tessellation and zone marking pro-
cess [18,117], (iv) the topological complexity based on a charac-
terisation of fracture intersection types [118] and the
termination of fractures at intersections with pre-existing frac-
tures [86], and (v) the mechanical interaction between neighbour-
ing fractures based on a stress shadow zone model [76,119,120].
It has to be mentioned that an alternative to the Poisson DFN
model may use the spacing distribution to locate fractures in
the stochastic generation [121], which is however considered
more suitable for highly persistent fracture systems. For more
detailed summary of the Poisson model and other improved mod-
els, the reader is referred to Dershowitz and Einstein [18], Staub
et al. [122] and Fadakar-Alghalandis [114].
A more systematic characterisation of the hierarchy, clustering
and scaling of natural fracture systems may involve the methods of
fractal geometry and power law models [73]. Extensive field obser-
vations suggest that fracturing occurs at all scales in the crust and
creates a hierarchical structure that can exhibit long-range correla-
tions from macroscale frameworks to microscale fabrics
[8,124,125]. The spatial organisation of natural fracture networks
can be characterised by the fractal dimension D, which quantifies
the manner whereby fractals cluster and spread in the Euclidean
space and can be measured using the box-counting method
[31,126,127] or the two-point correlation function [128,129]. The
density and length distribution of a fracture population can be then
described by a statistical model [32,130] given as: n(l, L) = aLDl
a,
for l 2 ½lmin ; lmax , where n(l, L)dl gives the number of fractures with
sizes belonging to the interval [l, l + dl] (dl  l) in an elementary
volume of characteristic size L, a is the power law length exponent,
a is the density term, and lmin and lmax are the smallest and largest
fracture sizes. The exponent a, which defines the respective pro-
portion of large and small fractures, can be derived from the cumu-
lative distribution or density distribution of fracture lengths
[72,82]. In theory, D is restricted to the range of [1, 2] for 2D and
[2, 3] for 3D, and a is restricted to [1, 1] for 2D and [2, 1] for 3D.
However, extensive measurements based on 2D trace maps reveal
that generally D varies between 1.5 and 2 and a falls between 1.3
and 3.5 [73] in natural fracture systems. The D and a measured
from 1D/2D samples can be extrapolated to derive 3D parameters
based on stereological relationships [123]. The density term a is
related to the total number of fractures in the system and varies
as a function of fracture orientations [130]. The extent of the power
law relation is bounded by an upper limit lmax that is probably
related to the thickness of the crust, and a lower limit lmin that is
constrained by a physical length scale (e.g. grain size) or the reso-
lution of measurement [73]. For numerical simulations, the model
size L usually meets lmin  L  lmax [131]. A fractal spatial distribu-
tion of fracture barycentres can be modelled through a multiplica-
tive cascade process governed by a prescribed D value, while
fracture lengths can be sampled from a power law distribution
having an exponent a [131]. Fracture orientations can be assigned
isotropically or based on a statistical distribution. Fractal fracture
networks can then be generated through a synthesis of the differ-
ent geometrical attributes modelled by independent random vari-
ables (Fig. 3). A D value of 2 (in 2D) and 3 (in 3D) represents a
homogeneous spatial distribution, i.e. ‘‘space filling”. As D
decreases, the fracture pattern becomes more clustered associated
with more empty areas. A small a value corresponds to a system
dominated by large fractures, while a ? 1 relates to a pattern with
all fractures having an equal size (i.e. lmin). The D and a values as
well as their relationship may control the connectivity, permeabil-
ity and strength of fractured rocks [131–134]. More interestingly,
when a = D + 1, the fracture network is self-similar and the connec-
tivity properties are scale invariant [131]. A self-similar fracture
pattern statistically exhibits a hierarchical characteristic whereby
a large fracture inhibits the propagation of smaller ones in its vicin-
ity, but not the converse [32,130]. Implementation of such a hier-
archical rule together with subcritical fracture growth laws leads
to a new DFN model that can simulate the sequential stages of frac-
ture network formation associated with nucleation, propagation
and arrest processes (Fig. 4a) [135].
 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176 155

Fig. 3. 2D and 3D fractal fracture networks generated with different values of the fractal dimension D and the power law length exponent a while the ratio of the domain size
to the minimum fracture length L/lmin = 1024 [123].

Fig. 4. Some new stochastic DFN models: (a) trace map views of a 3D sequential DFN model that simulates the nucleation, growth and arrest processes of natural fractures
(the ratio of the domain size L to the characteristic nuclei length lN is 100) [135] and (b) a 2D self-referencing DFN model that extrapolates a larger scale pattern recursively
from a smaller one through random walks with the preservation of natural fracture curvatures [40].

The assumption of a linear/planar fracture shape in the Poisson
model and fractal DFN model may be simplistic. Many field obser-
vations have shown that natural fracture geometries can be curved
and irregular [9]. To simulate the non-planarity of natural frac-
tures, Lei et al. [40] developed a novel DFN model that accommo-
dates discrete-time random walks in recursive self-referencing
lattices to extrapolate fracture geometries into larger domains
based on a small sample (Fig. 4b). This model was developed
mainly for generating cross-cutting fracture sets. For more com-
plex networks involving abutting relations, the random walk tech-
nique proposed by Horgan and Young [136] for simulating
polygonal crack patterns in soil may be employed. Furthermore,
the invasion percolation method that has been used to model
channel networks [137] may provide a way to simulate some
highly branched and tortuous fracture systems.
The stochastic DFN method, in essence, treats problems in a
probabilistic framework and regards the real physical system as
one possibility among simulated realisations sharing the same
statistics. Hence, a sufficient number of realisations based on a
Monte Carlo process are necessary to predict a bounded range. In
156 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176
practice, a balance exists between the benefits of collecting
detailed information to create more representative DFNs and the
increased cost of field measurements. The uncertainty can be
reduced by constraining the random process with deterministic
data, e.g. forcing the 3D DFN generator to reproduce a 2D trace pat-
tern such as one exposed on tunnel walls [138]. Calibration and
validation of stochastic DFN models are important for solving real
problems and can be conducted based on the in-situ data from
field mapping and/or hydraulic tests [138–144]. The random nat-
ure of the stochastic DFN method may be regarded as an advanta-
geous aspect, because uncertainty is unavoidable when analysing
complex geological systems, for which single-valued predictions
from deterministic methods may be more risky [7]. However, it
is still very important to continue improving the realism and accu-
racy of stochastic DFN models, because the predicted results (e.g.
permeability, strength, fragmentation, etc.) featured by upper
and lower bounds based on unrealistic DFNs may be systematically
biased from the truth. Understanding of the variability and uncer-
tainty in data collection and DFN generation are thus critical for
engineering applications [145]. Developments are needed towards
(i) a more thorough characterisation of the underlying statistics,
such as multifractals for which a single scaling exponent is not suf-
ficient [110], and (ii) a more precise and efficient generator to cre-
ate DFNs respecting more details of real fracture systems, such as
the diversity of individual fracture shapes and morphology [9],
the topological complexity in fracture populations [114,146]
and the correlation between geometrical properties [129,147,148].
The important difference between 2D and 3D fracture networks
with respect to connectivity and permeability [64,90,105] renders
another advantage of the stochastic method, which has an intrinsic
capability of generating 3D networks.
2.3. Geomechanically-grown fracture networks
Extensive studies have been conducted to interpret the geolog-
ical history and the formation mechanism behind field observa-
tions of natural fracture systems (e.g. patterns, statistics and
minerals) [8,9,28–30,149]. The increased knowledge of fracture
mechanics [5] promoted the development of geomechanically-
based DFN models that incorporate the physics of fracture growth
and simulate fracture network evolution as a geometrical response
to stress and deformation. By applying a geologically-inferred
palaeo-stress/strain condition, natural fracture patterns may be
reproduced by such a DFN simulator that progressively solves
the perturbation of stress fields and captures the nucleation, prop-
agation and coalescence of discrete fractures. Different numerical
methods have been proposed and the one based on linear elastic
fracture mechanics (LEFM) is frequently adopted.
In the LEFM model, fracture patterns can be simulated by four
main steps in an iterative fashion [150,151]: (1) generation of ini-
tial flaws to mimic the process that natural fractures initiate from
microcracks, (2) calculation of the perturbed stress field in the rock
caused by the presence and evolution of fractures under an
imposed boundary condition, (3) derivation of the stress intensity
factor (KI) at the tip of each fracture, and (4) propagation of
fractures which satisfy a growth criterion, e.g. a subcritical law
KO 6 KI 6 KIC, where KO is the stress corrosion limit and KIC is the
material toughness [152]. The stress field and stress intensity fac-
tor can be calculated based on analytical solutions [150] or (most
commonly) numerical methods such as the boundary element
method (BEM) [153] and finite element method (FEM) [151,154].
The propagation length in each growth iteration can be derived
according to a power law relation with the energy release rate G
(related to KI) associated with a velocity exponent j (or subcritical
index n) [152], while the propagation angle may be computed if
the curvature and coalescence effects are important, especially
when the tectonic stress field is quite isotropic [29].
The development of fracture networks is a sophisticated
feedback-loop process, in which the complexity of growth dynam-
ics is directly related to the complexity of the developing struc-
tures. Specifically, the propagation of a fracture is influenced by
the mechanical interaction with others, and the propagated frac-
ture geometries can conversely generate stress perturbations into
the system. The mechanical interaction of fractures was found
strongly dependent on the velocity exponent j: an increased j
tends to promote a localised fracture pattern [150,153,155–157].
The fracture pattern evolution is also affected by the density
[150,157] and orientation [158] of the initial flaws, and the 3D
layer confinement effect [155,156]. Such a fracture mechanics
model has been developed to mimic the evolution of a 2D single
set of straight fractures (Fig. 5a) [150,153,155], 2D orthogonal sets
of straight fractures [157], 2D curved fracture patterns (Fig. 5b)
[29,151,156,158,159], and 3D curved fracture geometries (Fig. 5c)
[160]. The generated DFN pattern can be further used to evaluate
the connectivity [91,157], permeability [161] and solute transport
properties [162] of natural fracture systems.
Apart from the LEFM approach, fracture patterns can also be
simulated using other numerical methods. Cowie et al. [163,164]
developed a lattice-based rupture model to simulate the anti-
plane shear deformation of a tectonic plate and the spatiotemporal
evolution of a multifractal fault system. Tang et al. [165] used a
damage mechanics FEM model to simulate the evolution of paral-
lel, laddering or polygonal fracture patterns formed under different
boundary conditions, i.e. uniaxial, anisotropic and isotropic tec-
tonic stretch, respectively. Spence and Finch [166] employed the
discrete element method (DEM) to simulate the fracture pattern
development in a sedimentary sequence embedded with stratified
nodular chert rhythmites. Asahina et al. [167] coupled the finite
volume multiphase flow simulator (i.e. TOUGH2) and a lattice-
based elasticity and fracture model (i.e. Rigid-Body-Spring Net-
work) to simulate the desiccation cracking in a mining waste mate-
rial under a hydromechanically coupled process. The finite-discrete
element method (FEMDEM) has also been used to simulate the for-
mation and evolution of geological fracture patterns [168,169].
The geomechanically-grown DFN model, as a process-oriented
approach, has the advantage of linking the geometry and topology
of fracture networks with the conditions and physics of their for-
mation. Another merit is the automatic correlation between the
geometrical attributes of individual fractures (e.g. length, orienta-
tion, aperture and shear displacement) linked by the governing
physics. To solve practical problems, such a DFN generator can
be constrained by the measurement of rock properties (e.g. the
subcritical index measured from core samples) and the informa-
tion of geological conditions (e.g. stress, strain, pore pressure and
diagenesis) to achieve rational predictions [170]. However, diffi-
culty and uncertainty still exist in creating fracture patterns con-
sistent with the real systems for which coupled tectonic,
hydrological, thermal and chemical processes may be involved.
2.4. Summary with a view towards HM modelling
The three types of DFN models have each its own strength and
also suffer from some limitations, as listed in Table 1. The
geologically-mapped DFNs can preserve many realistic features
of fractures but it is hard to apply this method to characterise deep
rocks and 3D structures. The stochastic DFN approach has the mer-
its of simplicity and efficiency as well as applicability for 3D prob-
lems. However, the strong geological hypotheses of fracture
geometries and topologies used in its construction may ignore
some important underlying mechanical and tectonic constraints
and thus possibly lead to errors or large uncertainties. The geome-
 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176 157

Fig. 5. Geomechanically-grown DFN patterns based on linear elastic fracture mechanics: (a) evolution of a 2D fracture set [150], (b) development of a 2D polygonal fracture
pattern [159] and (c) growth of 3D layer-restricted fractures [160].

Table 1
Comparison of different approaches for DFN geometrical representation.

Numerical Key inputs Strengths Limitations

models
Geological DFNs Analogue mapping, borehole imaging, aerial (i) Preservation of geological (i) Limited feasibility for application to deep
photographs, realisms. rocks.
LIDAR scan or seismic survey (ii) Deterministic description of a frac- (ii) Difficulty in building 3D structures.
ture system. (iii) Constraints from measurement scale and
resolution.
Stochastic DFNs Statistical data of fracture lengths, orientations, (i) Simplicity and convenience. (i) Oversimplification of fracture geometries
locations, (ii) Efficient generation. and topologies.
shapes and their correlations (iii) Applicability for both 2D and 3D. (ii) Uncertainties in statistical parameters.
(iv) Applicability for various scales. (iii) Requirement of multiple realisations.
(iv) Negligence of physical processes.
Geomechanical Palaeostress conditions, rock and fracture (i) Linking geometry with physical (i) Uncertainties in input properties and tec-
DFNs mechanical mechanisms. tonic conditions.
properties (ii) Correlation between different frac- (ii) Large computational time.
ture attributes. (iii) Negligence of hydrological, thermal and
chemical processes.

chanical DFN models, which may capture some mechanical charac-
teristics of natural fractures, are sensitive to the assumed/mea-
sured rock properties and inferred palaeostress fields. If they are
to be improved to become the preferred useable model, coupling
with hydrological, thermal and chemical mechanisms may be
needed to reproduce actual geological systems. To generate better
fracture networks, a future research direction that is attracting
much effort is the development of hybrid DFN models that assim-
ilate the advantages of different approaches. Some of the current
DFN models have already exhibited some of such features. For
example, Kattenhorn and Pollard [59] used mechanical simulation
to correct the 3D fault structures interpreted from seismic survey
data. In the sequential stochastic DFN model proposed by Davy
et al. [135], fractures develop following the subcritical growth
law and their interactions are governed by an arrest mechanism.
In the self-referencing DFN model by Lei et al. [40], a
geologically-mapped fracture pattern is populated to larger
domains based on a random walk algorithm that preserves key
geometric and geomechanical attributes.
The created DFN geometries can serve as important inputs for
modelling the geomechanical and/or hydrological behaviour of
fractured rocks. To reduce the numerical difficulties induced by
158 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176
DFN complexities whilst capturing the key characteristics, simpli-
fications may be applied. In conventional hydrological studies,
inactive fractures (i.e. dead-end or isolated fractures) and very
small fractures are often removed from original DFNs based on
an assumption of their negligible contribution to the overall con-
nectivity and permeability [67,87,95,99]. Such treatments may be
applicable for low permeability rocks with well-connected fracture
topologies, inside which the matrix flow can be neglected. How-
ever, when the fractures are poorly connected and/or the rock
matrix is highly permeable, the effects of fracture-matrix fluid
transfer and flow in the matrix can significantly affect (or even
govern) the hydrological processes in fractured rocks [171] and
thus the features of inactive fractures may need to be preserved.
Attentions may also be drawn when setting a truncation threshold
for deleting small-sized fractures which were commonly believed
to have little contribution to the flow. However, several studies
have suggested that when a > D + 1, the connectivity of fracture
networks is ruled by fractures much smaller than the system size
[131,172] and thus the removal of very small fractures may lead
to important biases. Apart from these potential hydrological
effects, even more importantly, the presence of inactive fractures
may also influence the mechanical behaviour of fractured rocks
(e.g. elastic or brittle deformation). New fractures may propagate
from the dead-end tips of pre-existing fractures under stress con-
centrations and link with other fractures to form critical pathways
for fluid flow. More discussions about choosing appropriate DFNs
for geomechanical and HM modelling will be presented in the fol-
lowing sections.
3. Modelling of the geomechanical behaviour of fracture
networks
The numerical methods for geomechanical modelling of frac-
tured rocks can be categorised as continuum and discontinuum
approaches with the classification based on their treatment of dis-
placement compatibility [12]. The preference for a continuum or
discontinuum modelling scheme depends on the scale of the prob-
lem and the complexity of the fracture system [22]. The continuum
approach has the advantage of its greater efficiency to handle
large-scale problems, whereas the discontinuum method can more
straightforwardly incorporate complex fracture networks and
model multi-fracturing and fragmentation processes. In this sec-
tion, commonly used models for simulating geomechanical beha-
viour of fractured rocks will be reviewed: continuum/extended-
continuum, block-type discontinuum, particle-based discontin-
uum and hybrid finite-discrete element approaches. It is worth
mentioning that the classification here is not intended to be abso-
lute, since the boundary between the continuum and discontin-
uum methods has become very vague. Some advanced
continuum techniques have included contact algorithms and frac-
ture mechanics to consider discontinuities, while many discontin-
uum models are able to deal with continuous deformations.
3.1. Continuum/extended-continuum models
The conventional continuum approach treats a rock domain as a
continuous body that can be solved by the finite element method
(FEM) or finite difference method (FDM). It may be applicable to
a fractured rock with only a few or a large number of fractures
[12]. If the system consists of only a few discontinuities associated
with only a small amount of displacement/rotation, the discrete
fractures can be modelled by special ‘‘interface elements” (or ‘‘joint
elements”) that are forced to have fixed connectivity with the solid
elements [173]. However, such a treatment renders difficult to
handle the dynamics and large displacement problems of natural
fracture systems. When the density of DFN fractures is very high,
the modelling domain may be divided into a finite number of grid
blocks assigned with equivalent properties derived from
homogenisation techniques (Fig. 6a). The equivalent properties,
such as bulk modulus and strength parameters, are usually calcu-
lated using empirical formulations that consider the degradation
effect caused by pre-existing fractures [174,175] or analytical solu-
tions based on the crack tensor theory [176,177]. The crack tensor
theory calculates the volume averaged parameters of all fractures
with respect to their geometrical properties (e.g. length, orienta-
tion and aperture) and was extended to consider the coupling
between stress and fluid flow [46,178]. Such a crack tensor method
has been integrated into the FEM [179,180] and FDM [101,181] sol-
vers to model the geomechanical and HM behaviour of fractured
rocks. The simulation results may be sensitive to the grid block dis-
cretisation especially when the block size is significantly smaller
than the representative elementary volume (REV) [101]. The
homogenisation-based continuum model may not adequately con-
sider the connectivity effect of very long fractures that penetrate
numerous grid blocks and the results may be even worse if aper-
tures are very heterogeneous and positively correlated with frac-
ture sizes. Thus, it may not be applicable to a fractal fracture
system with high variability in its density distribution (i.e. a small
fractal dimension D) and/or a large proportion of long fractures
having a size comparable to the problem domain (i.e. a small
power law length exponent a). The two conventional continuum
schemes (i.e. the one employing interface elements and the one
using homogenised properties) may be combined to explicitly
model large discontinuities (e.g. dominant faults) using interface
elements and then to characterise each isolated block as contin-
uum bodies with their bulk properties parameterised according
to the distribution of small fractures. Furthermore, the crack tensor
method cannot consider the interaction between fractures and
blocks as well as the resulting localised deformation and damage
in the rock.
To more explicitly capture the effects of discrete fractures, an
extended-continuum model has been developed by assuming frac-
tures to have certain numerical width (for connectivity preserva-
tion) and representing them as arrays of grid elements with
softening and weakening properties based on a very fine finite dif-
ference mesh (Fig. 6b) [41,182]. It treats the fractured rock as a
composite elasto-plastic solid system, in which the failure of intact
rocks or stress-displacement behaviour of fractures can be mod-
elled by a Mohr-Coulomb criterion with tension cut-off. Similar
‘‘weak material” representation of fractures has also been imple-
mented in the rock failure process analysis (RFPA) code (a damage
mechanics FEM model) [183] and the cellular automation model
[184]. Such a composite continuum model with explicit DFN repre-
sentations may be more suitable for simulating weakly cemented
fractures (i.e. mineral filled veins), whereas the physical rationale
is not intuitive if being applied to unfilled discontinuities having
clean wall surfaces. Furthermore, another important extended-
continuum model is the extended finite element method (XFEM),
which can represent fracture propagations without re-meshing
and has been recently developed for simulating fluid-driven frac-
turing [185,186].
3.2. Block-type discontinuum models
The block-type discontinuum models include the distinct ele-
ment method (DEM) with an explicit solution scheme and the dis-
continuous deformation analysis (DDA) method with an implicit
solution form. In this discontinuum modelling framework, the frac-
tured rock is represented as an assemblage of blocks (i.e. discrete
elements) bounded by a number of intersecting discontinuities.
The geometry of the interlocking block structures can be identified
 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176
Fig. 6. (a) A continuum modelling scheme: a fracture network is divided into a finite number of grid blocks with equivalent properties determined analytically or numerically
[101] and (b) an extended continuum modelling scheme: the domain is discretised by a regular finite difference grid and fractures are represented by softening and
weakening the grid elements intersected by fracture traces [41].
first by e.g. employing the techniques of combinatorial topology
[187]. In the subsequent mechanical computations, these blocks
can be treated as rigid bodies or deformable subdomains (further
discretised by finite difference/volume grids) with their interac-
tions continually tracked by spatial detections during their defor-
mation and motion processes.
3.2.1. Distinct element method (DEM)
The DEM method was originated by Cundall [188,189] and
gradually evolved to the commercial codes UDEC and 3DEC for
solving 2D and 3D problems, respectively [190,191]. Its basic com-
putational procedure can be summarised as four steps [192]: (1)
the contacts between blocks are identified/updated through a
space detection, (2) the contact forces between discrete bodies
are computed based on their relative positions, (3) the acceleration
induced by force imbalance for each discrete element is calculated
using Newton’s second law, and (4) the velocity and displacement
are further derived by time integration with new positions deter-
mined. An explicit time marching scheme is applied to solve the
problem iteratively until the block interaction process to be simu-
lated has been completed. The mechanical interaction between
blocks is captured by a compliant contact model that accommo-
dates virtual ‘‘interpenetrations” governed by assumed finite stiff-
nesses to derive normal and tangential contact forces. The
empirical joint constitutive laws derived from laboratory experi-
ments [14,193] can be implemented into the interaction calcula-
tion in an incremental form to simulate joint normal and
159
shearing behaviour [194–197]. A viscous damping parameter
may be introduced to reduce dynamic effects for modelling
quasi-static conditions [198].
The DEM approach is able to capture the stress-strain character-
istics of intact rocks, the opening/shearing of pre-existing fractures
and the interaction between multiple blocks and fractures. Com-
bined with DFN models, it has been widely applied to study the
mechanical behaviour of fractured rocks. Zhang and Sanderson
[44] studied the critical deformation behaviour of fractured rocks
and observed an abrupt increase in the deformability associated
with a power law scaling when the fracture density exceeds the
mechanical percolation threshold (slightly higher than the geo-
metrical threshold) (Fig. 7). Min and Jing [199] examined the scale
dependency of the equivalent elastic properties of a fractured rock
based on multiple DFN realisations conditioned by the same frac-
ture statistics (Fig. 8). In their study, a technique to derive the
fourth-order elastic compliance tensor has also been developed
for equivalent continuum representations. Min and Jing [200] fur-
ther found that the equivalent mechanical properties (i.e. elastic
modulus and Poisson’s ratio) of fractured rocks may also be stress
dependent. With the increase of stress magnitudes, the equivalent
elastic modulus significantly increases, while the Poisson’s ratio
generally decreases but can be well above 0.5 (i.e. the upper limit
for isotropic materials). Noorian-Bidgoli et al. [201] extended this
DEM-DFN model to a more systematic framework to derive the
strength and deformability of fractured rocks under different load-
ing conditions, which is further applied to study the anisotropy
160 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176

Fig. 7. Deformation of fractured rocks with (a) a relatively low fracture density of 5.25 m/m2 and (b) a high fracture density of 7.77 m/m2 under a uniaxial compression
loading. (c) The deformability Bs of the fractured rock exhibits a power law scaling behaviour when the fracture density d exceeds the mechanical percolation threshold of
6.5 m/m2, which is higher than the geometrical threshold of 4.0–5.5 m/m2 of the study networks [44].

Fig. 8. Variation of the elastic modulus of fractured rocks with the increase of the model size. The ratio of shear stiffness to normal stiffness of fractures is assumed to be 0.2.
Results are computed using the block-type DEM simulator (UDEC) based on multiple DFN realisations [199].

[202] and randomness [203] of the strength/deformability of
stochastic DFNs. Recently, Le Goc et al. [204] integrated 3D DFNs
into the 3DEC simulator and investigated the effects of fracture
density, sizes and orientations on the magnitude and scaling of
the equivalent elastic modulus of fractured rocks. Havaej et al.
[205] conducted 3D DEM-DFN modelling to analyse the effects of
geological structures on the quarry slope failure mechanism. In
addition, a considerable number of similar DEM-DFN models have
been developed where the main motive is to study the effect of
stresses on fluid flow. Such models were applied to explicitly cap-
ture the fracture opening, closing, shearing and dilational charac-
teristics in complex fracture networks under in-situ stresses,
after which the fluid flow implications were investigated [43–45,
96,98,206–209] (more details are presented in Section 4). Further-
more, some models that incorporate the pore fluid pressure effect
show that the pore pressure level can also significantly influence
the mechanical and hydrological properties of fractured rocks
[47,210].
The classic DEM formulation cannot simulate the propagation
of new fractures in intact rocks driven by stress concentrations,
although plastic yielding can capture some aspects of the rock
mass failure process [211]. Such a shortcoming was recently
addressed by introducing a Voronoi or Trigon discretisation in
matrix blocks that allows fracturing along the internal ‘‘grain”
boundaries governed by tensile and shear failure criteria [212–
216]. The DFN representation can also be integrated into such a
DEM model using a dual-scale tessellation, in which the primary
grid represents natural fractures and the secondary discretisation
mimics microstructures [217]. The uncertainty of the Voronoi/Tri-
gon DEM model due to mesh dependency indicates the necessity of
realistic representation of natural grain shapes and textures
[218].
 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176
Fig. 9. (a) A fractured rock with a geologically-mapped DFN pattern and (b) its dynamic collapsing process modelled by the DDA method [227].
3.2.2. Discontinuous deformation analysis (DDA)
The DDA method was proposed by Shi and Goodman [219,220]
to compute the deformation and motion of a multi-block system.
The discretisation of a DDA system is quite similar to the one for
the DEM, i.e. the medium is dissected into blocks by intersecting
discontinuities. However, a fundamental difference between the
two methods lies in their computational frameworks. The DEM
treats the kinematics of each block separately based on an explicit
time-marching scheme, while the DDA calculates the displacement
field based on a minimisation of the total potential energy of the
whole blocky system and an implicit solution to the established
system of equations through a matrix inversion. Thus, the DDA
method has an important advantage of fast convergence with
unconditional numerical stability compared to the DEM method
that requires a time step smaller than a critical threshold [12].
Important extensions of the original DDA method include the finite
element discretisation of rock matrix [221], the sub-block tech-
nique (similar to the Voronoi DEM) for simulating fracturing pro-
cesses [222], the formulation for modelling coupled solid
deformation and fluid flow [223,224], and the development of 3D
models [225]. The mechanical behaviour of fractured rocks has also
been investigated based on DDA-DFN simulations, with emphasis
on studying the stability of underground excavations and slope
engineering (Fig. 9) [226–228]. Recently, Tang et al. [229] com-
bined the RFPA (a damage mechanics FEM model) with the DDA
method to capture crack propagations and block kinematics of a
rock slope system.
3.3. Particle-based discontinuum models
The particle-based discontinuum model was originally intro-
duced by Cundall and Strack [230] to simulate granular materials
such as soils/sands and gradually evolved to a commercial code,
i.e. particle flow code (PFC) [231]. Similar to the block-type DEM
method, PFC calculates the inertial forces, velocities and displace-
ments of interacting particles by solving Newton’s second law
through an explicit time-marching scheme. The discrete particles
are assumed to be rigid with a circular (in 2D) or spherical (in
3D) shape, and can have variable sizes which are usually much lar-
ger than the physical grain scale. To extend PFC to model rock
materials, Potyondy and Cundall [232] developed a bonded-
particle model (BPM), in which intact rocks are represented as
assemblages of cemented rigid particles and the macroscopic frac-
turing is simulated as a result of the breakage and coalescence of
the inter-particle cohesive bonds. The interaction between parti-
cles can be characterised by two types of bond models in PFC, i.e.
the contact bond model and the parallel bond model. A contact
bond serves as a linear elastic spring with normal and shear
161
strengths, and transmits forces via the contact point between
two particles. A parallel bond with certain normal and shear stiff-
nesses joins two particles to resist against separation under ten-
sion, shear and rotation. The parallel bond model is more
suitable for simulating rock materials as it can capture the
cement-like behaviour with resistance to bending moments. To
overcome the original deficiency of PFC in reproducing a realistic
rock strength ratio (i.e. the ratio of uniaxial compressive strength
to tensile strength) and macroscopic friction angle, a hierarchical
bonding structure can be built based on a cluster logic [232] or a
clump logic [233]. The cluster/clump approach mimics the inter-
locking effect of irregular grains by defining a higher value of
intra-cluster bond strength (i.e. the strength between particles in
the same cluster) than that of the strength between cluster
boundaries.
The BPM representation using particles with idealised circular/-
spherical shapes can introduce unphysical asperities on disconti-
nuity surfaces, resulting in an additional resistance to frictional
sliding. To suppress such an artificial roughness effect, a smooth-
joint contact model (SJM) was proposed to simulate fracture wall
behaviour based on the actual geometry and morphology of dis-
continuities and independent of the arrangement of local particles
in contact [234]. The smooth contact is assigned to all particle pairs
lying on the fracture interface but belonging to opposite matrix
blocks, so that they can overlap and pass through each other
(Fig. 10a). The contact forces are calculated based on the relative
displacements and the smooth-joint stiffness in the normal and
tangential directions of the local surfaces. The SJM was found to
be able to capture the shear strength and dilational behaviour of
natural fractures associated with significant scale effects
[235,236]. By simulating intact rock matrix using the BPM and dis-
crete fractures using the SJM, a synthetic rock mass (SRM) mod-
elling approach (Fig. 10b) has been developed to characterise the
mechanical properties of fractured rocks including peak strength,
damage, fragmentation, brittleness, anisotropy and scale effects
[234]. Compared to the Hoek-Brown empirical approach for pre-
sumed isotropic rock masses, the SRM method that integrates
explicit DFN representations has an advantage to derive the
orientation-specific strength of naturally fractured rocks and con-
sider the influence of fracture length distribution and connectivity
[237]. The SRM model has been applied to reproduce the failure
behaviour of veined core samples under uniaxial compression tests
[238], to estimate the mechanical REV of a jointed rock mass near
an underground facility [239], to simulate fracture propagation in
jointed rock pillars [240,241], and to evaluate the stability of
wedges around a vertical excavation in a hard rock [242].
An open source particle-based modelling platform named YADE
[243,244] has recently been developed as an alternative to the
162 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176

Fig. 10. Integration of (a) a smooth-joint contact model in PFC to achieve (b) synthetic rock mass (SRM) modelling of fractured rocks with stochastic DFN geometries [234].

Fig. 11. Integration of (a) a fractal DFN into (b) the YADE bonded-particle model (BPM) for mechanical modelling of fractured rocks [134].

commercial code PFC. YADE represents intact rocks using glued
discs/spheres and models the fracturing process based on the rup-
ture of inter-particle bonds with the contact bond algorithms fol-
lowing a similar logic to PFC. To reproduce the high ratio of the
compressive strength to the tensile strength and the non-linear
failure envelope of brittle rocks, the concept of ‘‘interaction range”
was introduced by Scholtès and Donzé [245]. They mimic the
microstructural complexity by assembling constitutive particles
in neighbouring zones (not only the particles in direct contact). A
joint contact logic equivalent to the SJM has also been imple-
mented in YADE to avoid the particle interlocking effect between
sliding fracture surfaces [246]. By integrating 3D fractal DFNs into
the YADE BPM model associated with the smooth joint contact
treatment (Fig. 11), Harthong et al. [134] studied the influence of
fracture network properties (i.e. fractal dimension D, power law
length exponent a and fracture intensity P32) on the mechanical
behaviour of fractured rocks. The strength and elastic modulus of
rock masses decrease if P32 increases (i.e. more fractures) or a
decreases (i.e. higher proportion of larger fractures), while the spa-
tial heterogeneity and scaling of the mechanical properties are
 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176
affected by D. Such a combined BPM-DFN model has also been
applied to analyse the stability of fractured rock slopes, which is
controlled by the strengths of both pre-existing fractures and
intact rocks [247,248].
3.4. Hybrid finite-discrete element models
The hybrid finite-discrete element method (FEMDEM or FDEM)
combines the finite element analysis of stress/deformation evolution
with the discrete element solutions of transient dynamics, contact
detection and interaction. In such a discontinuum modelling scheme,
the internal stress field of each discrete matrix block is calculated by
the FEM solver, while the translation, rotation and interaction of
multiple rock blocks are traced by the DEM algorithms. Pre-
existing fractures in rocks are treated as the internal boundaries of
rock volumes. The FEMDEM approach also provides a natural solu-
tion route to modelling the transitional behaviour of brittle/quasi-
brittle materials from continuum to discontinuum (i.e. fracturing
processes) by integrating fracture mechanics principles into the for-
mulation. This section will review the two most commonly used
FEMDEM models, i.e. the commercial software ELFEN [249] and an
open source platform Y-code [250], which have been broadly used
to simulate the mechanical processes in geological media containing
pre-existing discontinuities. Although the two FEMDEM models
originated from the same concept [251], a fundamental difference
exists in their discontinuum implementations: ELFEN is based on a
partially-discontinuous discretisation with joint elements inserted
only in pre-existing or propagated fractures, while Y-code is based
on a fully-discontinuous mesh where joint elements are embedded
between the interfaces of all finite elements (i.e. along fractures
and inside intact domains). Some other differences also lie in the
computation of intact material deformation and determination of
local failure. For more details, the reader is referred to the work by
Lisjak and Grasselli [25].
3.4.1. ELFEN
The ELFEN code models the degradation of an initial continuous
domain into discrete bodies by inserting cracks into a finite ele-
Fig. 12. Integration of DFN geometries into the FEMDEM model of ELFEN for modelling strength of (a) a prefractured pillar [256] (note: rci is the uniaxial compressive
strength of intact rocks, P21 denotes the fracture intensity, i.e. total length of fractures per unit area) and (b) an open pit slope [265].
163
ment mesh. A nodal fracture scheme was introduced by construct-
ing a non-local failure map for the whole system [252]. The
feasibility of local failure is determined based on the evolution of
nodal damage indicators. The fracturing direction (if failure occurs)
is calculated based on the weighted average of the maximum fail-
ure strain directions of all surrounding elements. A new discrete
fracture is then inserted along the failure plane with the local mesh
topology updated through either the ‘‘intra-element” or ‘‘inter-
element” insertion algorithm with adaptive mesh refinement
applied if necessary [253]. ELFEN provides various material consti-
tutive models including the elastic, elasto-plastic and visco-plastic
laws, and many brittle/quasi-brittle failure models including the
rotating crack model, the Rankine material model, and the com-
pressive fracture model (i.e. Mohr-Coulomb failure criterion cou-
pled with a tensile crack model) [253,254]. The capability of
ELFEN for modelling 3D fracturing processes has also been demon-
strated through the simulation of laboratory strength tests [255].
The explicit DFN fracture geometries generated from e.g. FracMan
can be imported into the ELFEN platform by embedding fracture
entities into rock solids and representing each fracture as opposed
free surfaces [256,257]. To mesh such complex systems, special
geometrical treatments may be involved to avoid ill-posed ele-
ments caused by subparallel fractures intersecting at a very small
acute angle or a fracture tip terminating at the vicinity of another
fracture [249]. Both pre-existing and newly propagated fractures
are assigned with contact properties, e.g. fracture stiffness and fric-
tion coefficient, to simulate solid interactions through discontinu-
ity surfaces [257]. The degradation of natural fractures during
shearing can also be modelled by introducing roughness profiles
[258].
The combined FEMDEM-DFN model has been applied to tackle
the geomechanical problems for various engineering applications
[259]. The presence of natural fractures may dominate the strength
of slender pillars but have a reduced influence for wider pillars
(Fig. 12a) [256]. The orientation and length distribution of DFN
fractures also affect the failure mode of the pillar structures, which
can exhibit splitting with lateral kinematic releases or shearing of
critically inclined pre-existing fractures linked by new cracks
164 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176
through intact rock bridges [260]. This synthetic numerical model
has also been used to investigate the progressive failure of rock
slopes (Fig. 12b) [261,262], which in reality is usually triggered
by both the reactivation of natural fractures and the propagation
of new cracks [263]. The FEMDEM model is well suited to mimic
the staged failure processes of rock slopes including initiation,
transportation/comminution and deposition, which involve yield-
ing and fracturing of intact materials, shearing of fracture surfaces
and translational/rotational instabilities [264]. The rock mass fab-
rics and rock bridge properties can have important influences on
the stability of large-scale open pit slopes [261]. The caving-
induced rock mass deformations and associated surface subsidence
may be controlled by the orientation of joint sets and the location/
inclination of major faults [265]. The FEMDEM-DFN model has also
been applied to capture the progressive failure of the roof strata of
a coal mine tunnel [266]. All these engineering applications high-
lighted the advantage of the FEMDEM-DFN technique with the
capability of simulating the reactivation/interaction of pre-
existing fractures and initiation/propagation of new cracks.
3.4.2. Y-code
During the 1990s, many algorithmic solutions for 2D and 3D
FEMDEM simulation were developed by Munjiza et al. [251,267]
and Munjiza and Andrews [268,269]. Extensive developments
and applications of the FEMDEM model have been conducted after
the release of the open source Y-code [250], and different versions
have emerged including the code developed collaboratively by
Queen Mary University and Los Alamos National Laboratory
[270–272], the Y-Geo program by Toronto University
[25,273,274], and the VGeST (recently renamed ‘‘Solidity”) plat-
form by Imperial College London [15,39,275–281]. The Y-code
FEMDEM model accommodates the finite strain elasticity coupled
with a smeared crack model and is able to capture the complex
behaviour of discontinuous rock masses involving deformation,
rotation, interaction, fracturing and fragmentation.
In the Y-code, the fractured rock is represented by a discontin-
uous discretisation of the model domain using three-noded trian-
gular (in 2D) or four-noded tetrahedral (in 3D) finite elements
and four-noded (in 2D) or six-noded (in 3D) joint elements embed-
ded at the interface between finite elements. An important differ-
ence with the ELFEN code is that the joint elements are inserted for
all edges (in 2D) or surfaces (in 3D) of finite elements. The defor-
mation of the bulk material is captured by the linear-elastic
constant-strain finite elements with the impenetrability enforced
by a penalty function and the continuity constrained by a constitu-
tive relation [267], while the interaction of matrix bodies through
discontinuity interfaces is simulated by the penetration calculation
[269]. The joint elements are created and embedded between tri-
angular/tetrahedral element pairs before the numerical simulation,
and no further remeshing process is performed during later com-
putations. Pre-existing fractures can be represented by a series of
joint elements which are initially overlapped (but opposite sides
are separately defined) free surfaces [39]. The brittle failure of
intact materials is governed by both fracture energy parameters
(for mode I and mode II failure) and strength properties (e.g. tensile
strength, internal friction angle and cohesion) [25]. A numerical
calibration can be conducted to achieve consistency between the
input material strength parameters and simulated macroscopic
response [282]. Code development for modelling 3D crack propa-
gation has also been achieved by different research groups
[272,283–286].
The advantage of the FEMDEM model for simulating the degra-
dation of continuum into discrete pieces promoted the application
to tackle various engineering problems, such as rock blasting [287],
fracture development around excavations in isotropic/anisotropic
intact rocks [288–291] and mountain slope failure [292]. It has also
been used to model the mechanical behaviour of fractured rocks
embedded with pre-existing fractures. Lisjak et al. [293] integrated
DFN crack arrays into the FEMDEM model to imitate the anisotropy
of an argillaceous rock. The FEMDEM technique has been applied to
model geologically-mapped DFNs under various stress conditions
[15,39,40,280,294]. It can capture realistic geomechanical phe-
nomena such as deformation and rotation of matrix blocks, open-
ing, shearing, and dilation of pre-existing fractures as well as the
propagation of new cracks [281]. Due to the presence of natural
fractures, strongly heterogeneous stresses are distributed in the
rock mass (Fig. 13a) [39]. The in-situ stress can introduce new
cracks into the natural fracture system which potentially increases
the network connectivity (Fig. 13b) [15,280]. Highly variable aper-
tures can also be generated (Fig. 13c), which is related to the com-
plexity of DFN geometries and the condition of far-field stresses
[39,278,280]. The stress-dependent mechanical response of frac-
tured rocks (e.g. deformation and rotation of matrix blocks, open-
ing, shearing and dilation of pre-existing fractures and new crack
propagation) can significantly affect the fluid flow in fracture net-
works, which will be discussed in more details in Section 4.
3.5. Summary with a view towards HM modelling
As can be seen from above, geomechanical modelling of frac-
tured rocks can be achieved by continuum or discontinuum
approaches, which have important differences in the conceptuali-
sation of geological media and the treatment of displacement com-
patibility [12]. Table 2 presents a comparison of the continuum and
discontinuum models. The continuum modelling scheme mainly
reflects the material deformation of a geological system from a
more overarching view and attempts to bypass the geometrical
complexity by using specific constitutive laws and equivalent
material properties derived from homogenisation techniques.
However, it cannot adequately consider the effects of stress varia-
tions, fracture interactions, block displacements and rotations.
More importantly, the applicability of a homogenisation process
is based on the assumption of an REV, which may not exist for nat-
ural fracture systems [73]. On the other hand, the discontinuum
scheme treats the system as an assemblage of interacting individ-
ual components and permits the integration of complex constitu-
tive laws for rock materials and fracture interfaces. A
discontinuum model can be established at a specific scale of inves-
tigation without presuming the existence of an REV. However,
some of the input parameters (e.g. bonding strength, joint stiff-
ness) may need to be determined by indirect numerical calibra-
tions rather than from physical measurements. Furthermore, the
computational time for solving discontinuous problems can be
considerably larger than that for continuum models. To take
advantage of the two modelling technologies for tackling practical
issues, a discontinuum model may be used to derive the REV size
(if it exists) as the onset to treat a geological system as a
continuum.
In the conventional block-type DEM and DDA modelling, the
rock mass is dissected into blocks through topological identifica-
tions, during which the dead-ends and isolated fractures are usu-
ally removed for simplicity [44,199,206,221,227]. Such
treatments may be acceptable for modelling extremely hard rocks
or highly fragmented rocks, for which the effect of new crack prop-
agations may be negligible. However, for most rock types, fractures
are likely to propagate from the tips of pre-existing fractures dri-
ven by concentrated local stresses. This fracture propagation and
coalescence process can create new pathways for fluid migration
and thus affect the hydrological properties (e.g. permeability) of
fractured rocks, as has been revealed in numerical simulations
[15,39]. Furthermore, the curved natural fractures, which are often
assumed to be straight line (in 2D) or planar disc (in 3D) in the
 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176 165

Fig. 13. FEMDEM-DFN modelling results. (a) Heterogeneous distribution of local maximum principal stresses in fractured rocks represented by a geologically-mapped DFN
(the upper array of figures) and a statistically-equivalent random DFN (the lower array of figures) under in-situ stresses applied at different angles to the model [39]. (b)
Variation of fracture apertures in a geologically-mapped DFN network that accommodates further new crack propagations in response to a biaxial stress condition [15]. (c)
Highly inhomogeneous aperture distribution within a single fracture of a 3D persistent DFN system under a polyaxial stress condition [278].

Poisson DFN model, may accommodate localised dilations under
high differential stresses with localised increases in aperture and
this may have important HM effects [15,47,280]. Hence, towards
a better HM modelling of fractured rocks, the capability of pre-
cisely modelling the geomechanical response is prerequisite, for
which the important characteristics of fracture dead-ends and cur-
vatures need to be appropriately treated.
4. Modelling of the hydrological behaviour of fracture networks
under geomechanical effects
The presence of fractures can generate stress perturbations in
the rock, such as rotation of stress fields, stress shadows around
discontinuities and stress concentration at fracture tips [5]. The
resulting heterogeneous stress distribution may lead to variable
local normal/shear stresses loaded on different fractures having
distinct sizes and orientations, and produce various fracture
responses such as opening, closing, sliding, dilatancy and propaga-
tion. Since the conductivity of fractures is critically dependent on
the third power of fracture apertures [295], the geomechanical
conditions can considerably affect the hydrological properties of
fractured rocks including fluid pathways, bulk permeability and
mass transport [207]. Numerical models that integrate explicit
DFNs and non-linear rock/fracture behaviours provide powerful
(and so far irreplaceable) tools to investigate the geomechanical
effects on fluid flow in complex fracture networks [296].
4.1. Fluid pathways
Fracture networks usually serve as the major pathways for fluid
transport in subsurface rocks, especially if the matrix is almost
impermeable compared to the fractures [297]. The partitioning of
fluid flow within a fracture population relies on the spatial connec-
tivity of fracture geometries and the transmissivity of individual
fractures, both of which can be affected by the geomechanical
conditions.
Zhang et al. [45] used the UDEC DEM code to study the defor-
mation of a fractured rock based on a geologically-mapped DFN
pattern and found the closure of fractures under applied in-situ
stresses can re-organise the fluid pathways. They also found the
166 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176

Table 2
Comparison of different numerical models for DFN geomechanical modelling.

Numerical models Key inputs Strengths Limitations

Continuum models Equivalent material properties (i) Simplicity of geometries.
(ii) Efficient calculation.
(iii) Suitability for large-scale industrial
applications.
Block-type & Material properties for both fractures and (i) Explicit integration of DFNs.
particle-based rocks, damping coefficient, bonding strengths (ii) Simple particle/grain bonding logic.
discrete models (iii) Integrated constitutive laws for rocks/
fractures.
(iv) Capturing the interaction of multiple
fractures.
Hybrid FEMDEM Material properties for both fractures and (i) Explicit integration of DFNs.
models rocks, fracture energy release rate, damping (ii) Fracture propagation is based on both
coefficient the strength criterion and fracture
mechanics principles.
(iii) Integrated constitutive laws for rocks/
fractures.
(iv) Capturing the interaction of multiple
fractures.
(i) No consideration of fracture inter-
action, block displacement/inter-
locking/rotation.
(ii) Complexity in deriving equivalent
material parameters and constitu-
tive laws.
(iii) Valid only if an REV exists.
(i) Limited data on joint stiffness
parameters.
(ii) Calibration of input particle bond-
ing properties.
(iii) No fracture mechanics principle.
(iv) Large computational time.
(i) Limited data on joint stiffness
parameters.
(ii) Calibration of fracture energy
release rates.
(iii) Large computational time.

closed joints of one set can hinder fluids from passing through not
only themselves but open fractures of another set due to the net
effect. Zhang and Sanderson [43] applied the same technique to
different types of outcrop DFN patterns with/without systematic
fracture sets. The fluid flow in systematic fracture networks tend
to be dominated by the primary joint set if it is relatively open,
while for non-systematic networks containing fairly randomly ori-
ented small fractures, the flow channels tend to align the direction
of the maximum principal stress. Min et al. [206] incorporated the
fracture shear dilation behaviour in the DEM modelling of a
stochastic DFN having a power law distribution of fracture lengths
and a uniform distribution of initial (i.e. zero stress) apertures.
They observed that, under high differential stresses, a small portion
of fractures which have critical/near-critical orientations, good
connectivity and long lengths would dilate and form large flow
channels (Fig. 14a). The ‘‘critical orientations” here correspond to
a range of fracture orientations that would allow discontinuities
with no cohesive strength to slide under the given in-situ differen-
tial stress condition. The localised features would be augmented if
the initial apertures are broadly distributed (e.g. following a log-
normal distribution) and correlated with fracture lengths [98].
Latham et al. [15] employed the FEMDEM method combined with
a smeared crack model to study the geomechanical response and
fluid flow in an outcrop-based DFN. They found that bent natural
fractures under high differential stresses may exhibit evident dila-
tional jogs and can be linked by newly propagated cracks to form
major fluid pathways. Similar localised flow channels created by
the connection of pre-existing fractures have also been observed
by Figueiredo et al. [41] using a FDM simulator. Sanderson and
Zhang [47,96] calculated the fluid flow in the third dimension of
sedimentary rocks using an analytical pipe formula based on the
deformed 2D fracture networks. They found vertical flow becomes
extremely localised when the pore fluid pressure exceeds a critical
level and very large aperture channels emerge (especially in some
preferentially-oriented and well-connected curved fractures)
(Fig. 14b). The flow distribution also exhibits significant multifrac-
tality when the loading condition approaches the critical state. Lei
et al. [278] studied the fluid flow in a 3D persistent fracture net-
work under various polyaxial stress conditions. They observed a
transition from homogeneous flow under an isotropic stress field
to highly localised patterns under a deviatoric stress condition
(Fig. 14c). Recently, Lei et al. [280] also studied the impacts of
stress on reorganising vertical fluid flow through a 3D sedimentary
layer.
4.2. Permeability
There are two different notions of rock mass permeability, i.e.
equivalent permeability and effective permeability. The equivalent
permeability is defined as a constant tensor in Darcy’s law to rep-
resent flow in a heterogeneous medium, while the effective perme-
ability is an intrinsic material property based on the existence of an
REV at a large homogenisation scale [298]. Permeability here
mainly refers to the equivalent permeability of a fractured rock
at a specific study scale.
The permeability tensor was observed to be highly dependent
on both the geometrical attributes of fracture networks (e.g. den-
sity, lengths and orientations) and the in-situ stress conditions
(e.g. direction, magnitude and ratio of the principal stresses) [45].
When the differential stress ratio is relatively low, the permeability
decreases with the increase of burial depth (or mean stress) of the
fractured rock due to the closure of most fractures [43,206]. The
non-linear relationship between normal stress and fracture closure
results in a phenomenon that the permeability is more sensitive at
shallower depths (i.e. smaller mean stresses) and approaches a
minimum value when most fractures are closed to their residual
apertures under high mean stresses (Fig. 15a) [206]. The perme-
ability anisotropy of a fracture network with non-systematic frac-
tures is more dependent on the ratio and direction of the applied
principal stresses than that of a network with systematic fracture
sets which is more controlled by the fracture set orientations
[43]. With the increase of differential stresses, the permeability
exhibits a decrease and then an abrupt increase separated by a crit-
ical stress ratio that begins to cause continued shear dilations
along some preferentially oriented fractures (Fig. 15b) [206]. A
similar variation of permeability occurs when the pore fluid pres-
sure is elevated [41]. Simultaneously, the permeability anisotropy
is also enlarged by the increased stress ratio [206]. If initial aper-
tures are correlated with fracture lengths, the permeability of frac-
tured rocks is dominated by larger fractures with wider apertures.
This model tends to exhibit a permeability value much higher than
the model assigned with a constant initial aperture (Fig. 15c) [98].
 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176 167

Fig. 14. Fluid pathways in fracture networks under in-situ stresses. (a) With the increase of the boundary stress ratio, fluid flow becomes more concentrated in only part of
the fractures in the network due to the shear dilation effect [206]. (b) The vertical fluid flow through a jointed layer exhibits a highly localised pattern when the fractured rock
is deformed under a critical stress state [47]. (c) The vertical flow in a 3D persistent fracture network shifts from a homogeneous to localised pattern with the increase of the
horizontal stress ratio [278].

The permeability tensor may be destroyed but then reestablished
with the increase of differential stresses [98]. In addition to the
shear dilation of fractures, the increase of network connectivity
caused by brittle failure and crack propagation under geomechan-
ical loading can also significantly raise the permeability of frac-
tured rocks [15,41,151,157]. More interestingly, the emergence of
dilational jogs/bends associated with curved fractures in response
to high differential stresses may lead to even more increase of per-
meability in the third dimension [47,96,280] and result in a highly
anisotropic permeability tensor [278]. The permeability of 3D frac-
ture networks also exhibits a transition stage having steep growth
that occurs when the stress ratio is approaching the critical state
(Fig. 15d) [278]. It has to be mentioned that an increased fracture
density may not always lead to an increased permeability under
some tectonic conditions (e.g. an extensional regime), because
fractures may be closed due to mechanical interactions when they
are too densely spaced [299].
4.3. Transport
Mass transport in fractured rocks is governed by various mech-
anisms including advection, dispersion, matrix diffusion, interface
sorption and chemical reaction [17,297]. The heterogeneous fluid
velocity fields in geological media consisting of distributed frac-
tures and porous rocks can result in complex transport phenomena
in the system. Computational models employing solute compo-
nents or tracked particles have been developed to simulate migra-
tion processes in fractured rocks [6,300]. Recently, numerical
studies have also been conducted to investigate the effects of the
stress/deformation on the transport properties of fracture net-
works, as summarised below.
Zhao et al. [208] coupled the UDEC DEM code and a random
walk particle tracking code PTFR and investigated the stress effects
on the hydrodynamic dispersion of contaminant solutes in a
stochastic DFN system. They found that compressive stresses can
close fracture apertures and attenuate the dispersivity, but an
increased differential stress ratio could greatly intensify the
spreading phenomenon if it exceeds certain threshold for trigger-
ing shear dilations. Zhao et al. [209] extended this modelling tech-
nique to incorporate the effects of matrix diffusion and sorption,
and conducted a systematic study of the solute transport under
various stress conditions. Their results showed that the stress can
significantly affect the solute residence time in the fracture net-
work (Fig. 16a). When the stress ratio is increased but not very
high (<3), the breakthrough curve shifts to the right side (i.e. the
average residence time increases) compared to that of the initial
zero-stress condition, due to the closure of fractures under rela-
tively isotropic stresses. However, as the stress ratio exceeds 3,
fluid velocity is raised drastically in some dominant channels
formed by dilated fractures due to shearing, and thus the residence
time decreases with the breakthrough curve shifting backward.
They also observed that the breakthrough curve for interacting
tracers (i.e. with matrix diffusion) exhibits longer tails than that
for non-interacting tracers (i.e. without matrix diffusion) due to
the meandering of a small amount of particles passing through tor-
tuous fluid pathways. Such long tail phenomena were more signif-
icant when the pressure gradient is small, for which the matrix
diffusion tends to play a dominant role in solute transport
(Fig. 16b). Rutqvist et al. [101] used an extended multiple interact-
ing continua model combined with the crack tensor approach to
simulate the advection-dominated transport (under a high hydrau-
lic gradient) and diffusion-retarded transport (under a low hydrau-
lic gradient). In addition to the stress-dependent transport
168 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176

Fig. 15. Variation of permeability of fractured rocks in response to the change of stress conditions. (a) Permeability change versus stress change with a fixed principal stress
ratio of 1.3 [206]. (b) Permeability change with the increase of stress ratio for a DFN with a constant initial aperture [206]. (c) Permeability change with the increase of stress
ratio for a DFN with a lognormally distributed and length correlated apertures under rotated stress fields [98]. (d) Increase of vertical permeability with the increase of
horizontal stress ratio for a 3D persistent fracture network [278].

Fig. 16. Breakthrough curves for interacting tracers (i.e. with matrix diffusion) in a 2D DFN network stressed by various ratios of horizontal to vertical stresses (i.e. K) under a
horizontal hydraulic pressure gradient of (a) 1  104 Pa/m, and (b) 10 Pa/m [209]. (Note: c0 is the initial concentration along upstream boundary, c is the concentration
observed at the downstream boundary, K = 0 denotes a zero stress condition, and the two figures have different time scales).
 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176
behaviour, they also observed a delayed breakthrough in low pres-
sure gradient scenarios due to the residence of solutes in the por-
ous rock matrix. Wang et al. [301] further applied this
multicontinuum method to demonstrate the contribution of inac-
tive fractures (i.e. isolated cracks or dead-ends of fractures) for
stagnating solutes by providing additional surface areas for diffu-
sive transfer into/out of matrix pores. Zhao et al. [102] compared
the stress-flow models of five different research groups, which
showed consistency in predicting the stress dependency of mass
transport in a fractured rock. Recently, Kang et al. [302] studied
the impact of in-situ stresses and aperture changes on the anoma-
lous transport in a fractured rock based on an outcrop-based DFN
model. Apart from the stress-induced aperture change that can
affect the transport behaviour, Nick et al. [162] found that the
propagation of fractures under tectonic loading can also vary the
breakthrough properties due to the increased fracture density
and network connectivity.
4.4. Comments on the use of DFNs for HM modelling
The geomechanical effects on fluid flow in fractured rocks as
observed in numerical simulations demonstrate the importance
of using appropriate fracture network representations and con-
ducting systematic geomechanical computations for characterising
the coupled HM behaviour of fractured geological formations. Sim-
ulation results rely strongly on the suitability and accuracy of the
constructed DFNs and the geomechanical models of rocks and
fractures.
Natural fracture networks have been observed to exhibit fractal
characteristics and long-range correlations [8,31,32,73,124–127,1
30]. The fractal (or clustered) characteristics have been found to
have significant impacts on the connectivity, permeability and flow
pattern of fracture networks [131–133]. The scattered distribution
of fracture sizes (e.g. power law distribution) also considerably
affects the hydrological [92–94], geomechanical [134] and HM
[40] behaviour of fractured rocks. These important features may
need to be prudently incorporated when creating DFNs. The
hydraulically inactive fractures (i.e. dead-end and isolated frac-
tures) were found playing important roles in propagating new frac-
tures that can link with pre-existing fractures
[15,39,41,151,161,280], and in transferring fluids and solutes
into/out of matrix pores [101,171,301]. Thus, they may need to
be preserved in DFNs for HM modelling of fractured rocks. The bent
fractures that can accommodate large shear dilations and induce
localised fluid flows [15,47] may also need to be adequately char-
acterised in DFN representations. Such curvature characteristics
may provide additional surface areas for diffusive transfer and
are therefore important for solute transport modelling. Special
attention is needed when defining fracture apertures in DFNs for
HM modelling. Fracture apertures have been suggested to have
intrinsic correlations with fracture sizes [5,80], which tend to exhi-
bit a significant power law scaling over a broad spectrum of length
scales [8,73]. This long-range correlation has been found to have
critical influences on the permeability and flow distribution of frac-
ture networks [93,94,97,99] as well as their stress dependency
[40,98]. In addition, the geomechanical simulation tool also needs
to be carefully adopted or developed for HM modelling. The follow-
ing aspects may be of concern: (1) whether the elastic/elasto-
plastic deformation of rock matrix can be modelled; (2) whether
the non-linear deformation of natural fractures (opening/closing,
shearing and dilation) can be captured; (3) whether the fracture
propagation can be simulated; and (4) whether the geomechanical
model can be coupled with fluid flow solvers.
It can be noted that this section is focused on the research of
modelling geomechanical impacts on the hydrological behaviour
of fractured rocks. The reverse effect of fluid flow on rock mass
169
deformations is also a very important issue which is relevant to
many engineering problems such as hydraulic fracturing, CO2
sequestration and reservoir compaction. Recent developments in
exploitation of shale gas resources have highlighted the demand
for a better understanding of the interaction between hydraulic
fractures and natural fractures. Several studies have been done to
develop coupled HM models to simulate fluid-driven fracture
propagations in pre-existing DFN networks [185,186,303–308].
5. Outstanding issues and concluding remarks
Modelling the geomechanical evolution and the resulting
hydrological characteristics of fractured rocks is a challenging
issue. The previous HM simulation results have shown consistency
with field measurements: (i) only a small portion of fractures are
conductive [6,144], (ii) permeability is less sensitive to depth in
deep rocks [2], and (iii) critically stressed faults tend to have much
higher hydraulic conductivity [16,309]. However, a number of
important issues of difficulty still exist in applying DFNs for HM
modelling of fractured rocks. The first task will be the development
of realistic DFNs that can represent the important geometrical and
geostatistical characteristics of natural fracture systems (e.g. clus-
tering, scaling, connectivity, intersection, termination, curvature
and correlation), and that can further be used to capture the key
geomechanical and hydrological characteristics of fractured rocks
(e.g. strength, deformation, permeability and mass transport).
The geometry of the created DFNs needs to be consistent with
the geomechanics such as the in-situ stress field so that the
geometrically- and geomechanically-dependent fluid flow can be
properly modelled. Another critical issue is to develop appropriate
upscaling approaches to DFN models to evaluate the large-scale
behaviour, and this would require the preservation of geostatistical
and geomechanical characteristics. Some techniques have been
proposed to construct large-scale heterogeneous models based
on upscaled stress-dependent permeability tensors of local grid
blocks [310–312] or using self-referencing random walks associ-
ated with stress-dependent aperture information [40]. More efforts
are also needed in the future for many other aspects, such as devel-
oping more advanced ‘‘two-way” coupling schemes, modelling
geomechanical effects on multiphase flow and, importantly, exten-
sion to complex 3D systems.
The development of calibration and validation procedures for
DFN models based on field measurements is a crucial issue if the
numerical models are to be used for practical applications. The
DFN model can be examined by comparing the geometrical proper-
ties of a simulated 2D trace network with those of the trace pattern
exposed on an outcrop or excavation wall [117,138,141]. Mine-by
experiments in underground research laboratories [313,314] pro-
vide also valuable patterns of fracture traces that intersect tunnel
walls and measurement data of rock mass mechanical responses
(e.g. breakout characteristics, spalling depth, tunnel radial defor-
mation, etc.). Such data might be very useful for the DFN calibra-
tion study. Hydraulic test data such as groundwater inflow
recorded at boreholes/shafts, pressure variations and derivatives
obtained from well/interference tests, and particle concentration
monitored during tracer tests can also be used to tune the hydro-
logical performance of DFN models [138–140,142–144]. Further-
more, the oil recovery case history information in reservoir
engineering [315] offers useful references for the calibration of
DFN models used for multiphase flow simulation. Recently, an
inverse model constrained by in-situ hydraulic data from sequen-
tial pumping tests has been developed to derive the spatial distri-
bution of flow channels on the bedding plane of a karstic formation
[316]. This new technique can be employed to calculate best-fit
aperture distributions for DFN models based on an optimisation
170 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176
process using multiple objective functions. Other possible calibra-
tion approaches may include those based on heat transfer experi-
ments and electrical resistivity measurements of fractured rocks.
In summary, the present paper began by presenting an over-
view of various discrete fracture network (DFN) models for simu-
lating the geometry and topology of natural discontinuity
systems. Different continuum and discontinuum approaches that
integrate DFN geometries to model the geomechanical behaviour
of fractured rocks were then surveyed. Numerical results of the
fracture-dependent mechanical response and stress-dependent
hydrological characteristics of fractured rocks suggest that it is
important to use appropriate DFN representations and conduct
geomechanical computations to better characterise the bulk beha-
viour (e.g. strength, deformation, permeability and mass transport)
of highly disordered geological media embedded with naturally
occurring discontinuities. Some of the outstanding issues that need
to be addressed in the near future will be those related to genera-
tion of more realistic DFNs, improvement of computational effi-
ciency, study of coupled multiphysics processes, calibration and
validation of numerical models as well as upscaling for large-
scale practical applications.
Acknowledgements
We thank the industry consortium for supporting part of the
lead author’s PhD research on the project of ‘‘Improved simulation
of faulted and fractured reservoirs, itf-ISF Phase 3: gravity assisted
recovery processes”. The lead author also acknowledges the Janet
Watson scholarship awarded by the Department of Earth Science
and Engineering, Imperial College London. The second author is
grateful for the NERC RATE HYDROFRAME project of ‘‘Hydrome-
chanical and biogeochemical processes in fractured rock masses
in the vicinity of a geological disposal facility for radioactive waste”
and the EU H2020-SURE project of ‘‘Novel productivity enhance-
ment concept for a sustainable utilization of a geothermal
resource”. The third author gratefully acknowledges the support
of the Swedish Geological Survey (SGU) through its grant, number
1724, to Uppsala University.
References
[1] Zimmerman RW, Main I. Hydromechanical behavior of fractured rocks. In:
Gueguen Y, Bouteca M, editors. Mechanics of fluid-saturated
rocks. London: Elsevier; 2004. p. 363–421.
[2] Rutqvist J, Stephansson O. The role of hydromechanical coupling in fractured
rock engineering. Hydrogeol J 2003;11:7–40. http://dx.doi.org/10.1007/
s10040-002-0241-5.
[3] Tsang C-F, Bernier F, Davies C. Geohydromechanical processes in the
Excavation Damaged Zone in crystalline rock, rock salt, and indurated and
plastic clays—in the context of radioactive waste disposal. Int J Rock Mech
Min Sci 2005;42:109–25. http://dx.doi.org/10.1016/j.ijrmms.2004.08.003.
[4] Tsang C-F, Neretnieks I, Tsang Y. Hydrologic issues associated with nuclear
waste repositories. Water Resour Res 2015;51:6923–72. http://dx.doi.org/
10.1002/2015WR017641.
[5] Pollard DD, Segall P. Theoretical displacements and stresses near fractures in
rock: with applications to faults, joints, veins, dikes, and solution surfaces. In:
Atkinson BK, editor. Fracture mechanism of. rock. San Diego: Academic Press;
1987. p. 277–349.
[6] Tsang C-F, Neretnieks I. Flow channeling in heterogeneous fractured rocks.
Rev Geophys 1998;36:275–98. http://dx.doi.org/10.1029/97RG03319.
[7] Herbert AW. Modelling approaches for discrete fracture network flow
analysis. In: Stephansson O, Jing L, Tsang C-F, editors. Coupled thermo-
hydro-mechanical processes of fractured media,
79. Amsterdam: Elsevier; 1996. p. 213–29.
[8] Lei Q, Wang X. Tectonic interpretation of the connectivity of a multiscale
fracture system in limestone. Geophys Res Lett 2016;43:1551–8. http://dx.
doi.org/10.1002/2015GL067277.
[9] Pollard DD, Aydin A. Progress in understanding jointing over the past century.
Geol Soc Am Bull 1988;100:1181–204. http://dx.doi.org/10.1130/0016-7606
(1988) 100<1181:PIUJOT>2.3.CO;2.
[10] Schultz R. Growth of geologic fractures into large-strain populations: review
of nomenclature, subcritical crack growth, and some implications for rock
engineering. Int J Rock Mech Min Sci 2000;37:403–11. http://dx.doi.org/
10.1016/S1365-1609(99)00115-X.
[11] Einstein HH, Baecher GB. Probabilistic and statistical methods in engineering
geology. Rock Mech Rock Eng 1983;16:39–72. http://dx.doi.org/10.1007/
BF01030217.
[12] Jing L. A review of techniques, advances and outstanding issues in numerical
modelling for rock mechanics and rock engineering. Int J Rock Mech Min Sci
2003;40:283–353. http://dx.doi.org/10.1016/S1365-1609(03)00013-3.
[13] Hoek E, Martin CD. Fracture initiation and propagation in intact rock – a
review. J Rock Mech Geotech Eng 2014;6:287–300. http://dx.doi.org/10.1016/
j.jrmge.2014.06.001.
[14] Barton N, Bandis S, Bakhtar K. Strength, deformation and conductivity
coupling of rock joints. Int J Rock Mech Min Sci Geomech Abstr
1985;22:121–40. http://dx.doi.org/10.1016/0148-9062(85)93227-9.
[15] Latham J-P, Xiang J, Belayneh M, Nick HM, Tsang C-F, Blunt MJ. Modelling
stress-dependent permeability in fractured rock including effects of
propagating and bending fractures. Int J Rock Mech Min Sci
2013;57:100–12. http://dx.doi.org/10.1016/j.ijrmms.2012.08.002.
[16] Zoback MD. Reservoir geomechanics. Cambridge: Cambridge University
Press; 2007.
[17] Bear J, Tsang C-F, de Marsily G. Flow and contaminant transport in fractured
rock. San Diego: Academic Press Inc.; 1993.
[18] Dershowitz WS, Einstein HH. Characterizing rock joint geometry with joint
system models. Rock Mech Rock Eng 1988;21:21–51. http://dx.doi.org/
10.1007/BF01019674.
[19] Liu R, Li B, Jiang Y, Huang N. Review: mathematical expressions for estimating
equivalent permeability of rock fracture networks. Hydrogeol J 2016. http://
dx.doi.org/10.1007/s10040-016-1441-8.
[20] Adler PM, Thovert J-F. Fractures and fracture networks. Dordrecht: Kluwer
Academic Publishers; 1999.
[21] Adler PM, Thovert J-F, Mourzenko VV. Fractured porous
media. Oxford: Oxford University Press; 2012.
[22] Jing L, Hudson JA. Numerical methods in rock mechanics. Int J Rock Mech Min
Sci 2002;39:409–27. http://dx.doi.org/10.1016/S1365-1609(02)00065-5.
[23] Yuan SC, Harrison JP. A review of the state of the art in modelling progressive
mechanical breakdown and associated fluid flow in intact heterogeneous
rocks. Int J Rock Mech Min Sci 2006;43:1001–22. http://dx.doi.org/10.1016/j.
ijrmms.2006.03.004.
[24] Bobet A, Fakhimi A, Johnson S, Morris J, Tonon F, Ronald Yeung M. Numerical
models in discontinuous media: review of advances for rock mechanics
applications. J Geotech Geoenviron Eng 2009;135:1547–61. http://dx.doi.org/
10.1061/(ASCE)GT.1943-5606.0000133.
[25] Lisjak A, Grasselli G. A review of discrete modeling techniques for fracturing
processes in discontinuous rock masses. J Rock Mech Geotech Eng
2014;6:301–14. http://dx.doi.org/10.1016/j.jrmge.2013.12.007.
[26] Zhang X, Sanderson DJ. Numerical modelling and analysis of fluid flow and
deformation of fractured rock masses. Oxford: Pergamon; 2002.
[27] Segall P, Pollard DD. Joint formation in granitic rock of the Sierra Nevada. Geol
Soc Am Bull 1983;94:563. http://dx.doi.org/10.1130/0016-7606(1983)
94<563:JFIGRO>2.0.CO;2.
[28] Engelder T, Geiser P. On the use of regional joint sets as trajectories of
paleostress fields during the development of the Appalachian Plateau. New
York. J Geophys Res 1980;85:6319. http://dx.doi.org/10.1029/
JB085iB11p06319.
[29] Olson J, Pollard D. Inferring paloestresses from natural fracture patterns: a
new method. Geology 1989;17:345–8. http://dx.doi.org/10.1130/0091-7613
(1989) 017<0345.
[30] Petit J-P, Mattauer M. Palaeostress superimposition deduced from mesoscale
structures in limestone: the Matelles exposure, Languedoc, France. J Struct
Geol 1995;17:245–56. http://dx.doi.org/10.1016/0191-8141(94)E0039-2.
[31] La Pointe PR. A method to characterize fracture density and connectivity
through fractal geometry. Int J Rock Mech Min Sci Geomech Abstr
1988;25:421–9. http://dx.doi.org/10.1016/0148-9062(88)90982-5.
[32] Bour O, Davy P, Darcel C, Odling N. A statistical scaling model for fracture
network geometry, with validation on a multiscale mapping of a joint
network (Hornelen Basin, Norway). J Geophys Res 2002;107. http://dx.doi.
org/10.1029/2001JB00017. ETG4-1–ETG4-12.
[33] Belayneh M, Cosgrove JW. Fracture-pattern variations around a major fold
and their implications regarding fracture prediction using limited data: an
example from the Bristol Channel Basin. In: Cosgrove JW, Engelder T, editors.
Initiation, propagation, and arrest of joints and other fractures. London: Geological
Society, London, Special Publications; 2004. p. 89–102.
[34] Masihi M, King PR. Connectivity prediction in fractured reservoirs with
variable fracture size: analysis and validation. SPE J 2008;13:12–5. http://dx.
doi.org/10.2118/100229-PA.
[35] Belayneh M, Geiger S, Matthäi SK. Numerical simulation of water injection
into layered fractured carbonate reservoir analogs. Am Assoc Pet Geol Bull
vol.
2006;90:1473–93. http://dx.doi.org/10.1306/05090605153.
[36] Belayneh MW, Matthäi SK, Blunt MJ, Rogers SF. Comparison of deterministic
with stochastic fracture models in water-flooding numerical simulations. Am
Assoc Pet Geol Bull 2009;93:1633–48. http://dx.doi.org/10.1306/
07220909031.
[37] Geiger S, Matthäi SK. What can we learn from high-resolution numerical
simulations of single- and multi-phase fluid flow in fractured outcrop
analogues? In: Spence GH et al., editors. Advances in the study of fractured
reservoirs, vol. 374. Geological Society, London, Special Publications; 2012.
 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176
[38] Geiger S, Cortis A, Birkholzer JT. Upscaling solute transport in naturally
fractured porous media with the continuous time random walk method.
Water Resour Res 2010;46:W12530. http://dx.doi.org/10.1029/
2010WR009133.
[39] Lei Q, Latham J-P, Xiang J, Tsang C-F, Lang P, Guo L. 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
2014;70:507–23. http://dx.doi.org/10.1016/j.ijrmms.2014.06.001.
[40] Lei Q, Latham J-P, Tsang C-F, Xiang J, Lang P. A new approach to upscaling
fracture network models while preserving geostatistical and geomechanical
characteristics. J Geophys Res Solid Earth 2015;120:4784–807. http://dx.doi.
org/10.1002/2014JB011736.
[41] Figueiredo B, Tsang C-F, Rutqvist J, Niemi A. A study of changes in deep
fractured rock permeability due to coupled hydro-mechanical effects. Int J
Rock Mech Min Sci 2015;79:70–85. http://dx.doi.org/10.1016/j.
ijrmms.2015.08.011.
[42] Zhang X, Sanderson DJ. Anisotropic features of geometry and permeability in
fractured rock masses. Eng Geol 1995;40:65–75. http://dx.doi.org/10.1016/
0013-7952(95)00040-2.
[43] Zhang X, Sanderson DJ. Effects of stress on the two-dimensional permeability
tensor of natural fracture networks. Geophys J Int 1996;125:912–24. http://
dx.doi.org/10.1111/j.1365-246X.1996.tb06034.x.
[44] Zhang X, Sanderson DJ. Numerical study of critical behaviour of deformation
and permeability of fractured rock masses. Mar Pet Geol 1998;15:535–48.
http://dx.doi.org/10.1016/S0264-8172(98)00030-0.
[45] Zhang X, Sanderson DJ, Harkness RM, Last NC. Evaluation of the 2-D
permeability tensor for fractured rock masses. Int J Rock Mech Min Sci
Geomech Abstr 1996;33:17–37. http://dx.doi.org/10.1016/0148-9062(95)
00042-9.
[46] Brown SR, Bruhn RL. Fluid permeability of deformable fracture networks. J
Geophys Res 1998;103:2489–500. http://dx.doi.org/10.1029/97JB03113.
[47] Sanderson DJ, Zhang X. Critical stress localization of flow associated with
deformation of well-fractured rock masses, with implications for mineral
deposits. In: Mccavvrey KJW et al., editors. Fractures, fluid flow and
mineralization, vol. 155. London: Geological Society, London, Special
Publications; 1999. p. 69–81. http://dx.doi.org/10.1144/GSL.
SP.1999.155.01.07.
[48] Griffith WA, Sanz PF, Pollard DD. Influence of outcrop scale fractures on the
effective stiffness of fault damage zone rocks. Pure Appl Geophys
2009;166:1595–627. http://dx.doi.org/10.1007/s00024-009-0519-9.
[49] Sturzenegger M, Stead D. Close-range terrestrial digital photogrammetry and
terrestrial laser scanning for discontinuity characterization on rock cuts. Eng
Geol 2009;106:163–82. http://dx.doi.org/10.1016/j.enggeo.2009.03.004.
[50] Sturzenegger M, Stead D, Elmo D. Terrestrial remote sensing-based
estimation of mean trace length, trace intensity and block size/shape. Eng
Geol 2011;119:96–111. http://dx.doi.org/10.1016/j.enggeo.2011.02.005.
[51] Jacquemyn C, Swennen R, Ronchi P. Mechanical stratigraphy and (palaeo-)
karstification of the Murge area (Apulia, southern Italy). In: Garland J, Neilson
JE, Laubach SE, Whidden KJ, editors. Advances in carbonate exploration and
reservoir analysis, vol. 370. London: Geological Society, London, Special
Publications; 2012. p. 169–86. http://dx.doi.org/10.1144/SP370.4.
[52] Jones RR, Pearce MA, Jacquemyn C, Watson FE. Robust best-fit planes from
geospatial data. Geosphere 2016;12:196–202. http://dx.doi.org/10.1130/
GES01247.1.
[53] Jourde H, Flodin EA, Aydin A, Durlofsky LJ, Wen X-H. Computing permeability
of fault zones in eolian sandstone from outcrop measurements. Am Assoc Pet
Geol Bull 2002;86:1187–200. http://dx.doi.org/10.1306/61EEDC4C-173E-
11D7-8645000102C1865D.
[54] Leckenby RJ, Sanderson DJ, Lonergan L. Estimating flow heterogeneity in
natural fracture systems. J Volcanol Geotherm Res 2005;148:116–29. http://
dx.doi.org/10.1016/j.jvolgeores.2005.03.017.
[55] Taylor WL, Pollard DD, Aydin A. Fluid flow in discrete joint sets: field
observations and numerical simulations. J Geophys Res
1999;104:28983–9006. http://dx.doi.org/10.1029/1999JB900179.
[56] Wilson CE, Aydin A, Karimi-Fard M, Durlofsky LJ, Sagy A, Brodsky EE, et al.
From outcrop to flow simulation: constructing discrete fracture models from
a LIDAR survey. Am Assoc Pet Geol Bull 2011;95:1883–906. http://dx.doi.org/
10.1306/03241108148.
[57] Jacquemyn C, Huysmans M, Hunt D, Casini G, Swennen R. Multi-scale three-
dimensional distribution of fracture- and igneous intrusion- controlled
hydrothermal dolomite from digital outcrop model, Latemar platform,
Dolomites, northern Italy. Am Assoc Pet Geol Bull 2015;99:957–84. http://
dx.doi.org/10.1306/10231414089.
[58] Wu H, Pollard D. Imaging 3-D fracture networks around boreholes. Am Assoc
Pet Geol Bull 2002;4:593–604. http://dx.doi.org/10.1306/61EEDB52-173E-
11D7-8645000102C1865D.
[59] Kattenhorn SA, Pollard DD. Integrating 3-D seismic data, field analogs, and
mechanical models in the analysis of segmented normal faults in the Wytch
Farm oil field, southern England, United Kingdom. Am Assoc Pet Geol Bull
2001;85:1183–210. http://dx.doi.org/10.1306/8626CA91-173B-11D7-
8645000102C1865D.
[60] Balberg I, Binenbaum N. Computer study of the percolation threshold in a
two-dimensional anisotropic system of conducting sticks. Phys Rev B
1983;28:3799–812. http://dx.doi.org/10.1103/PhysRevB.28.3799.
171
[61] Robinson PC. Connectivity of fracture systems-a percolation theory approach.
J Phys A Math Gen 1983;16:605–14. http://dx.doi.org/10.1088/0305-4470/
16/3/020.
[62] Robinson PC. Numerical calculations of critical densities for lines and planes. J
Phys A Math Gen 1984;17:2823. http://dx.doi.org/10.1088/0305-4470/17/14/
025.
[63] Long JCS, Remer JS, Wilson CR, Witherspoon PA. Porous media equivalents for
networks of discontinuous fractures. Water Resour Res 1982;18:645. http://
dx.doi.org/10.1029/WR018i003p00645.
[64] Long JCS, Gilmour P, Witherspoon PA. A model for steady fluid flow in random
three-dimensional networks of disc-shaped fractures. Water Resour Res
1985;21:1105–15. http://dx.doi.org/10.1029/WR021i008p01105.
[65] Andersson J, Shapiro AM, Bear J. A stochastic model of a fractured rock
conditioned by measured information. Water Resour Res 1984;20:79. http://
dx.doi.org/10.1029/WR020i001p00079.
[66] Andersson J, Dverstorp B. Conditional simulations of fluid flow in three-
dimensional networks of discrete fractures. Water Resour Res 1987;23:1876.
http://dx.doi.org/10.1029/WR023i010p01876.
[67] Long JCS, Billaux DM. From field data to fracture network modeling: an
example incorporating spatial structure. Water Resour Res 1987;23:1201.
http://dx.doi.org/10.1029/WR023i007p01201.
[68] Baecher GB. Statistical analysis of rock mass fracturing. Math Geol
1983;15:329–48. http://dx.doi.org/10.1007/BF01036074.
[69] Priest SD, Hudson Ja. Estimation of discontinuity spacing and trace length
using scanline surveys. Int J Rock Mech Min Sci Geomech Abstr
1981;18:183–97. http://dx.doi.org/16/0148-9062(81)90973-6.
[70] Kulatilake PHSW, Wu TH. Estimation of mean trace length of discontinuities.
Rock Mech Rock Eng 1984;17:215–32. http://dx.doi.org/10.1007/
BF01032335.
[71] Zhang L, Einstein HH. Estimating the intensity of rock discontinuities. Int J
Rock Mech Min Sci 2000;37:819–37. http://dx.doi.org/10.1016/S1365-1609
(00)00022-8.
[72] Davy P. On the frequency-length distribution of the San Andreas Fault
System. J Geophys Res 1993;98:12141–51. http://dx.doi.org/10.1029/
93JB00372.
[73] Bonnet E, Bour O, Odling NE, Davy P, Main I, Cowie P, et al. Scaling of fracture
systems in geological media. Rev Geophys 2001;39:347–83. http://dx.doi.
org/10.1029/1999RG000074.
[74] Dershowitz WS, Herda HH. Interpretation of fracture spacing and intensity.
In: 33th U.S. symposium on rock mechanism, Santa Fe. p. 757–66.
[75] Hudson JA, Priest SD. Discontinuity frequency in rock masses. Int J Rock Mech
Min Sci 1983;20:73–89. http://dx.doi.org/10.1016/0148-9062(83)90329-7.
[76] Rives T, Razack M, Petit J-P, Rawnsley KD. Joint spacing: analogue and
numerical simulations. J Struct Geol 1992;14:925–37. http://dx.doi.org/
10.1016/0191-8141(92)90024-Q.
[77] Snow D. The frequency and apertures of fractures in rock. Int J Rock Mech Min
Sci Geomech Abstr 1970;7:23–40. http://dx.doi.org/10.1016/0148-9062(70)
90025-2.
[78] Barton CA, Zoback MD. Self-similar distribution and properties of
macroscopic fractures at depth in crystalline rock in the Cajon Pass
Scientific Drill Hole. J Geophys Res 1992;97:5181. http://dx.doi.org/
10.1029/91JB01674.
[79] Hooker JN, Gale JFW, Gomez LA, Laubach SE, Marrett R, Reed RM. Aperture-
size scaling variations in a low-strain opening-mode fracture set, Cozzette
Sandstone, Colorado. J Struct Geol 2009;31:707–18. http://dx.doi.org/
10.1016/j.jsg.2009.04.001.
[80] Olson JE. Sublinear scaling of fracture aperture versus length: an exception or
the rule? J Geophys Res 2003;108:2413. http://dx.doi.org/10.1029/
2001JB000419.
[81] Laslett GM. Censoring and edge effects in areal and line transect sampling of
rock joint traces. J Int Assoc Math Geol 1982;14:125–40. http://dx.doi.org/
10.1007/BF01083948.
[82] Pickering G, Bull JM, Sanderson DJ. Sampling power-law distributions.
Tectonophysics 1995;248:1–20. http://dx.doi.org/10.1016/0040-1951(95)
00030-Q.
[83] Mauldon M. Estimating mean fracture trace length and density from
observations in convex windows. Rock Mech Rock Eng 1998;31:201–16.
http://dx.doi.org/10.1007/s006030050021.
[84] Warburton PM. Stereological interpretation of joint trace data: influence of
joint shape and implications for geological surveys. Int J Rock Mech Min Sci
Geomech Abstr 1980;17:305–16. http://dx.doi.org/10.1016/0148-9062(80)
90513-6.
[85] Berkowitz B, Adler PM. Stereological analysis of fracture network structure in
geological formations. J Geophys Res 1998;103:15339–60.
[86] Golder Associate Inc. FracMan7 User Documentation. Seattle (USA); 2011.
[87] Hestir K, Long JCS. Analytical expressions for the permeability of random
two-dimensional Poisson fracture networks based on regular lattice
percolation and equivalent media theories. J Geophys Res 1990;95:21565.
http://dx.doi.org/10.1029/JB095iB13p21565.
[88] Zimmerman RW, Bodvarsson GS. Effective transmissivity of two-dimensional
fracture networks. Int J Rock Mech Min Sci Geomech Abstr 1996;33:433–8.
http://dx.doi.org/10.1016/0148-9062(95)00067-4.
[89] Bour O, Davy P. Connectivity of random fault networks following a power law
fault length distribution. Water Resour Res 1997;33:1567–83. http://dx.doi.
org/10.1029/96WR00433.
172 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176
[90] Bour O, Davy P. On the connectivity of three-dimensional fault networks.
Water Resour Res 1998;34:2611–22. http://dx.doi.org/10.1029/98WR01861.
[91] Renshaw CE. Connectivity of joint networks with power law length
distributions. Water Resour Res 1999;35:2661–70. http://dx.doi.org/
10.1029/1999WR900170.
[92] de Dreuzy J-R, Davy P, Bour O. Hydraulic properties of two-dimensional
random fracture networks following a power law length distribution: 1.
Effective connectivity. Water Resour Res 2001;37:2065–78. http://dx.doi.org/
10.1029/2001WR900011.
[93] de Dreuzy J-R, Davy P, Bour O. Hydraulic properties of two-dimensional
random fracture networks following a power law length distribution: 2.
Permeability of networks based on lognormal distribution of apertures.
Water Resour Res 2001;37:2079–95.
[94] de Dreuzy J-R, Davy P, Bour O. Hydraulic properties of two-dimensional
random fracture networks following power law distributions of length and
aperture. Water Resour Res 2002;38:1276. http://dx.doi.org/10.1029/
2001WR001009.
[95] Min K-B, Jing L, Stephansson O. Determining the equivalent permeability
tensor for fractured rock masses using a stochastic REV approach: METHOD
and application to the field data from Sellafield, UK. Hydrogeol J
2004;12:497–510. http://dx.doi.org/10.1007/s10040-004-0331-7.
[96] Sanderson DJ, Zhang X. Stress controlled localisation of deformation and fluid
flow in fractured rocks. In: Cosorove JW, Engelder T, editors. Initiation,
propagation, and arrest of joints and other fractures. London: Geological
Society, London, Special Publications; 2004. p. 299–314. http://dx.doi.org/
10.1144/GSL.SP.2004.231.01.18.
[97] Baghbanan A, Jing L. Hydraulic properties of fractured rock masses with
correlated fracture length and aperture. Int J Rock Mech Min Sci
2007;44:704–19. http://dx.doi.org/10.1016/j.ijrmms.2006.11.001.
[98] Baghbanan A, Jing L. Stress effects on permeability in a fractured rock mass
with correlated fracture length and aperture. Int J Rock Mech Min Sci
2008;45:1320–34. http://dx.doi.org/10.1016/j.ijrmms.2008.01.015.
[99] Klimczak C, Schultz R, Parashar R, Reeves D. Cubic law with aperture-length
correlation: implications for network scale fluid flow. Hydrogeol J
2010;18:851–62. http://dx.doi.org/10.1007/s10040-009-0572-6.
[100] Leung CTO, Zimmerman RW. Estimating the hydraulic conductivity of two-
dimensional fracture networks using network geometric properties. Transp
Porous Media 2012;93:777–97. http://dx.doi.org/10.1007/s11242-012-9982-3.
[101] Rutqvist J, Leung C, Hoch A, Wang Y, Wang Z. Linked multicontinuum and
crack tensor approach for modeling of coupled geomechanics, fluid flow and
transport in fractured rock. J Rock Mech Geotech Eng 2013;5:18–31. http://
dx.doi.org/10.1016/j.jrmge.2012.08.001.
[102] Zhao Z, Rutqvist J, Leung C, Hokr M, Liu Q, Neretnieks I, et al. Impact of stress
on solute transport in a fracture network: a comparison study. J Rock Mech
Geotech Eng 2013;5:110–23. http://dx.doi.org/10.1016/j.jrmge.2013.01.002.
[103] Zhang Z, Lei Q. Object-oriented modeling for three-dimensional multi-block
systems. Comput Geotech 2013;48:208–27. http://dx.doi.org/10.1016/
j.compgeo.2012.07.008.
[104] Zhang Z, Lei Q. A morphological visualization method for removability
analysis of blocks in discontinuous rock masses. Rock Mech Rock Eng
2014;47:1237–54. http://dx.doi.org/10.1007/s00603-013-0471-y.
[105] Lang PS, Paluszny A, Zimmerman RW. Permeability tensor of three-
dimensional fractured porous rock and a comparison to trace map
predictions. J Geophys Res Solid Earth 2014;119:6288–307. http://dx.doi.
org/10.1002/2014JB011027.
[106] Elmo D, Rogers S, Stead D, Eberhardt E. Discrete Fracture Network approach
to characterise rock mass fragmentation and implications for geomechanical
upscaling. Min Technol 2014;123:149–61. http://dx.doi.org/10.1179/
1743286314Y.0000000064.
[107] Ebigbo A, Lang PS, Paluszny A, Zimmerman RW. Inclusion-based effective
medium models for the permeability of a 3D fractured rock mass. Transp
Porous Media 2016;113:137–58. http://dx.doi.org/10.1007/s11242-016-
0685-z.
[108] Odling NE, Webman I. A ‘‘conductance” mesh approach to the permeability of
natural and simulated fracture patterns. Water Resour Res 1991;27:2633–43.
http://dx.doi.org/10.1029/91WR01382.
[109] Odling NE. Network properties of a two-dimensional natural fracture pattern.
Pure Appl Geophys 1992;138:95–114. http://dx.doi.org/10.1007/
BF00876716.
[110] Berkowitz B, Hadad A. Fractal and multifractal measures of natural and
synthetic fracture networks. J Geophys Res 1997;102:12205. http://dx.doi.
org/10.1029/97JB00304.
[111] Brzovic A, Rogers S, Webb G, Hurtado JP, Marin N, Schachter P, et al. Discrete
fracture network modelling to quantify rock mass pre-conditioning at the El
Teniente Mine, Chile. Min Technol 2015;124:163–77. http://dx.doi.org/
10.1179/1743286315Y.0000000019.
[112] Billaux D, Chiles JP, Hestir K, Long J. Three-dimensional statistical modelling
of a fractured rock mass—an example from the Fanay-Augères Mine. Int J
Rock Mech Min Sci Geomech Abstr 1989;26:281–99. http://dx.doi.org/
10.1016/0148-9062(89)91977-3.
[113] Xu C, Dowd P. A new computer code for discrete fracture network modelling.
Comput Geosci 2010;36:292–301. http://dx.doi.org/10.1016/j.
cageo.2009.05.012.
[114] Fadakar-Alghalandis Y. Stochastic modeling of fractures in rock
masses. Adelaide: University of Adelaide; 2014.
[115] Masihi M, King PR. A correlated fracture network: modeling and percolation
properties. Water Resour Res 2007;43:1–9. http://dx.doi.org/10.1029/
2006WR005331.
[116] Shekhar R, Gibson RL. Generation of spatially correlated fracture models for
seismic simulations. Geophys J Int 2011;185:341–51. http://dx.doi.org/
10.1111/j.1365-246X.2011.04937.x.
[117] Meyer T, Einstein HH. Geologic stochastic modeling and connectivity
assessment of fracture systems in the Boston area. Rock Mech Rock Eng
2002;35:23–44. http://dx.doi.org/10.1007/s006030200007.
[118] Manzocchi T. The connectivity of two-dimensional networks of spatially
correlated fractures. Water Resour Res 2002;38:1162. http://dx.doi.org/
10.1029/2000WR000180.
[119] Josnin J, Jourde H, Fénart P, Bidaux P. A three-dimensional model to simulate
joint networks in layered rocks. Can J Earth Sci 2002;39:1443–55. http://dx.
doi.org/10.1139/e02-043.
[120] Jourde H, Fenart P, Vinches M, Pistre S, Vayssade B. Relationship between the
geometrical and structural properties of layered fractured rocks and their
effective permeability tensor. A simulation study. J Hydrol 2007;337:117–32.
http://dx.doi.org/10.1016/j.jhydrol.2007.01.027.
[121] Lu P, Latham J-P. Developments in the assessment of in-situ block size
distributions of rock masses. Rock Mech Rock Eng 1999;32:29–49. http://dx.
doi.org/10.1007/s006030050042.
[122] Staub I, Fredriksson A, Outters N. Strategy for a rock mechanics site
descriptive model: development and testing of the theoretical approach
[SKB Rep. R-02-02]. Golder Associates AB; 2002.
[123] Darcel C, Bour O, Davy P. Stereological analysis of fractal fracture networks. J
Geophys Res Solid Earth 2003;108:2451. http://dx.doi.org/10.1029/
2002JB002091.
[124] Allègre CJ, Le Mouel JL, Provost A. Scaling rules in rock fracture and possible
implications for earthquake prediction. Nature 1982;297:47–9. http://dx.doi.
org/10.1038/297047a0.
[125] Barton CC. Fractal analysis of scaling and spatial clustering of fractures. In:
Barton CC, La Pointe PR, editors. Fractals in the earth sciences. New
York: Plenum Press; 1995. p. 141–78.
[126] Chilès JP. Fractal and geostatistical methods for modeling of a fracture
network. Math Geol 1988;20:631–54. http://dx.doi.org/10.1007/
BF00890581.
[127] Ehlen J. Fractal analysis of joint patterns in granite. Int J Rock Mech Min Sci
2000;37:909–22. http://dx.doi.org/10.1016/S1365-1609(00)00027-7.
[128] Hentschel HGE, Procaccia I. The infinite number of generalized dimensions of
fractals and strange attractors. Phys D 1983;8:435–44. http://dx.doi.org/
10.1016/0167-2789(83)90235-X.
[129] Bour O, Davy P. Clustering and size distributions of fault patterns: theory and
measurements. Geophys Res Lett 1999;26:2001–4. http://dx.doi.org/
10.1029/1999GL900419.
[130] Davy P, Le Goc R, Darcel C, Bour O, de Dreuzy J-R, Munier R. A likely universal
model of fracture scaling and its consequence for crustal hydromechanics. J
Geophys Res 2010;115:B10411. http://dx.doi.org/10.1029/2009JB007043.
[131] Darcel C, Bour O, Davy P, de Dreuzy J-R. Connectivity properties of two-
dimensional fracture networks with stochastic fractal correlation. Water
Resour Res 2003;39:1272. http://dx.doi.org/10.1029/2002WR001628.
[132] de Dreuzy J-R, Darcel C, Davy P, Bour O. Influence of spatial correlation of
fracture centers on the permeability of two-dimensional fracture networks
following a power law length distribution. Water Resour Res 2004;40. http://
dx.doi.org/10.1029/2003WR00226.
[133] Davy P, Bour O, de Dreuzy J-R, Darcel C. Flow in multiscale fractal fracture
networks. In: Cello G, Malamud BD, editors. Fractal analysis for natural
hazards, vol. 261. London: Geological Society, London, Special Publications;
2006. p. 31–45. http://dx.doi.org/10.1144/GSL.SP.2006.261.01.03.
[134] Harthong B, Scholtès L, Donzé FV. Strength characterization of rock masses,
using a coupled DEM-DFN model. Geophys J Int 2012;191:467–80. http://dx.
doi.org/10.1111/j.1365-246X.2012.05642.x.
[135] Davy P, Le Goc R, Darcel C. A model of fracture nucleation, growth and arrest,
and consequences for fracture density and scaling. J Geophys Res Solid Earth
2013;118:1393–407. http://dx.doi.org/10.1002/jgrb.50120.
[136] Horgan GW, Young IM. An empirical stochastic model for the geometry of
two-dimensional crack growth in soil (with Discussion). Geoderma
2000;96:263–76. http://dx.doi.org/10.1016/S0016-7061(00)00015-X.
[137] Ronayne MJ, Gorelick SM. Effective permeability of porous media containing
branching channel networks. Phys Rev E 2006;73:1–10. http://dx.doi.org/
10.1103/PhysRevE.73.026305.
[138] Dverstorp B, Andersson J. Application of the discrete fracture network
concept with field data: possibilities of model calibration and validation.
Water Resour Res 1989;25:540–50. http://dx.doi.org/10.1029/
WR025i003p00540.
[139] Cacas MC, Ledoux E, de Marsily G, Tillie B, Barbreau A, Durand E, et al.
Modeling fracture flow with a stochastic discrete fracture network:
calibration and validation: 1. The flow model. Water Resour Res
1990;26:479. http://dx.doi.org/10.1029/WR026i003p00479.
[140] Cacas MC, Ledoux E, de Marsily G, Barbreau A, Calmels P, Gaillard B, et al.
Modeling fracture flow with a stochastic discrete fracture network:
calibration and validation: 2. The transport model. Water Resour Res
1990;26:491–500. http://dx.doi.org/10.1029/WR026i003p00491.
[141] Kulatilake PHSW, Wathugala DN, Stephansson O. Joint network modelling
with a validation exercise in Stripa Mine, Sweden. Int J Rock Mech Min Sci
 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176
Geomech Abstr 1993;30:503–26. http://dx.doi.org/10.1016/0148-9062(93)
92217-E.
[142] Sarda S, Jeannin L, Basquet R, Bourbiaux B. Hydraulic characterization of
fractured reservoirs: simulation on discrete fracture models. SPE Reserv Eval
Eng 2002;5:154–62. http://dx.doi.org/10.2118/77300-PA.
[143] Casciano C, Ruvo L, Volpi B, Masserano F. Well test simulation through
Discrete Fracture Network modelling in a fractured carbonate reservoir. Pet
Geosci 2004;10:331–42. http://dx.doi.org/10.1144/1354-079303-590.
[144] Follin S, Hartley L, Rhén I, Jackson P, Joyce S, Roberts D, et al. A methodology
to constrain the parameters of a hydrogeological discrete fracture network
model for sparsely fractured crystalline rock, exemplified by data from the
proposed high-level nuclear waste repository site at Forsmark, Sweden.
Hydrogeol J 2014;22:313–31. http://dx.doi.org/10.1007/s10040-013-1080-2.
[145] Elmo D, Stead D, Rogers S. Guidelines for the quantitative description of
discontinuities for use in discrete fracture network engineering. In:
Proceedings of the 13th ISRM congress, Montreal. p. 587.
[146] Sanderson DJ, Nixon CW. The use of topology in fracture network
characterization. J Struct Geol 2015;72:55–66. http://dx.doi.org/10.1016/j.
jsg.2015.01.005.
[147] Darcel C, Bour O, Davy P. Cross-correlation between length and position in
real fracture networks. Geophys Res Lett 2003;30:1–4. http://dx.doi.org/
10.1029/2003GL017174.
[148] Neuman SP. Multiscale relationships between fracture length, aperture,
density and permeability. Geophys Res Lett 2008;35:L22402. http://dx.doi.
org/10.1029/2008GL035622.
[149] Segall P, Pollard DD. Nucleation and growth of strike slip faults in granite. J
Geophys Res 1983;88:555. http://dx.doi.org/10.1029/JB088iB01p00555.
[150] Renshaw CE, Pollard DD. Numerical simulation of fracture set formation: a
fracture mechanics model consistent with experimental observations. J
Geophys Res 1994;99:9359–72. http://dx.doi.org/10.1029/94JB00139.
[151] Paluszny A, Matthäi SK. Numerical modeling of discrete multi-crack growth
applied to pattern formation in geological brittle media. Int J Solids Struct
2009;46:3383–97. http://dx.doi.org/10.1016/j.ijsolstr.2009.05.007.
[152] Atkinson BK. Subcritical crack growth in geological materials. J Geophys Res
1984;89:4077. http://dx.doi.org/10.1029/JB089iB06p04077.
[153] Olson JE. Joint pattern development: effects of subcritical crack growth and
mechanical crack interaction. J Geophys Res 1993;98:12251–65. http://dx.
doi.org/10.1029/93JB00779.
[154] Paluszny A, Zimmerman RW. Numerical simulation of multiple 3D fracture
propagation using arbitrary meshes. Comput Meth Appl Mech Eng
2011;200:953–66. http://dx.doi.org/10.1016/j.cma.2010.11.013.
[155] Olson JE. Predicting fracture swarms – the influence of subcritical crack
growth and the crack-tip process zone on joint spacing in rock. In: Cosgrove
JW, Engelder T, editors. Initiation, propagation, and arrest of joints and other
fractures, vol. 231. London: Geological Society, London, Special Publications;
2004. p. 73–88. http://dx.doi.org/10.1144/GSL.SP.2004.231.01.05.
[156] Olson JE. Fracture aperture, length and pattern geometry development under
biaxial loading: a numerical study with applications to natural, cross-jointed
systems. In: Lewis H, Couples GD, editors. Relationship between damage
localization, vol. 289. London: Geological Society, London, Special
Publications; 2007. p. 123–42. http://dx.doi.org/10.1144/SP289.8.
[157] Renshaw CE. Influence of subcritical fracture growth on the connectivity of
fracture networks. Water Resour Res 1996;32:1519–30. http://dx.doi.org/
10.1029/96WR00711.
[158] Olson JE, Pollard DD. The initiation and growth of en échelon veins. J Struct
Geol 1991;13:595–608. http://dx.doi.org/10.1016/0191-8141(91)90046-L.
[159] Olson JE, Laubach SE, Lander RH. Combining diagenesis and mechanics to
quantify fracture aperture distributions and fracture pattern permeability. In:
Lonergan L, Jolly RJH, Rawnsley K, Sanderson DJ, editors. Fractured reservoirs,
vol. 270. Geological Society, London, Special Publications; 2007. p. 101–16.
http://dx.doi.org/10.1144/GSL.SP.2007.270.01.08.
[160] Paluszny A, Zimmerman RW. Numerical fracture growth modeling using
smooth surface geometric deformation. Eng Fract Mech 2013;108:19–36.
http://dx.doi.org/10.1016/j.engfracmech.2013.04.012.
[161] Paluszny A, Matthai SK. Impact of fracture development on the effective
permeability of porous rocks as determined by 2-D discrete fracture growth
modeling. J Geophys Res 2010;115:B02203. http://dx.doi.org/10.1029/
2008JB006236.
[162] Nick HM, Paluszny A, Blunt MJ, Matthäi SK. Role of geomechanically grown
fractures on dispersive transport in heterogeneous geological formations.
Phys Rev E 2011;84:056301. http://dx.doi.org/10.1103/PhysRevE.84.056301.
[163] Cowie PA, Vanneste C, Sornette D. Statistical physics model for the
spatiotemporal evolution of faults. J Geophys Res 1993;98:21809–21.
http://dx.doi.org/10.1029/93jb02223.
[164] Cowie PA, Sornette D, Vanneste C. Multifractal scaling properties of a growing
fault population. Geophys J Int 1995;122:457–69. http://dx.doi.org/10.1111/
j.1365-246X.1995.tb07007.x.
[165] Tang CA, Zhang YB, Liang ZZ, Xu T, Tham LG, Kou SQ, et al. Fracture spacing in
layered materials and pattern transition from parallel to polygonal fractures.
Phys Rev E 2006:1–9. http://dx.doi.org/10.1103/PhysRevE.73.056120.
[166] Spence GH, Finch E. Influences of nodular chert rhythmites on natural
fracture networks in carbonates: an outcrop and two-dimensional discrete
element modelling study. In: Spence GH, Redfern J, Aguilera R, Bevan TG,
Cosgrove JW, Couples GD, et al., editors. Advanced study in fractured
reservoirs, vol. 374. Geological Society, London, Special Publications; 2014. p.
211–49. http://dx.doi.org/10.1144/SP374.18.
173
[167] Asahina D, Houseworth JE, Birkholzer JT, Rutqvist J, Bolander JE. Hydro-
mechanical model for wetting/drying and fracture development in
geomaterials. Comput Geosci 2014;65:13–23. http://dx.doi.org/10.1016/j.
cageo.2013.12.009.
[168] Rockfield Software Ltd. Personal communication; 2012.
[169] Guo L, Latham J-P, Xiang J, Lei Q. A numerical investigation of fracture pattern
and fracture aperture development in multi-layered rock using a combined
finite-discrete element method. In: 47th U.S. rock mechanics/geomechanics
symposium, San Francisco. p. 686.
[170] Olson JE, Yuan Q, Holder J, Rijken P. Constraining the spatial distribution of
fracture networks in naturally fractured reservoirs using fracture mechanics
and core measurements. In: SPE annual technical conference on
exhibition. Society of Petroleum Engineers; 2001. p. 71342. http://dx.doi.
org/10.2118/71342-MS.
[171] Matthäi SK, Belayneh M. Fluid flow partitioning between fractures and a
permeable rock matrix. Geophys Res Lett 2004;31:L07602. http://dx.doi.org/
10.1029/2003GL019027.
[172] Berkowitz B, Bour O, Davy P, Odling N. Scaling of fracture connectivity in
geological formations. Geophys Res Lett 2000;27:2061–4. http://dx.doi.org/
10.1029/1999GL011241.
[173] Goodman R, Taylor R, Brekke T. A model for the mechanics of jointed rock. J
Soil Mech Found Div 1968;94:637–59.
[174] Hoek E, Brown ET. Practical estimates of rock mass strength. Int J Rock Mech
Min Sci 1997;34:1165–86. http://dx.doi.org/10.1016/S1365-1609(97)80069-
X.
[175] Sitharam TG, Sridevi J, Shimizu N. Practical equivalent continuum
characterization of jointed rock masses. Int J Rock Mech Min Sci
2001;38:437–48. http://dx.doi.org/10.1016/S1365-1609(01)00010-7.
[176] Oda M. A method for evaluating the effect of crack geometry on the
mechanical behavior of cracked rock masses. Mech Mater 1983;2:163–71.
http://dx.doi.org/10.1016/0167-6636(83)90035-2.
[177] Oda M. Similarity rule of crack geometry in statistically homogeneous rock
masses. Mech Mater 1984;3:119–29. http://dx.doi.org/10.1016/0167-6636
(84)90003-6.
[178] Oda M. An equivalent continuum model for coupled stress and fluid flow
analysis in jointed rock masses. Water Resour Res 1986;22:1845–56. http://
dx.doi.org/10.1029/WR022i013p01845.
[179] Oda M, Yamabe T, Ishizuka Y, Kumasaka H, Tada H, Kimura K. Elastic stress
and strain in jointed rock masses by means of crack tensor analysis. Rock
Mech Rock Eng 1993;26:89–112. http://dx.doi.org/10.1007/BF01023618.
[180] Kobayashi A, Fujita T, Chijimatsu M. Continuous approach for coupled
mechanical and hydraulic behavior of a fractured rock mass during
hypothetical shaft sinking at Sellafield, UK. Int J Rock Mech Min Sci
2001;38:45–57. http://dx.doi.org/10.1016/S1365-1609(00)00063-0.
[181] Gan Q, Elsworth D. A continuum model for coupled stress and fluid flow in
discrete fracture networks. Geomech Geophys Geo-Energy Geo-Resour
2016;2:43–61. http://dx.doi.org/10.1007/s40948-015-0020-0.
[182] Rutqvist J, Bäckström A, Chijimatsu M, Feng X-T, Pan P-Z, Hudson J, et al. A
multiple-code simulation study of the long-term EDZ evolution of geological
nuclear waste repositories. Environ Geol 2009;57:1313–24. http://dx.doi.org/
10.1007/s00254-008-1536-1.
[183] Tham LG, Yang TH, Tang CA. Progressive failure of jointed rocks. Int J Rock
Mech Min Sci 2004;41:182–7. http://dx.doi.org/10.1016/j.
ijrmms.2004.03.039.
[184] Pan P-Z, Feng X-T, Huang X-H, Cui Q, Zhou H. Coupled THM processes in EDZ
of crystalline rocks using an elasto-plastic cellular automaton. Environ Geol
2009;57:1299–311. http://dx.doi.org/10.1007/s00254-008-1463-1.
[185] Faivre M, Paul B, Golfier F, Giot R, Massin P, Colombo D. 2D coupled HM-
XFEM modeling with cohesive zone model and applications to fluid-driven
fracture network. Eng Fract Mech 2016;159:115–43. http://dx.doi.org/
10.1016/j.engfracmech.2016.03.029.
[186] Youn D-J. Hydro-mechanical coupled simulation of hydraulic fracturing using
the extended finite element method (XFEM). Golden: Colorado School of
Mines; 2016.
[187] Jing L. Block system construction for three-dimensional discrete element
models of fractured rocks. Int J Rock Mech Min Sci 2000;37:645–59. http://
dx.doi.org/10.1016/S1365-1609(00)00006-X.
[188] Cundall PA. A computer model for simulating progressive, large-scale
movements in blocky rock systems. Proceedings of the international
symposium on rock mechanism, Nancy, France, vol. 1. p. 129–36.
[189] Cundall PA. Formulation of a three-dimensional distinct element model—Part
I. A scheme to detect and represent contacts in a system composed of many
polyhedral blocks. Int J Rock Mech Min Ellipsis 1988;25:107–16. http://dx.
doi.org/10.1016/0148-9062(88)92293-.
[190] Itasca Consulting Group Inc. UDEC (Version 5.0) manual. Minneapolis (USA);
2013.
[191] Itasca Consulting Group Inc. 3DEC (Version 5.0) manual. Minneapolis (USA);
2013.
[192] Jing L, Stephansson O. Fundamentals of discrete element methods for rock
engineering: theory and applications. Oxford: Elsevier; 2007.
[193] Bandis SC, Lumsden AC, Barton NR. Fundamentals of rock joint deformation.
Int J Rock Mech Min Sci Geomech Abstr 1983;20:249–68. http://dx.doi.org/
10.1016/0148-9062(83)90595-8.
[194] Saeb S, Amadei B. Modelling joint response under constant or variable normal
stiffness boundary conditions. Int J Rock Mech Min Sci Geomech Abstr
1990;27:213–7. http://dx.doi.org/10.1016/0148-9062(90)94330-V.
174 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176
[195] Saeb S, Amadei B. Modelling rock joints under shear and normal loading. Int J
Rock Mech Min Sci Geomech Abstr 1992;29:267–78. http://dx.doi.org/
10.1016/0148-9062(92)93660-C.
[196] Jing L, Nordlund E, Stephansson O. A 3-D constitutive model for rock joints
with anisotropic friction and stress dependency in shear stiffness. Int J Rock
Mech Min Sci Geomech Abstr 1994;31:173–8. http://dx.doi.org/10.1016/
0148-9062(94)92808-8.
[197] Souley M, Homand F, Amadei B. An extension to the Saeb and Amadei
constitutive model for rock joints to include cyclic loading paths. Int J Rock
Mech Min Sci Geomech Abstr 1995;32:101–9. http://dx.doi.org/10.1016/
0148-9062(94)00039-6.
[198] Hart R, Cundall PA, Lemos J. Formulation of a three-dimensional distinct
element model—Part II. Mechanical calculations for motion and interaction of
a system composed of many polyhedral blocks. Int J Rock Mech Min Sci
Geomech Abstr 1988;25:117–25. http://dx.doi.org/10.1016/0148-9062(88)
92294-2.
[199] Min K-B, Jing L. Numerical determination of the equivalent elastic
compliance tensor for fractured rock masses using the distinct element
method. Int J Rock Mech Min Sci 2003;40:795–816. http://dx.doi.org/
10.1016/S1365-1609(03)00038-8.
[200] Min K-B, Jing L. Stress dependent mechanical properties and bounds of
Poisson’s ratio for fractured rock masses investigated by a DFN-DEM
technique. Int J Rock Mech Min Sci 2004;41:3–8. http://dx.doi.org/10.1016/
j.ijrmms.2003.12.072.
[201] Noorian-Bidgoli M, Zhao Z, Jing L. Numerical evaluation of strength and
deformability of fractured rocks. J Rock Mech Geotech Eng 2013;5:419–30.
http://dx.doi.org/10.1016/j.jrmge.2013.09.002.
[202] Noorian-Bidgoli M, Jing L. Anisotropy of strength and deformability of
fractured rocks. J Rock Mech Geotech Eng 2014;6:156–64. http://dx.doi.org/
10.1016/j.jrmge.2014.01.009.
[203] Noorian-Bidgoli M, Jing L. Stochastic analysis of strength and
deformability of fractured rocks using multi-fracture system realizations.
Int J Rock Mech Min Sci 2015;78:108–17. http://dx.doi.org/10.1016/j.
ijrmms.2015.05.010.
[204] Le Goc R, Darcel C, Davy P, Pierce M, Brossault MA. Effective elastic
properties of 3D fractured systems. In: Proceedings of the 1st
international conference on discrete fracture network engineering,
Vancouver, Canada. p. 142.
[205] Havaej M, Coggan J, Stead D, Elmo D. A combined remote sensing–numerical
modelling approach to the stability analysis of Delabole Slate Quarry,
Cornwall, UK. Rock Mech Rock Eng 2016;49:1227–45. http://dx.doi.org/
10.1007/s00603-015-0805-z.
[206] Min K-B, Rutqvist J, Tsang C-F, Jing L. Stress-dependent permeability of
fractured rock masses: a numerical study. Int J Rock Mech Min Sci
2004;41:1191–210. http://dx.doi.org/10.1016/j.ijrmms.2004.05.005.
[207] Tsang C-F, Rutqvist J, Min K-B. Fractured rock hydromechanics: from
borehole testing to solute transport and CO2 storage. In: David C, Le
Ravalec-Dupin M, editors. Rock physics and geomechanics in the study of
reservoirs and repositories, vol. 284. London: Geological Society, London,
Special Publications; 2007. p. 15–34. 10.1144/SP284.2.
[208] Zhao Z, Jing L, Neretnieks I. Evaluation of hydrodynamic dispersion
parameters in fractured rocks. J Rock Mech Geotech Eng 2010;2:243–54.
http://dx.doi.org/10.3724/SP.J.1235.2010.00243.
[209] Zhao Z, Jing L, Neretnieks I, Moreno L. Numerical modeling of stress effects on
solute transport in fractured rocks. Comput Geotech 2011;38:113–26. http://
dx.doi.org/10.1016/j.compgeo.2010.10.001.
[210] Noorian-Bidgoli M, Jing L. Water pressure effects on strength and
deformability of fractured rocks under low confining pressures. Rock Mech
Rock Eng 2015;48:971–85. http://dx.doi.org/10.1007/s00603-014-0628-3.
[211] Shen B, Barton N. The disturbed zone around tunnels in jointed rock Masses.
Int J Rock Mech Min Sci 1997;34:117–25. http://dx.doi.org/10.1016/S1365-
1609(97)80037-8.
[212] Lin M, Kicker D, Damjanac B, Board M, Karakouzian M. Mechanical
degradation of emplacement drifts at Yucca Mountain-A modeling case
study-Part I: nonlithophysal rock. Int J Rock Mech Min Sci 2007;44:351–67.
http://dx.doi.org/10.1016/j.ijrmms.2006.07.011.
[213] Kazerani T, Zhao J. Micromechanical parameters in bonded particle method
for modelling of brittle material failure. Int J Numer Anal Meth Geomech
2010;34:1877–95. http://dx.doi.org/10.1002/nag.884.
[214] Kazerani T, Yang ZY, Zhao J. A discrete element model for predicting shear
strength and degradation of rock joint by using compressive and tensile test
data. Rock Mech Rock Eng 2012;45:695–709. http://dx.doi.org/10.1007/
s00603-011-0153-6.
[215] Gao FQ, Stead D. The application of a modified Voronoi logic to brittle fracture
modelling at the laboratory and field scale. Int J Rock Mech Min Sci
2014;68:1–14. http://dx.doi.org/10.1016/j.ijrmms.2014.02.003.
[216] Gao FQ, Kang HP. Effects of pre-existing discontinuities on the residual
strength of rock mass - insight from a discrete element method simulation. J
Struct Geol 2016;85:40–50. http://dx.doi.org/10.1016/j.jsg.2016.02.010.
[217] Ghazvinian E, Diederichs MS, Quey R. 3D random Voronoi grain-based
models for simulation of brittle rock damage and fabric-guided micro-
fracturing. J Rock Mech Geotech Eng 2014;6:506–21. http://dx.doi.org/
10.1016/j.jrmge.2014.09.001.
[218] Mayer JM, Stead D. Exploration into the causes of uncertainty in UDEC Grain
Boundary Models Keywords. Comput Geotech 2017;82:110–23. http://dx.
doi.org/10.1016/j.compgeo.2016.10.003.
[219] Shi G-H, Goodman RE. Two dimensional discontinuous deformation analysis.
Int J Numer Anal Methods Geomech 1985;9:541–56. http://dx.doi.org/
10.1002/nag.1610090604.
[220] Shi G, Goodman RE. Generalization of two-dimensional discontinuous
deformation analysis for forward modelling. Int J Numer Anal Meth
Geomech 1989;13:359–80. http://dx.doi.org/10.1002/nag.1610130403.
[221] Jing L. Formulation of discontinuous deformation analysis (DDA) - an implicit
discrete element model for block systems. Eng Geol 1998;49:371–81. http://
dx.doi.org/10.1016/S0013-7952(97)00069-0.
[222] Lin CT, Amadei B, Jung J, Dwyer J. Extensions of discontinuous deformation
analysis for jointed rock masses. Int J Rock Mech Min Sci Geomech Abstr
1996;33:671–94. http://dx.doi.org/10.1016/0148-9062(96)00016-2.
[223] Kim Y-I, Amadei B, Pan E. Modeling the effect of water, excavation sequence
and rock reinforcement with discontinuous deformation analysis. Int J Rock
Mech Min Sci 1999;36:949–70. http://dx.doi.org/10.1016/S0148-9062(99)
00046-7.
[224] Jing L, Ma Y, Fang Z. Modeling of fluid flow and solid deformation for
fractured rocks with discontinuous deformation analysis (DDA) method. Int J
Rock Mech Min Sci 2001;38:343–55. http://dx.doi.org/10.1016/S1365-1609
(01)00005-3.
[225] Jiang QH, Yeung MR. A model of point-to-face contact for three-dimensional
discontinuous deformation analysis. Rock Mech Rock Eng 2004;37:95–116.
http://dx.doi.org/10.1007/s00603-003-0008-x.
[226] Wu J-H, Ohnishi Y, Nishiyama S. Simulation of the mechanical behavior of
inclined jointed rock masses during tunnel construction using Discontinuous
Deformation Analysis (DDA). Int J Rock Mech Min Sci 2004;41:731–43.
http://dx.doi.org/10.1016/j.ijrmms.2004.01.010.
[227] Hatzor YH, Arzi AA, Zaslavsky Y, Shapira A. Dynamic stability analysis of
jointed rock slopes using the DDA method: King Herod’s Palace, Masada,
Israel. Int J Rock Mech Min Sci 2004;41:813–32. http://dx.doi.org/10.1016/j.
ijrmms.2004.02.002.
[228] Bakun-Mazor D, Hatzor YH, Dershowitz WS. Modeling mechanical layering
effects on stability of underground openings in jointed sedimentary rocks. Int
J Rock Mech Min Sci 2009;46:262–71. http://dx.doi.org/10.1016/j.
ijrmms.2008.04.001.
[229] Tang CA, Tang SB, Gong B, Bai HM. Discontinuous deformation and
displacement analysis: from continuous to discontinuous. Sci China
2015;58:1567–74. http://dx.doi.org/10.1007/s11431-015-5899-8.
[230] Cundall PA, Strack ODL. A discrete numerical model for granular assemblies.
Géotechnique 1979;29:47–65. http://dx.doi.org/10.1680/geot.1979.29.1.47.
[231] Itasca Consulting Group Inc. PFC (Version 5.0) manual. Minneapolis (USA);
2014.
[232] Potyondy DO, Cundall PA. A bonded-particle model for rock. Int J Rock Mech
Min Sci 2004;41:1329–64. http://dx.doi.org/10.1016/j.ijrmms.2004.09.011.
[233] Cho N, Martin CD, Sego DC. A clumped particle model for rock. Int J Rock
Mech Min Sci 2007;44:997–1010. http://dx.doi.org/10.1016/j.
ijrmms.2007.02.002.
[234] Mas Ivars D, Pierce ME, Darcel C, Reyes-Montes J, Potyondy DO, Paul Young R,
et al. The synthetic rock mass approach for jointed rock mass modelling. Int J
Rock Mech Min Sci 2011;48:219–44. http://dx.doi.org/10.1016/j.
ijrmms.2010.11.01.
[235] Lambert C, Coll C. Discrete modeling of rock joints with a smooth-joint
contact model. J Rock Mech Geotech Eng 2014;6:1–12. http://dx.doi.org/
10.1016/j.jrmge.2013.12.003.
[236] Bahaaddini M, Hagan PC, Mitra R, Hebblewhite BK. Scale effect on the shear
behaviour of rock joints based on a numerical study. Eng Geol
2014;181:212–23. http://dx.doi.org/10.1016/j.enggeo.2014.07.018.
[237] Poulsen BA, Adhikary DP, Elmouttie MK, Wilkins A. Convergence of synthetic
rock mass modelling and the Hoek-Brown strength criterion. Int J Rock Mech
Min Sci 2015;80:171–80. http://dx.doi.org/10.1016/j.ijrmms.2015.09.022.
[238] Vallejos JA, Suzuki K, Brzovic A, Mas Ivars D. Application of synthetic rock
mass modeling to veined core-size samples. Int J Rock Mech Min Sci
2016;81:47–61. http://dx.doi.org/10.1016/j.ijrmms.2015.11.003.
[239] Esmaieli K, Hadjigeorgiou J, Grenon M. Estimating geometrical and
mechanical REV based on synthetic rock mass models at Brunswick Mine.
Int J Rock Mech Min Sci 2010;47:915–26. http://dx.doi.org/10.1016/j.
ijrmms.2010.05.010.
[240] Zhang Y, Stead D. Modelling 3D crack propagation in hard rock pillars using a
synthetic rock mass approach. Int J Rock Mech Min Sci 2014;72:199–213.
http://dx.doi.org/10.1016/j.ijrmms.2014.09.014.
[241] Zhang Y, Zhao X. Characterisation of confinement effect on jointed rock
pillars using a Synthetic Rock Mass approach. Int J Numer Anal Meth
Geomech 2016;40:1690–711. http://dx.doi.org/10.1002/nag.2509.
[242] Hadjigeorgiou J, Esmaieli K, Grenon M. Stability analysis of vertical
excavations in hard rock by integrating a fracture system into a PFC model.
Tunn Undergr Sp Technol 2009;24:296–308. http://dx.doi.org/10.1016/
j.tust.2008.10.002.
[243] Kozicki J, Donzé FV. A new open-source software developed for numerical
simulations using discrete modeling methods. Comput Meth Appl Mech Eng
2008;197:4429–43. http://dx.doi.org/10.1016/j.cma.2008.05.023.
[244] Kozicki J, Donzé FV. YADE-OPEN DEM: an open-source software using a
discrete element method to simulate granular material. Eng Comput
2009;26:786–805. http://dx.doi.org/10.1108/02644400910985170.
[245] Scholtès L, Donzé F-V. A DEM model for soft and hard rocks: role of grain
interlocking on strength. J Mech Phys Solids 2013;61:352–69. http://dx.doi.
org/10.1016/j.jmps.2012.10.005.
 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176
[246] Scholtès L, Donzé F-V, Khanal M. Scale effects on strength of geomaterials,
case study: coal. J Mech Phys Solids 2011;59:1131–46. http://dx.doi.org/
10.1016/j.jmps.2011.01.009.
[247] Scholtès L, Donzé F-V. Modelling progressive failure in fractured rock masses
using a 3D discrete element method. Int J Rock Mech Min Sci 2012;52:18–30.
http://dx.doi.org/10.1016/j.ijrmms.2012.02.009.
[248] Bonilla-Sierra V, Scholtès L, Donzé FV, Elmouttie MK. Rock slope stability
analysis using photogrammetric data and DFN–DEM modelling. Acta Geotech
2015;10:497–511. http://dx.doi.org/10.1007/s11440-015-0374-z.
[249] Rockfield Software Ltd. ELFEN User’s Manual. Swansea (UK); 2011.
[250] Munjiza A. The combined finite-discrete element method. London: Wiley;
2004.
[251] Munjiza A, Owen DRJ, Bicanic N. A combined finite-discrete element method
in transient dynamics of fracturing solids. Eng Comput 1995;12:145–74.
http://dx.doi.org/10.1108/02644409510799532.
[252] Owen DRJ, Feng YT. Parallelised finite/discrete element simulation of multi-
fracturing solids and discrete systems. Eng Comput 2001;18:557–76.
[253] Klerck PA, Sellers EJ, Owen DRJ. Discrete fracture in quasi-brittle materials
under compressive and tensile stress states. Comput Meth Appl Mech Eng
2004;193:3035–56. http://dx.doi.org/10.1016/j.cma.2003.10.015.
[254] Owen DRJ, Feng YT, de Souza Neto EA, Cottrell MG, Wang F, Pires FMA, et al.
The modelling of multi-fracturing solids and particulate media. Int J Numer
Methods Eng 2004;60:317–39. http://dx.doi.org/10.1002/nme.964.
[255] Hamdi P, Stead D, Elmo D. Damage characterization during laboratory
strength testing: a 3D-finite-discrete element approach. Comput Geotech
2014;60:33–46. http://dx.doi.org/10.1016/j.compgeo.2014.03.011.
[256] Pine RJ, Coggan JS, Flynn ZN, Elmo D. The development of a new numerical
modelling approach for naturally fractured rock masses. Rock Mech Rock Eng
2006;39:395–419. http://dx.doi.org/10.1007/s00603-006-0083-x.
[257] Pine RJ, Owen DRJ, Coggan JS, Rance JM. A new discrete fracture modelling
approach for rock masses. Géotechnique 2007;57:757–66. http://dx.doi.org/
10.1680/geot.2007.57.9.757.
[258] Karami A, Stead D. Asperity degradation and damage in the direct shear test:
a hybrid FEM/DEM approach. Rock Mech Rock Eng 2008;41:229–66. http://
dx.doi.org/10.1007/s00603-007-0139-6.
[259] Elmo D, Stead D, Eberhardt E, Vyazmensky A. Applications of finite/discrete
element modeling to rock engineering problems. Int J Geomech
2013;13:565–81. http://dx.doi.org/10.1061/(ASCE)GM.1943-5622.0000238.
[260] Elmo D, Stead D. An integrated numerical modelling-discrete fracture
network approach applied to the characterisation of rock mass strength of
naturally fractured pillars. Rock Mech Rock Eng 2010;43:3–19. http://dx.doi.
org/10.1007/s00603-009-0027-3.
[261] Vyazmensky A, Stead D, Elmo D, Moss A. Numerical analysis of block caving-
induced instability in large open pit slopes: a finite element/discrete element
approach. Rock Mech Rock Eng 2010;43:21–39. http://dx.doi.org/10.1007/
s00603-009-0035-3.
[262] Havaej M, Stead D, Eberhardt E, Fisher BR. Characterization of bi-planar and
ploughing failure mechanisms in footwall slopes using numerical modelling.
Eng Geol 2014;178:109–20. http://dx.doi.org/10.1016/j.enggeo.2014.06.003.
[263] Eberhardt E, Stead D, Coggan JS. Numerical analysis of initiation and
progressive failure in natural rock slopes—the 1991 Randa rockslide. Int J
Rock Mech Min Sci 2004;41:69–87. http://dx.doi.org/10.1016/S1365-1609
(03)00076-5.
[264] Stead D, Eberhardt E, Coggan JS. Developments in the characterization of
complex rock slope deformation and failure using numerical modelling
techniques. Eng Geol 2006;83:217–35. http://dx.doi.org/10.1016/j.
enggeo.2005.06.033.
[265] Vyazmensky A, Elmo D, Stead D. Role of rock mass fabric and faulting in the
development of block caving induced surface subsidence. Rock Mech Rock
Eng 2010;43:533–56. http://dx.doi.org/10.1007/s00603-009-0069-6.
[266] Coggan J, Gao F, Stead D, Elmo D. Numerical modelling of the effects of weak
immediate roof lithology on coal mine roadway stability. Int J Coal Geol
2012;90–91:100–9. http://dx.doi.org/10.1016/j.coal.2011.11.003.
[267] Munjiza A, Andrews KRF, White JK. Combined single and smeared crack
model in combined finite-discrete element analysis. Int J Numer Meth Eng
1999;44:41–57. http://dx.doi.org/10.1002/(SICI)1097-0207(19990110)
44:1<41::AID-NME487>3.0.CO;2-A.
[268] Munjiza A, Andrews KRF. NBS contact detection algorithm for bodies of
similar size. Int J Numer Meth Eng 1998;43:131–49. http://dx.doi.org/
10.1002/(SICI)1097-0207(19980915)43:1<131::AID-NME447>3.0.CO;2-S.
[269] Munjiza A, Andrews KRF. Penalty function method for combined finite-
discrete element systems comprising large number of separate bodies. Int J
Numer Meth Eng 2000;49:1377–96. http://dx.doi.org/10.1002/1097-0207
(20001220)49:11<1377::AID-NME6>3.0.CO;2-B.
[270] Munjiza A, Knight EE, Rougier E. Computational mechanics of
discontinua. Chichester: John Wiley & Sons; 2011.
[271] Munjiza A, Knight EE, Rougier E. Large strain finite element method: a
practical course. Chichester: John Wiley & Sons; 2015.
[272] Rougier E, Knight EE, Broome ST, Sussman AJ, Munjiza A. Validation of a
three-dimensional finite-discrete element method using experimental
results of the split hopkinson pressure bar test. Int J Rock Mech Min Sci
2014;70:101–8. http://dx.doi.org/10.1016/j.ijrmms.2014.03.011.
[273] Mahabadi OK, Grasselli G, Munjiza A. Y-GUI: A graphical user interface and
pre-processor for the combined finite-discrete element code, Y2D,
incorporating material heterogeneity. Comput Geosci 2010;36:241–52.
http://dx.doi.org/10.1016/j.cageo.2009.05.010.
175
[274] Mahabadi OK, Lisjak A, Munjiza A, Grasselli G. Y-Geo: a new combined finite-
discrete element numerical code for geomechanical applications. Int J
Geomech 2012;153. http://dx.doi.org/10.1061/(ASCE)GM.1943-
5622.000021.
[275] Xiang J, Munjiza A, Latham J-P. Finite strain, finite rotation quadratic
tetrahedral element for the combined finite-discrete element method. Int J
Numer Meth Eng 2009;79:946–78. http://dx.doi.org/10.1002/nme.2599.
[276] Xiang J, Munjiza A, Latham J-P, Guises R. On the validation of DEM and FEM/
DEM models in 2D and 3D. Eng Comput 2009;26:673–87. http://dx.doi.org/
10.1108/02644400910975469.
[277] Munjiza A, Xiang J, Garcia X, Latham J-P, D’Albano GGS, John NWM. The
Virtual Geoscience Workbench, VGW: open source tools for discontinuous
systems. Particuology 2010;8:100–5. http://dx.doi.org/10.1016/
j.partic.2009.04.008.
[278] Lei Q, Latham J, Xiang J, Tsang C. Polyaxial stress-induced variable aperture
model for persistent 3D fracture networks. Geomech Energy Environ
2015;1:34–47. http://dx.doi.org/10.1016/j.gete.2015.03.003.
[279] Lei Q. Characterisation and modelling of natural fracture networks:
geometry, geomechanics and fluid flow [Ph.D. thesis]. London: Imperial
College London; 2016.
[280] Lei Q, Wang X, Xiang J, Latham J-P. Influence of stress on the
permeability of a three-dimensional fractured sedimentary layer. In:
Proceedings of the 50th US rock mechanics/geomechanics symposium,
Houston, USA. p. 586.
[281] Lei Q, Latham J-P, Xiang J. Implementation of an empirical joint constitutive
model into finite-discrete element analysis of the geomechanical behaviour
of fractured rocks. Rock Mech Rock Eng 2016. http://dx.doi.org/10.1007/
s00603-016-1064-3.
[282] Tatone BSA, Grasselli G. A calibration procedure for two-dimensional
laboratory-scale hybrid finite–discrete element simulations. Int J Rock
Mech Min Sci 2015;75:56–72. http://dx.doi.org/10.1016/j.
ijrmms.2015.01.011.
[283] Mahabadi OK, Kaifosh P, Marschall P, Vietor T. Three-dimensional FDEM
numerical simulation of failure processes observed in Opalinus Clay
Laboratory samples. J Rock Mech Geotech Eng 2014;6:591–606. http://dx.
doi.org/10.1016/j.jrmge.2014.10.005.
[284] Guo L. Development of a three-dimensional fracture model for the combined
finite-discrete element method [Ph.D. thesis]. London: Imperial College
London; 2014.
[285] Guo L, Latham J-P, Xiang J. Numerical simulation of breakages of concrete
armour units using a three-dimensional fracture model in the context of the
combined finite-discrete element method. Comput Struct 2015;146:117–42.
http://dx.doi.org/10.1016/j.compstruc.2014.09.001.
[286] Guo L, Xiang J, Latham J-P, Izzuddin B. A numerical investigation of mesh
sensitivity for a new three-dimensional fracture model within the combined
finite-discrete element method. Eng Fract Mech 2016;151:70–91. http://dx.
doi.org/10.1016/j.engfracmech.2015.11.006.
[287] Munjiza A, Latham J-P, Andrews KRF. Detonation gas model for combined
ÿnite-discrete element simulation of fracture and fragmentation. Int J Numer
Meth Eng 2000:1495–520. http://dx.doi.org/10.1002/1097-0207(20001230)
49:12<1495::AID-NME7>3.0.CO;2-5.
[288] Lisjak A, Grasselli G, Vietor T. Continuum–discontinuum analysis of failure
mechanisms around unsupported circular excavations in anisotropic clay
shales. Int J Rock Mech Min Sci 2014;65:96–115. http://dx.doi.org/10.1016/j.
ijrmms.2013.10.006.
[289] Lisjak A, Figi D, Grasselli G. Fracture development around deep underground
excavations: insights from FDEM modelling. J Rock Mech Geotech Eng
2014;6:493–505. http://dx.doi.org/10.1016/j.jrmge.2014.09.003.
[290] Lisjak A, Tatone BSA, Mahabadi OK, Grasselli G, Marschall P, Lanyon
GW, et al. Hybrid finite-discrete element simulation of the EDZ
formation and mechanical sealing process around a microtunnel in
Opalinus Clay. Rock Mech Rock Eng 2015. http://dx.doi.org/10.1007/
s00603-015-0847-2.
[291] Lisjak A, Garitte B, Grasselli G, Müller HR, Vietor T. The excavation of a
circular tunnel in a bedded argillaceous rock (Opalinus Clay): short-term rock
mass response and FDEM numerical analysis. Tunn Undergr Space Technol
2015;45:227–48. http://dx.doi.org/10.1016/j.tust.2014.09.014.
[292] Barla M, Piovano G, Grasselli G. Rock slide simulation with the combined
finite-discrete element method. Int J Geomech 2012;12:711–21. http://dx.
doi.org/10.1061/(ASCE)GM.1943-5622.0000204.
[293] Lisjak A, Tatone BSA, Grasselli G, Vietor T. Numerical modelling of the
anisotropic mechanical behaviour of Opalinus Clay at the laboratory-scale
using FEM/DEM. Rock Mech Rock Eng 2014;47:187–206. http://dx.doi.org/
10.1007/s00603-012-0354-7.
[294] Gao K, Harrison JP, Lei Q, Latham J-P. Influence of boundary constraint
stiffness on stress heterogeneity modelling. In: Proceedings of the 50th US
rock mechanics/geomechanics symposium, Houston, USA. p. 178.
[295] Witherspoon PA, Wang JSY, Iwai K, Gale JE. Validity of cubic law for fluid flow
in a deformable rock fracture. Water Resour Res 1980;16:1016–24. http://dx.
doi.org/10.1029/WR016i006p01016.
[296] Jing L, Min K-B, Baghbanan A, Zhao Z. Understanding coupled stress, flow and
transport processes in fractured rocks. Geosyst Eng 2013;16:2–25. http://dx.
doi.org/10.1080/12269328.2013.780709.
[297] Berkowitz B. Characterizing flow and transport in fractured geological media:
a review. Adv Water Resour 2002;25:861–84. http://dx.doi.org/10.1016/
S0309-1708(02)00042-8.
176 Q. Lei et al. / Computers and Geotechnics 85 (2017) 151–176
[298] Renard P, de Marsily G. Calculating equivalent permeability: a review. Adv
Water Resour 1997;20:253–78. http://dx.doi.org/10.1016/S0309-1708(96)
00050-4.
[299] Bai T, Pollard DD. Getting more for less: the unusual efficiency of fluid flow in
fractures. Geophys Res Lett 2001;28:65–8. http://dx.doi.org/10.1029/
2000GL006124.
[300] Moreno L, Tsang YW, Tsang CF, Hale FV, Neretnieks I. Flow and tracer
transport in a single fracture: a stochastic model and its relation to some field
observations. Water Resour Res 1988;24:2033. http://dx.doi.org/10.1029/
WR024i012p02033.
[301] Wang Z, Rutqvist J, Wang Y, Leung C, Hoch A, Dai Y. The effect of stress on
flow and transport in fractured rock masses using an extended multiple
interacting continua method with crack tensor theory. Nucl Technol
2014;187:1–8. http://dx.doi.org/10.13182/NT13-76.
[302] Kang PK, Lei Q, Juanes R. Impact of stress on anomalous transport in a
fractured rock: from a single rough fracture to discrete fracture networks. In:
Proceedings of the 12th world congress on computational mechanism, Seoul,
Korea. p. 151232.
[303] Rogers S, Elmo D, Dunphy R, Bearinger D. Understanding hydraulic fracture
geometry and interactions in the Horn River Basin through DFN and
numerical modeling. In: International petroleum conference on the
proceedings of Canadian unconventional resources, Calgary. p. 137488.
http://dx.doi.org/10.2118/137488-MS.
[304] Kresse O, Weng X, Gu H, Wu R. Numerical modeling of hydraulic fractures
interaction in complex naturally fractured formations. Rock Mech Rock Eng
2013;46:555–68. http://dx.doi.org/10.1007/s00603-012-0359-2.
[305] Nagel NB, Sanchez-Nagel MA, Zhang F, Garcia X, Lee B. Coupled numerical
evaluations of the geomechanical interactions between a hydraulic
fracture stimulation and a natural fracture system in shale formations.
Rock Mech Rock Eng 2013;46:581–609. http://dx.doi.org/10.1007/s00603-
013-0391-x.
[306] Fu P, Johnson SM, Carrigan CR. An explicitly coupled hydro-geomechanical
model for simulating hydraulic fracturing in arbitrary discrete fracture
networks. Int J Numer Anal Meth Geomech 2013;37:2278–300. http://dx.doi.
org/10.1002/nag.2135.
[307] Zou Y, Zhang S, Ma X, Zhou T, Zeng B. Numerical investigation of hydraulic
fracture network propagation in naturally fractured shale formations. J Struct
Geol 2016;84:1–13. http://dx.doi.org/10.1016/j.jsg.2016.01.004.
[308] Obeysekara A, Lei Q, Salinas P, Pavlidis D, Latham J-P, Xiang J, et al. A fluid-
solid coupled approach for numerical modeling of near-wellbore hydraulic
fracturing and flow dynamics with adaptive mesh refinement. In:
Proceedings of the 50th US rock mechanics/geomechanics symposium,
Houston, USA. p. 13.
[309] Barton CA, Zoback MD, Moos D. Fluid flow along potentially active faults in
crystalline rock. Geology 1995;23:683–6. http://dx.doi.org/10.1130/0091-
7613(1995) 023<0683:FFAPAF>2.3.CO;2.
[310] Zhang X, Sanderson DJ. Scale up of two-dimensional conductivity tensor for
heterogenous fracture networks. Eng Geol 1999;53:83–99.
[311] Blum P, Mackay R, Riley MS, Knight JL. Performance assessment of a nuclear
waste repository: upscaling coupled hydro-mechanical properties for far-
field transport analysis. Int J Rock Mech Min Sci 2005;42:781–92. http://dx.
doi.org/10.1016/j.ijrmms.2005.03.015.
[312] Blum P, Mackay R, Riley MS. Stochastic simulations of regional scale
advective transport in fractured rock masses using block upscaled hydro-
mechanical rock property data. J Hydrol 2009;369:318–25. http://dx.doi.org/
10.1016/j.jhydrol.2009.02.009.
[313] Read RS. 20 years of excavation response studies at AECL’s Underground
Research Laboratory. Int J Rock Mech Min Sci 2004;41:1251–75. http://dx.
doi.org/10.1016/j.ijrmms.2004.09.012.
[314] Hudson JA, Bäckström A, Rutqvist J, Jing L, Backers T, Chijimatsu M, et al.
Characterising and modelling the excavation damaged zone in crystalline
rock in the context of radioactive waste disposal. Environ Geol
2009;57:1275–97. http://dx.doi.org/10.1007/s00254-008-1554-z.
[315] Oliver DS, Chen Y. Recent progress on reservoir history matching: a review.
Comput Geosci 2011;15:185–221. http://dx.doi.org/10.1007/s10596-010-
9194-2.
[316] Wang X, Jardani A, Jourde H, Lonergan L, Cosgrove J, Gosselin O, et al.
Characterisation of the transmissivity field of a fractured and karstic aquifer,
Southern France. Adv Water Resour 2016;87:106–21. http://dx.doi.org/
10.1016/j.advwatres.2015.10.014.