2021JB022348 Skib BNGST

RESEARCH ARTICLE How Does In Situ Stress Rotate Within a Fault Zone?
10.1029/2021JB022348
Insights From Explicit Modeling of the Frictional,
Key Points:
• S tress variations in fault zones
Fractured Rock Mass
are explicitly simulated with a 2D Shihuai Zhang1 and Xiaodong Ma1
multilayer model incorporating
macroscopic fractures Department of Earth Sciences, ETH Zürich, Zürich, Switzerland
1
• Fracture deformation and boundary
conditions synergistically contribute
to the variations of stress and fault
zone elastic properties Abstract We quantitatively investigate the spatial stress variations within the fault zone rock mass
• The major principal stress always by explicitly incorporating macroscopic fractures into a 2D multilayer model. Based on elastic crack
rotates to a limiting angle of 45° to theory, we first derive a unified constitutive law for frictional fractures, featuring elastic and plastic shear
the fault plane, upon which plastic
flow occurs deformation and shear-induced dilatancy. To honor the varying degrees of damage across fault zones, the
multilayer model is composed of varying densities of randomly oriented frictional fractures from layer
to layer. Under the boundary conditions specific to fault zones, the global mechanical response of each
Correspondence to:
layer is quantitatively related to the local fracture deformation. We show that the major principal stress
X. Ma,
xiaodong.ma@erdw.ethz.ch always rotates toward a limiting angle of 45° with respect to the fault plane and that the differential stress
invariantly decreases with increasing fracture density. Approaching the fault core, the mean stress can
Citation: either increase or decrease, depending on whether the fault strikes at a high (>45°) or low (<45°) angle to
Zhang, S., & Ma, X. (2021). How does the regional major principal stress. Accumulated damage also results in the decrease and increase of the
in situ stress rotate within a fault effective Young's modulus and Poisson's ratio of the fractured rock mass, respectively. Both the fracture
zone? Insights from explicit modeling
properties and pore pressure affect the stress variations by modulating the fracture-associated deformation
of the frictional, fractured rock mass.
Journal of Geophysical Research: Solid and the relative proportion of the elastic and plastic components. Our model illuminates the systematic
Earth, 126, e2021JB022348. https://doi. variations of in situ stresses and effective elastic properties within the damage zone of a mature fault.
org/10.1029/2021JB022348
Plain Language Summary Knowledge of the stress state around faults is of fundamental
Received 3 MAY 2021
Accepted 4 NOV 2021 importance to earthquake processes and subsurface engineering. Stress magnitude and orientation
can vary significantly near a fault, however, which is not thoroughly understood yet. Due to such
Author Contributions:
stress variations, fault slips can occur on some seemingly stable faults, which may compromise certain
Formal analysis: Shihuai Zhang engineering design schemes (e.g., underground excavation, hydraulic fracturing, etc.). Here, we propose
Investigation: Shihuai Zhang, numerical models to simulate such stress variations. We demonstrate that the stress variations and rock
Xiaodong Ma mass softening accompany each other toward the center of a fault. Our simulations illuminate systematic
Methodology: Shihuai Zhang
Software: Shihuai Zhang variations in the stress magnitude and orientation around a fault zone.
Supervision: Xiaodong Ma
Visualization: Shihuai Zhang
Writing – original draft: Shihuai 1. Introduction
Zhang
Writing – review & editing: Xiaodong Characterization of the in situ stresses is key to better understand the crustal deformation processes, such
Ma
as fault slip (Scholz, 2019) and to tailor subsurface engineering designs (Cornet et al., 2007; Ma & Zo-
back, 2017). In general, the stress field within the intra-plate area is relatively uniform at regional scales
(M. L. Zoback, 1992). However, the presence of ubiquitous discontinuities (e.g., veins, joints, fractures, and
faults) at various scales in the crustal rock mass modifies the local stress fields (Pollard & Segall, 1987). In
particular, faults, as complex geological structures, can exert significant influence on and interact with the
local stress fields in an intricate manner.
The state of stress is often not adequately understood in the vicinity of faults due to the complexity of
the stress conditions and the availability of stress measurements therein (Stephansson & Zang, 2012). In
© 2021 The Authors. general, stress fields at great depths are often inferred from earthquake focal mechanism inversions (Mi-
This is an open access article under
the terms of the Creative Commons chael, 1984; Vavryčuk, 2014), while borehole/drillcore measurements are typically available for the stress
Attribution-NonCommercial License, estimation at shallow depths (Funato & Ito, 2017; Haimson & Cornet, 2003; Pierdominici & Heidbach, 2012;
which permits use, distribution and Sjöberg et al., 2003). The integration of these stress estimation methodologies has provided ample field
reproduction in any medium, provided
the original work is properly cited and evidence, indicating that regional far-field stresses can rotate significantly toward major tectonic faults
is not used for commercial purposes. (Lin et al., 2010; Zoback et al., 1987) and reservoir-scale faults (Obara & Sugawara, 2003; Tamagawa &
ZHANG AND MA 1 of 23
Journal of Geophysical Research: Solid Earth 10.1029/2021JB022348
Table 1
Compilation of Measured Elastic Property Contrasts in Selected Fault Zones
Fault zone thickness EHR, GPa EDZ, GPa EFC, GPa vHR vFC References
∼15 km 75 – 30–45 0.25 – Schmalzle et al. (2006)
100 m 77 ± 6.3 – 52.0 ± 5.7 0.21 ± 0.06 0.23 Heesakkers et al. (2011)
∼50 m 55.4–57.3 22.6–27.1 16.2 0.24 0.26–0.39 Isaacs et al. (2008)
40 m – 7.6–17 4.2 – – Bauer et al. (2015)
30 m 30–40 >25 14–19 0.3 0.34–0.37 Carpenter et al. (2014)
20 m 40 – 22 – – Jeanne et al. (2012)
60 13
10 m 5 1 0.4 – – Jeanne et al. (2017)
47 6.5–12.8 6
10 m 54–88 – 26–46 0.15–0.27 0.16–0.3 Leclère et al. (2015)
10 m 54.0 a
2.8 a
0.03 a
– – Cappa et al. (2007)
26.4b 1.4b 0.02b
Note. E: Young's modulus, v: Poisson's ratio, HR: host rock, DZ: damage zone, FC: fault core.
a
Bulk modulus. bShear modulus.
Pollard, 2008; Yale, 2003). The mechanisms responsible for such stress rotations are complicated. It is likely
the result of a self-adjustment of the stress field, depending on the fault structure, fault zone rock mass
rheology, and in situ pore pressure (Faulkner et al., 2010; Stephansson & Zang, 2012; M. D. Zoback, 2010).
A mature fault in the Earth's upper crust can be simplified as a fault core sandwiched by two damage zones
(Caine et al., 1996; Chester et al., 1993; Mitchell & Faulkner, 2009), which border the host rock. The fault
core is composed of highly comminuted rocks that accommodates the majority of the cumulative shear
slip; the damage zones feature decreasing fracture density away from the fault core with only limited shear
displacement (e.g., Faulkner et al., 2003; Lockner et al., 2009), which is accompanied by the variations of
the elastic properties and permeability of the rock mass therein (Faulkner et al., 2010; Lockner et al., 2009).
Table 1 shows that Young's modulus of the in situ rock mass generally decreases monotonically from the
host rock, through the damage zone, toward the fault core, while the change of Poisson's ratio shows an
opposite trend. Based on these characteristics, a multilayer model was proposed to simulate the stress varia-
tions in a strike-slip fault damage zone (Faulkner et al., 2006). In their model, the stress rotations are chiefly
attributed to the elastic property changes as originally suggested by Casey (1980), but further related to the
gradual increase of microcrack density as the fault core is approached.
The multilayer model has been mainly applied to the strike-slip faults oriented at high angles to the major
dicts a rotation of SHmax from a large angle (∼80°) to ∼40° and decreased mean stress from afar to the fault
principal stress S1 (i.e., the maximum horizontal stress SHmax) at far-field (e.g., Faulkner et al., 2006). It pre-
plane. Such stress variations plausibly provide static weakening mechanisms for the unfavorably oriented
faults. On the other hand, stress variations around faults that are oriented at low angles (<45°) to the far-
field SHmax are less frequently studied. The multilayer model predicts, in the latter case, that SHmax rotates
further to the fault plane (decreasing angle) and the mean stress increases toward the fault core. However,
among other examples, a series of elaborate stress measurements recently at the Grimsel Test Site (GTS;
et al. (2019) observed that, as a subvertical fault zone is approached, SHmax rotates markedly from ∼15° to
Krietsch et al., 2019) have revealed an opposite trend to the prediction of the multilayer model. Krietsch
∼45° with respect to the fault strike direction, and that both horizontal stress components (and thus the
mean stress) decrease significantly. Such a contradiction calls for systematic study on the stress variations
around fault zones. The multilayer model requires further examination and further developments to accom-
modate different scenarios.
To this end, the objective of this study is to improve the understanding and prediction capability of stress
variations around faults, oriented at different angles with respect to the far-field S1. With the multilayer
Figure 1. Constitutive relationship of a frictional fracture under plane strain condition. (a) Geometric features and local stress state of a fracture. (b) Normal
stress versus normal displacement. When a fracture is subject to uniform normal compression, its normal displacement is controlled by the normal stiffness of
the fracture walls, while the resistance to the fracture opening is provided by the matrix. (c) Shear stress versus shear displacement. Shear deformation of the
fracture consists of elastic and plastic components; the latter emerges when the shear stress τf exceeds its frictional strength (τμ = μ·σn).
model, we particularly focus on the characterization of individual layers using fractal discrete fractures
rather than microcracks as in previous studies. The fractal fractures within the fault rock mass are likely to
cover a wide range of scales (e.g., Ostermeijer et al., 2020). Since friction is important to the fractured rock
mass deformation, we explicitly incorporate the elastic and frictional properties of fractures, and quantita-
tively relate fracture deformation to stress variations as well as effective elastic properties through layers.
This study is organized as follows. We first propose a unified constitutive law for frictional fractures. Then
we build a multilayer model and derive the deformation of fractures in each layer, honoring specific bound-
ary conditions. The fracture deformation is further related to the effective elastic properties and stresses of
each layer. In particular, we investigate the effect of fracture density, fracture stiffness, fracture friction, and
pore pressure on the variations of effective elastic properties and stresses. Finally, we discuss the assump-
tions associated with the simulations, the stress rotation limit predicted by the multilayer model, and the
spatio-temporal variations of the in situ stress within fault zones.
2. Constitutive Behavior of Frictional Fractures

In this section, the mechanical behavior of a frictional fracture under plane strain condition (Zhang &
Ma, 2021) is extended to include elastic deformation. In particular, a complete constitutive law for the fric-
tional fracture is derived under loading and unloading conditions, respectively, which is used to calculate
the fracture-induced stress changes inside the multilayer model in Section 3.
2.1. Stress-Displacement Relationship for Fractures Under Loading Condition
We consider an elastic body in a state of plane strain with only in-plane displacements for an embedded
fracture, as illustrated in Figure 1a. The planar fracture has a normal unit vector n in the xn coordinate
direction, negligible thickness, and a length of l. The surrounding elastic matrix is linear, homogeneous,
and isotropic with Poisson's ratio v and shear modulus G. Under the remote (effective) normal and shear
stresses (σn and τ), the relative normal and shear displacements (un and us) between the fracture walls can
be quantified by the following theoretical derivations. All parameters and corresponding symbols are sum-
marized in Table 2.
If the normal stress is compressive (σn > 0), the displacement normal to the fracture wall is given as:
un   n / kn
(1)
where kn is the normal stiffness of the fracture walls. When the fracture contains high pore pressure, the
resultant normal stress σn acting on the fracture wall can be tensional (σn < 0). In this case, the pure Mode I
deformation can be well quantified by elastic crack theory (Pollard & Segall, 1987), which gives the relative
normal displacement as:
Table 2
List of Physical Parameters and Their Definitions and Values (Units) in This Study
Parameter Symbol Value Unit
Matrix Poisson's ratio v 0.25 –
Matrix shear modulus G 25 GPa
Fracture length l 0.01 ∼ 1L Unit length
Fracture half-length δ 0.5 l Unit length
Frictional coefficient of fracture μ 0.2; 0.4; 0.6 –
Fracture normal stiffness kn 12,000 GPa/m
Fracture shear stiffness ks 10; 20; 30 GPa/m
Effective normal stress acting on fracture σn MPa
Remote shear stress acting on fracture τ MPa
Shear stress on fracture walls τf MPa
Fracture frictional strength τμ μ·σn MPa
Fracture normal displacement un Unit length
Fracture shear displacement us Unit length
Fracture dilatancy factor β 0.05 –
Matrix resistance to fracture opening (average) E km (km) GPa/m
Far-field major principal stress S1 MPa
Far-field minor principal stress S3 MPa
Pore pressure Pp MPa
Angle between the fault strike and S1 θ 10; 30; 60; 80 °
Initial effective normal stress of each layer σxx σyy MPa
Initial shear stress of each layer σxy σyx MPa
Constant strain along the strike direction εyy –
Length of the representative rock mass volume L unit length
Density term in the fracture length distribution α 0; 0.32; 0.8; 1.6; 2.4; 3.2; 4 –
Fractal dimension in the fracture length distribution D 3 –
Power law length exponent in the fracture length distribution a 2 –
un   n / km
(2)
The matrix resistance to the opening of fracture walls km is given by:
G
 
0.5
km
(3)  2  xs2
2 1  v 
where δ is the half-length of the fracture and the local coordinate xs ∈ [−δ, δ].
Realistically, the matrix resistance km varies from fracture to fracture and along the fracture wall. The av-
erage matrix resistance
E (km) for a specific fracture can be further obtained by integrating Equation 3 across
the whole fracture length:
1  G 2G
 
0.5
km 
(4)    2  xs2 dxs 
2 2 1  v   1  v 
Therefore, Equation 2 can be written as:
un   n / km
(5)
The relationship between normal stress and normal displacement in Equations 1 and 5 is illustrated in
Figure 1b.
The shear deformation of a fracture is assumed to be driven by the stress difference between the remote
shear stress τ and the shear stress on the fracture τf (Pollard & Segall, 1987). In this paper, we extend the
shear behavior of fractures by including the elastic shear deformation before the fracture reaches its fric-
tional sliding state. Specifically, when the shear stress on the fracture τf is less than its frictional strength τμ
(=μ·σn), the fracture is in the elastic stage, where the relative shear displacement is:
us   f / ks
(6)
with ks the fracture shear stiffness, which is assumed to be constant over the whole fracture plane. Simulta-
neously, the elastic matrix also experiences the same shear deformation, leading to the matrix shear stress
τm:

(7)
m km  us
Considering the concept of shear stress partition (Davy et al., 2018), the remote shear stress τ can be ex-
pressed as the sum of fracture shear stress τf and matrix shear stress τm:
  f  m

(8)
By substituting τf and τm into Equations 6 and 7, we further obtain:


us 
(9) ks  km
If τf reaches the frictional strength τμ (=μ·σn), frictional sliding occurs and the shear motion is driven by the
shear stress difference. Based on elastic crack theory, the shear displacement is given as:
      n
us 
(10) 
km km
In Figure 1c, the shear stress-shear displacement relationship defined jointly by Equations 9 and 10 is
shown. Straightforwardly, we propose the following criterion to estimate whether a fracture is in the elastic
or the plastic stage:
 km  ks
     , elastic
 ks

(11)
    km  ks , plastic

 ks
To honor shear-induced dilatancy commonly observed in the brittle rock masses, we further utilize the dila-
tancy factor β to relate the normal dilational displacement to the plastic shear displacement when frictional
sliding occurs. The relative normal displacement defined in Equation 1 can be modified as:
n
un 
(12)kn
   us    /ks  
which further gives the resultant displacement vector:
u un n  us s
(13)
It should be noted that the normal stiffness kn for natural fractures is generally much larger than the shear
stiffness ks. Therefore, we can neglect the compressive normal displacement for computational efficiency
without much loss of accuracy, which has been numerically verified by Davy et al. (2018).
2.2. Stress-Displacement Relationship for Fractures Under Unloading Condition
Unloading occurs on a fracture due to the decrease of remote shear (Δτ) or normal (Δσn) stress. For a frac-
ture in the elastic stage, the corresponding decrease of its shear stress Δτf is:
Δ
Δ f 
(14) ks
ks  km
For a fracture in the plastic stage, however, two scenarios are possible during its unloading process. First,
the unloaded fracture is still in the plastic stage. The shear stress change on the fracture Δτf is:
Δ f   Δ n
(15)
Second, the unloaded fracture returns to the elastic stage. Accordingly, the shear stress reduction on the
fracture Δτf now has a different expression:
  Δ
Δ f  Δ n 

(16) ks
ks  km
Despite the fact that fractures feature high nonlinearity (e.g., David et al., 2012), it is hypothesized that the
shear stiffness upon unloading is the same as the loading stiffness (Figure 1c), since unloading is secondary
to the main issue in this study. The shear stress reduction on the fracture defined by Equation 14 through
Equation 16 can be adopted to calculate the recovered elastic shear displacement Δus during unloading:
Δus  Δ f / ks
(17)
With regard to the normal displacement, it consists of dilational and elastic components. The former is
often assumed to be plastic and irreversible, while the latter is reversible. Under unloading condition, the
recovered elastic normal displacement Δun is expressed as:
Δun  Δ n / kn
(18)
3. Effective Properties of Fractured Rock Mass Within a Fault Damage Zone

We first briefly review the basic assumptions and boundary conditions of the multilayer model for fault
damage zone. For each layer of the model, we calculate the displacements and strains of individual frac-
tures according to the proposed fracture constitutive relationship, and relate them to the boundary strain
response using effective medium theory. Due to specific boundary conditions, we further propose a new
method to update the stress field of each layer to accommodate the incremental strain component induced
by fractures. In this way, it resolves the final stress field and effective elastic properties of each layer.
3.1. Boundary Conditions of a Fault Damage Zone
Existing multilayer models (e.g., Casey, 1980; Faulkner et al., 2006) assume that fault damage zone is subject to
constant strain along the slip direction and to constant stress in the off-fault direction. Following such assump-
tions, the conceptual multilayer model is shown schematically in Figure 2a for a completely unlocked section
of a mature strike-slip fault zone. Given the applied far-field stresses, the stress components for the outermost
host rock can be resolved based on the Mohr diagram, as shown in Figure 2b. For these consecutive layers, the
normal and shear stresses (Sxx and Sxy) acting on the interfaces are constant for mechanical continuity, while
constant strain εyy is applied in the fault strike direction assuming no slip between layers. Such boundary con-
ditions also apply to dip-slip faults featuring a fault plane parallel to the intermediate principal stress.
Specifically, we set the outermost layer representing intact rock as a benchmark, which gives the following
far-field stress components:
S1  S3 S1  S3
S xx
(19)  cos 2
2 2
Figure 2. (a) Configuration and boundary conditions of the multilayer model for a strike-slip fault zone. The model is
configured under plane strain condition (εzz = 0) so that the intermediate principal stress (vertical stress) is not relevant.
(b) Far-field state of stress and the corresponding stress components in the local coordinate system represented by
the Mohr diagram. The inset shows the initial stress state and boundary conditions applied to individual layers in the
simulations.
S1  S3 S1  S3
S yy
(20)  cos 2
2 2
S  S3
S xy  1
(21) sin 2
2
where S1 and S3 are the far-field total major and minor principal stresses, respectively, and θ is the angle
between the fault strike direction and S1. In the following, effective stresses (σxx = Sxx − Pp, σyy = Syy − Pp,
with pore pressure Pp subtracted from the total stresses) are used for calculation.
3.2. Updating Stress Components on the Strain Boundary
When the model is subject to some arbitrary external stresses, the mechanical behavior of each layer de-
pends jointly on the elastic matrix and fracture deformation. In other words, the local deformation of in-
dividual fractures contributes to the total strain of the model. In Appendix A, we provide the derivations
quantifying the fracture contribution (i.e., Equation A3) to the model's global strain based on effective medi-
um theory. Note that the strain εyy in the fault strike direction is maintained as constant across all layers (Fig-
ure 2a). Therefore, the additional strain in the y-direction contributed by fractures entails the adjustment of
strain in the elastic matrix. Equivalently, σyy should be modified to accommodate the corresponding strain
increments (Zhang & Ma, 2021). The interplay between local fracture deformation and specific boundary
conditions of the model drives the stress variations in a self-regulated manner. Specifically, the bisection
method (see Appendix B) is used to search for the stress change of σyy, which gives an updated state of stress
in each layer.
In Figure 3, we graphically illustrate how the stress state evolves after updating in the scenarios of low θ
(<45°) and large θ (>45°), respectively. Note that, all layers in the multilayer model are initially subject to
the same boundary conditions as the outermost purely elastic layer. Therefore, the far-field stress/strain
conditions are consistently the initial conditions for each layer. In Figure 3a, the resolved σyy is initially
larger than σxx so that the fracture deformation induces contractional strain in the y-direction. Because of
the constant strain boundary in the y-direction, σyy needs to decrease E (to  yy) to reduce the corresponding
strain component in the elastic matrix, accommodating the fracture contribution. As a result, it leads to an
increase of θ (to θ’). By contrast, when the resolved σyy is less than σxx due to the high θ in the latter case,
E (to  yy) and thus θ decreases (to θ’). The boundary conditions dictate that the shear stresses
σyy increases
between layers are continuously equal and maintained constant, which restrict the stress paths associated
with y-direction to horizontal lines, as shown by the dash arrows in Figures 3a and 3b. Moreover, it is worth
noting that a limiting condition that σyy = σxx (zero differential stress) exists according to the minimum en-
ergy principle, which corresponds to the pure plasticity with the effective Poisson's ratio of 0.5 and effective
Figure 3. Schematics of stress update under the condition of strain boundary along the fault strike direction: (a) S1 is
initially oriented at an angle of θ < 45°. (b) S1 is initially oriented at an angle of θ > 45°. In the Mohr diagrams, effective
stress components are adopted and the initial stress state is colored in gray for reference.
Young's modulus of 0. Therefore, the major principal stress in either scenario always rotates toward the
angle of 45° with respect to the fault strike direction.
4. Model Experimentation and Results

In this section, six sets of representative rock mass volumes are generated with different fracture densities
to represent different portions within a fault zone. Based on the aforementioned methods, the influences of
fracture shear stiffness ks, fracture frictional coefficient μ, pore pressure Pp, and the far-field stress angle θ
are systematically investigated under the general assumption of quasi-static deformation.
4.1. Generation of Discrete Fracture Networks
Field observations indicate that fault damage zones often feature a broad range of fracture lengths. The
length distribution can be quantified by a power law (Odling et al., 2004; Ostermeijer et al., 2020) or by a
log-normal law (Rizzo et al., 2017). In this study, we adopt the double power-law (Bonnet et al., 2001; Davy
et al., 1990) to describe the fracture length density distribution:
n  l, L   LD l  a , l  lmin , lmax  ,

(22)
in which n(l, L)dl is the number of fractures whose length is in the range [l, l + dl] in a representative
volume with side length L, α is the density-related term, D is the fractal dimension, and a is the power-law
length exponent. Numerous two-dimensional outcrop datasets suggest that D falls in the range of 1.3–2 and
a varies between 1.3 and 3.5 (Bonnet et al., 2001; Renshaw, 1999). In particular, when a = D + 1, the fracture
network is self-similar and its connectivity is scale-invariant (Bour et al., 2002; Darcel et al., 2003).
Despite the complexity of the fracture networks related to fault zones and the limited field characterization,
it is acknowledged that fractures in fault damage zones are generally not randomly oriented (e.g., Ostermei-
jer et al., 2020; Wilson et al., 2003), but the controls of preferential orientations on the stress variations are
not yet well understood. In addition, the fracture length distribution within the fault damage zone is anoth-
er intractable issue. Given this, we set D = 2 and a = 3 in this study to generate six sets of scale-independent
discrete fracture networks with different fracture densities for illustrative purposes. In each representative
rock mass volume, the minimum length of fractures is set to be 0.01 L for normalization. Note also that
uniform distributions of fracture center and orientation are assumed only for convenience here, although
fracture clustering and interactions generally exist within fault zones (e.g., Ostermeijer et al., 2020). In Sec-
tion 5.1, we substantiate more arguments about this heuristic treatment. The density term α is prescribed
as 4, 3.2, 2.4, 1.6, 0.8, and 0.32, respectively, for each set to reflect linearly decreasing fracture density. Note
that a power-law or exponential decay of fracture density would be more appropriate according to field
observations (Johri et al., 2014; Savage & Brodsky, 2011), which, however, can be conveniently included by
changing the values of the density term. With these parameters, 10 realizations are generated for each layer
using the multiplicative cascade process (Darcel et al., 2003; Kim, 2007; Moein et al., 2019). In Figure 4, we
further show the fractal features of the 10 realizations of each layer, where the cumulative number N(l) of
fractures with length greater than l is well captured by the integrated form of Equation 22:
  
log N  l / L  log 
(23) 
 a 1
D  a 1 
L      a  1 log  l / L 

in which fracture length is normalized by the model size. The constant

coefficient (a−1) of the second right-hand term represents a fractal di-
mension of the length distribution.
Assuming that fractures can hardly interpenetrate under compression

(Davy et al., 2018), the normal stiffness is set to 12,000 GPa/m for all
fractures. The high normal stiffness value is several hundreds of times
larger than that of shear stiffness adopted in Section 4.2, the latter of
which is selected according to laboratory shear experiments on rock
joints (e.g., Grasselli & Egger, 2003; Yoshinaka & Yamabe, 1986). The
dilatancy factor β is set to 0.05, corresponding to a dilation angle (2.86°)
characteristic of rock fractures (Farahmand et al., 2018; Min et al., 2004).
As implied by laboratory experiments (e.g., Cheng et al., 2019; Zhao &
Cai, 2010), the dilation angle may change with normal stress and plastic
Figure 4. Accumulative numbers of fractures with length greater than
l are plotted as functions of l for all layers (based on 10 realizations). The shear strain, and finally approach a constant value following large plastic
decreasing value of α represents decreasing fracture density. The parallel shear strains. Therefore, it is reasonable to assume a constant dilatancy
dash lines indicate that the fracture networks of all layers are characteristic factor for these pre-existing fractures in this study since they are likely to
of a fractal dimension of D = 2 and a power law length exponent of a = 3. have experienced large normal stresses and large plastic strains. Initially,
The position of stiffness length is also marked (red dash dot line).
(S1 = 150 MPa, S3 = 75 MPa, Pp = 40 MPa). A buried depth of ∼4 km is

all layers are loaded instantaneously under the same stress conditions
assumed so that the vertical stress, or the intermediate principal stress

(S2), albeit not explicitly considered in the model, is about 90 MPa, corresponding to a strike-slip faulting
regime. Such a state of stress also honors that the far-field stress in the upper crust is near frictional equilib-
rium with hydrostatic pore pressure (Zoback & Townend, 2001). A purely elastic layer without any fractures
is set as reference to highlight the effect of fracture density in the following analysis.
4.2. Effect of Fracture Shear Stiffness on Stress Variations
The large normal stiffness precludes the excessive contribution of the normal fracture deformation under
compression. Thus, we first explore the effect of fracture shear stiffness (ks = 10, 20, and 30 GPa/m) on stress
variations with increasing fracture density, while keeping μ and Pp constant at 0.6 and 40 MPa, respectively.
Figure 5a summarizes the stress rotations in four series of simulations (the far-field θ = 10°, 30°, 60°, and 80°,
respectively). For all cases, S1 constantly rotates to the angle of 45° with the increase of fracture density. For the
same fracture density, the rotation of S1 is impeded by the increase of ks. As quantitatively shown in Figure 10a,
a larger value of ks limits fracture deformation and consequently stress rotation due to the constant strain in
the strike direction. In addition, the increase of fracture density generally brings about higher uncertainty as
indicated by the increasing dispersion for constant ks. For a certain fracture density, the uncertainty becomes
more significant when θ approaches 45°. This is because the stress state becomes more isotropic when it is near
the pure shear state (θ = 45°), implying that fractures of any orientations can deform preferentially.
To better illustrate the stress variations, the mean values of both principal stresses of 10 realizations, for
ks = 20 GPa/m, are represented in Figure 5b by Mohr circles. It can be seen that the Mohr circles display
distinct changes as θ changes. For the case of θ = 10°, the stress variations manifest themselves as a series
of contracting Mohr circles, moving leftwards. In other words, both the mean stress and differential stress
decrease with the increase of fracture density. As we assume frictional equilibrium at the far-field, the local
stress states in the fault zone suggest a more stable fault as we normally expect. To activate such a stable
fault, a higher pore pressure or a lower frictional strength, or a combination of both, are needed. Note that
this observation is diametrically opposite to what is predicted by the existing multilayer model (Faulkner
et al., 2006) but consistent with the model by Casey (1980). As for the other end-member (θ = 80°), Mohr
circle also contracts with fracture density but moves to higher stress levels. It indicates reduced differential
stress and increasing mean stress as the fault core is progressively approached. The scenario with a high
Figure 5. (a) Stress rotation as a function of (normalized) fracture density and fracture shear stiffness ks in four series
of simulations (θ = 10°, 30°, 60°, 80°). Datasets are based on 10 realizations of each parameter combination. (b) Mohr
diagrams represent stress variations with increasing fracture density for different values of θ. Each Mohr circle denotes
the average stress state of 10 realizations (ks = 20 GPa/m).
value of θ has been widely studied. The results here are well consistent with previous work (Faulkner
et al., 2006; Healy, 2008; Heap et al., 2010). To re-activate such unfavorably orientated faults, static weaken-
ing mechanisms invoking high pore pressure and low friction strength are required.
As shown in Figure 5a, the middle two cases show a pure shear state at high fracture density levels. Unlike
the existing multilayer model, the self-regulated process in our model does not allow S1 to rotate beyond
45°. As discussed in Section 5.2, the innermost layer in this circumstance is purely plastic. Our model is thus
applicable to combine the broadly defined brittle fault zone, as the existing multilayer model, and the plastic
gouge layer, as originally indicated by Casey's model (Casey, 1980). Of all cases, the local S3 has a minimum
when θ = 30° for the same fracture density and the local S1 has a maxi-
mum when θ = 60°. In particular, faults that are oriented at around 30°
to S1 are considered to be critical according to the Coulomb frictional fail-
ure theory together with laboratory-derived frictional coefficients (Byer-
lee, 1978). The results of the θ = 30° case show that inner layers become
more critical when such a fault is already favorably oriented with respect
to the far-field stress field. The significance of this modeling observation
is interesting, and the modeled stress variations in this case still need to
be examined by in situ stress characterizations.
4.3. Effect of Fracture Frictional Coefficient on Stress Variations
In this section, we further examine the effect of fracture frictional coef-

ficient (μ = 0.2, 0.4, and 0.6). For all cases, ks and Pp are set to 20 GPa/m
and 40 MPa, respectively. The resulting stress rotations are shown in
Figure 6. It suggests that the decrease of μ from 0.6 to 0.4 has no sig-
nificant effect on the stress rotation for the same fracture density, while
its further reduction to 0.2 causes significant rotation especially at high
fracture density levels. When θ = 30° and 60°, the influence of μ becomes
less significant. We note that, for all cases, S1 in the series of θ = 60° first
rotates to the angle of 45° (when the normalized fracture density is 0.8).
Figure 6. Stress rotations as a function of fracture density and fracture
frictional coefficient μ for θ = 10°, 30°, 60°, and 80°, respectively. Data sets This suggests that the model is prone to plasticity in this configuration.
are based on 10 realizations of each parameter combination. Therefore, a further increase in fracture density does not promote further
Figure 7. Differential stress (left) and mean stress (right) as a function of fracture density and fracture frictional
coefficient μ. Error bars show the corresponding uncertainties based on the results of 10 realizations. Horizontal dash
lines denote the reference far-field stresses.
rotation of S1. In contrast, the case of θ = 30° exhibits nearly plastic behavior only in the innermost layer.
Furthermore, marked μ effect can be observed in both end-members, which feature significant rotations
and uncertainties, when μ decreases to 0.2.
The differential stress and mean stress are further plotted versus the fracture density for different μ in
Figure 7. Similar to Figure 5b, it is clearly seen from Figures 7a–7d that the decline of differential stress is
monotonic with fracture density in all cases. That said, fracture deformation always consumes the strain
energy of the model, causing differential stress to decrease. When θ = 45°, no fracture deforms due to
stress isotropy (σyy = σxx), and the principal stresses remain unchanged with varying fracture density. When
θ < 45°, the mean stress becomes less compressive with increasing fracture density, and it is more compres-
sive when θ > 45° (Figures 7e–7h). As a result, the principal stresses would rotate to be at 45° to the fault
strike direction for all values of θ, except for 0° and 90° in which differential stress still relaxes but in the
absence of stress rotations, that is, contracting Mohr circles with one end fixed (see Zhang & Ma, 2021).
These results are in good consistency with those predicted by Casey's (1980) model but substantiated by
physics-based mechanisms.
4.4. Effect of Local Pore Pressure on Stress Variations
In addition to the foregoing two factors, the effect of different pore pressures (Pp = 40, 50, and 60 MPa) is
investigated, while ks and μ are set to 20 GPa/m and 0.6, respectively. Different from the influences of ks and
μ, the elevated pore pressure has a trivial effect on the stress variations
in our simulations, as shown in Figure 8. Since the fracture density has
a more prominent effect, the pore pressure effect is likely masked by the
former.
The negligible pore pressure effect in these simulations can be account-

ed for by the fracture constitutive behavior. As pointed out by Davy
et al. (2018), the mechanical behavior of the fracture network is main-
ly controlled by the stiffness length ls, which is defined as the fracture
length for which the fracture shear stiffness equals to the average shear
stiffness of matrixE(i.e., ks  km):
4G
ls 
(24)
 1  v  ks
which is 0.141 L for the simulations in this section. For fractures much
shorter than ls, frictional resistance is dominated by the matrixE as km is
much largerEthan ks. For fractures much longer than ls, the plastic limit
in Equation 11 can be simplified as τμ (= μ·σn) Esince ks  km. As seen
Figure 8. Stress rotations as a function of fracture density and pore from Figure 4, a small number of fractures are longer than ls in these
pressure Pp for θ = 10°, 30°, 60°, and 80°, respectively. Datasets are based simulations. Therefore, the increase of pore pressure only facilitates these
on 10 realizations of each parameter combination. few long fractures to slip frictionally while the remaining vast number of
short fractures remain unaffected. Equivalently, the increased pore pres-
sure only contributes to a subtle increase in the plastic shear deforma-
tion, even at a high fracture density level (we expand this argument in Section 4.5). A more significant pore
pressure effect is expected with a larger fractal dimension a or in a weaker model.
It is worth noting that, the fracture constitutive relationship we develop in this paper aims originally at the
brittle damage rock mass. For mature fault cores, weak and interconnected foliated microstructures, such as
phyllosilicate foliae, might be pervasive (Collettini et al., 2019). Frictional sliding along fault-parallel fabric
can be simply described by the Coulomb criterion and Equation 10, so that significant effects of elevated
pore pressure (especially overpressure) can be expected on stress rotations therein. Healy (2008) argued
that, the deviatoric effect of pore pressure could exert an important influence on stress rotations when con-
sidering microcrack-related anisotropy, which, however, is beyond the scope of this study since the fractures
are set to be randomly oriented.
4.5. Influences of Fracture Deformation on the Effective Elastic Properties
Our simulations show that the local stress state can be modified, to different extents, by the fracture net-
work properties in various scenarios. Since the variations of parameters in previous sections essentially
change the magnitudes of fracture deformation (i.e., global strain increments), we further relate the fracture
deformation to the effective elastic properties. For each layer, we calculate its effective elastic properties
following Hooke's law based on the updated global stress and strain. In Figure 9, effective elastic properties
are plotted against the fracture-contributed strain in the fault strike (y-) direction based on all foregoing
numerical tests. It can be seen that fracture deformation induces compaction in the y-direction when θ is
less than 45° while dilation emerges when θ is larger than 45°, which results in the decrease and increase of
σyy, respectively, due to the constant strain boundary as illustrated in Figure 2.
In general, the fracture deformation (and the resultant strain increment in the y-direction) increases with
fracture density, which leads to decreasing effective Young's modulus and increasing effective Poisson's ra-
tio, as corroborated by the measured elastic properties in the field (Table 1). In other words, stress variations
and modifications of the effective elastic properties accompany each other within the fault damage zone.
Both effective elastic parameters show nonlinear dependence on the fracture-related strain. The nonlinear-
ity becomes more prominent when the fracture density increases. The data clusters at low fracture densities
(i.e., normalized fracture density of 0, 0.08, and 0.2) indicate that the relative changes in both effective
Figure 9. Effective elastic properties as a function of the strain increment in the fault strike direction: (a) Effective
Young's modulus. (b) Effective Poisson's ratio. Compression is taken as positive. The effective elastic properties are
obtained from the simulations as shown in Figures 5, 6, and 8. Note that the parameter combination (ks = 20 GPa/m,
μ = 0.6, Pp = 40 MPa) is set as a benchmark, results of which are linked by red dash lines in each series. When one of
the parameters varies, the remaining two are kept constant. The normalized fracture density levels are also qualitatively
indicated.
elastic parameters are dramatic in response to the fracture deformation. When the fracture density increas-
es, the relative changes gradually diminish but incur higher uncertainty. For the same fracture density, in
addition, more fracture deformation is induced when θ deviates further away from 45° (i.e., θ = 10° and
80°), which requires more substantial stress adjustment to maintain the constant strain in the y-direction.
When θ approaches 45°, both effective elastic properties are more sensitive to fracture deformation. In
other words, a small amount of fracture deformation results in significant changes in the effective elastic
properties. Therefore, it plausibly explains why it is much easier for the middle two cases (θ = 30° and 60°)
to reach a plastic state.
Other than fracture density, we further consider the individual influences of other factors. According to the
fracture constitutive relationship, we divide the fracture deformation into elastic and plastic deformation.
Therefore, it is able to quantify the dependence of each deformation component on fracture shear stiffness,
frictional coefficient, and pore pressure. In Figure 10, the fracture-induced total strain increment in the
fault strike direction and its plastic component are plotted versus each of the influence factors in the case of
θ = 10°. As with Figure 9, the total strain increment constantly increases with fracture density. For a certain
fracture density level, the variations of each influence factor induce dif-
ferent degrees of deformation. Figure 10a illustrates the control of ks on
the elastic fracture deformation, since all fractures are in the elastic stage
(i.e., no frictional slip occurs) in the simulations. As ks increases from 10
to 30 GPa/m, the elastic fracture deformation decreases accordingly due
to higher resistance. Figure 10b shows that most of the deformation is
elastic when μ = 0.6 and 0.4. When μ decreases further to 0.2, both the to-
tal and plastic fracture strain increase considerably. It demonstrates that
the low values of μ avail many fractures to frictional sliding. The effect of
μ, as well as the uncertainty, becomes more significant for higher fracture
densities. This justifies the observations in Section 4.3 that the local stress
states of the inner layers are close to the pure shear state when μ = 0.2.
The effect of μ also explains why the increasing pore pressure only re-
sults in trivial changes in our simulations (Section 4.4), in addition to the
stiffness length effect. Since μ is set to 0.6 in those simulations, all the
fractures are subject to shear stresses well below their frictional strength.
Even when pore pressure increases up to 60 MPa, which is substantially
over-pressured and close to the minor principal stress (75 MPa), only a
small amount of plastic deformation is induced.
5. Discussion
5.1. Further on the Assumptions Associated With Fractures in
the Multilayer Model
Our simulations are performed based on the multilayer model, which

has been previously used to assess the stress variations resulting from
the changes in elastic moduli (Faulkner et al., 2006; Healy, 2008; Heap
et al., 2010). Instead of directly varying elastic properties through lay-
ers, we explicitly relate the deformation of macroscopic fractures to the
Figure 10. Total strain increment and its plastic component induced by global mechanical behavior of each layer. We argue that, both the stress
fractures in the fault strike direction in the case of θ = 10°, as functions variations and the elastic property changes result from the synergistic ef-
of fracture density and (a) fracture shear stiffness ks, (b) fracture frictional
coefficient μ, and (c) pore pressure Pp.
fect of fracture deformation and boundary constraints. Specifically, this
bottom-up approach is primarily based on the following assumptions
about fractures: (a) the frictional fractures are assumed to be pre-existing
without considering the complex fault slip and seismic history; (b) these fractures exist at different scales, if
not all scales, following power-law distribution; and (c) fractures are randomly distributed while potential
clustering and interactions are not considered.
First, the off-fault damage is believed to be produced by earthquake ruptures due to transient loading of
the stress waves emitted from ruptures. Such coseismic damage generally occurs over relatively short time
scale. In the context of long-term mechanical response, our model begins with these pre-existing fractures
without considering their complex dynamic origins. In addition, the quasi-static deformation of these frac-
tures is deemed as a reasonable approximation for the static response to long-term loading. Therefore, the
assumption of constant fracture mechanical properties (i.e., shear and normal stresses, friction), independ-
ent of fracture length, is reasonable, although they certainly result in simplifications that depart from the
real cases.
Second, fracture length distribution associated with most geological outcrops has been commonly quanti-
fied by power law (e.g., Bonnet et al., 2001), but it should also be noted that the power law is not universal
for all cases, for example, fractures in some fault damage zones (Rizzo et al., 2017). In particular, Knott
et al. (1996) pointed out that secondary fault throw within fault damage zones can also be characterized by
power law, and fault length has a linear relationship with the fault throw (Cowie et al., 1996). Consequently,
power-law distribution can be reasonably hypothesized for fracture length within fault damage zones. We
only use self-similar and scale-invariant fracture networks in this study for illustrative purposes, and the
largest fracture length is limited by the model size L. Such model configuration yields trivial pore pressure
effect due to the control of the stiffness length, as elaborated in Section 4.4. More realistic pore pressure
effect is expected by underlining frictional slips with a larger power-law length exponent by underlining
frictional slips.
Third, Mello et al. (2010) pointed out that the maximum strain rate direction rotates when an earthquake
rupture propagates at the sub-Rayleigh wave speed, from fault-parallel to fault-perpendicular direction.
In this sense, this dynamic fracturing pattern could support the random distribution of fracture orienta-
tion within the damage zone. We are also cognizant that complex earthquake cycles can generally produce
fractures with preferential orientations, for example, parallel fractures inclined at a low angle to the fault
plane (Wilson et al., 2003). If such a fracture pattern is included, our model is still applicable but will likely
show significant anisotropy of mechanical properties, which warrants further investigation. Furthermore,
fracture clustering and interactions may further promote stress rotations and elastic moduli weakening,
since more strain energy stored in the model will be dissipated so, leading to a more isotropic state of stress.
Hence, simulations in the absence of fracture clustering and interactions in this study can be considered as
the most conservative case.
5.2. Stress Rotation Limit Imposed by the Multilayer Model and Field Observations
In the existing multilayer model (Faulkner et al., 2006; Healy, 2008; Heap et al., 2010), S1 is able to rotate
past the angle of 45°. In such cases, the mechanical equilibrium gives the solution to the final  yy of each
layer as:
 Eeff v
 yy
(25) 2
 yy  eff  xx
1  veff 1  veff
where Eeff and veff are effective Young's modulus and effective Poisson's ratio, respectively, of a specific layer,
and the parameters εyy and σxx are prescribed as those of purely elastic layer, representing the far-field con-
ditions. In the following, this equation and its application to stress variations are re-examined considering
the elastic moduli changes.
In the case of θ > 45°, σxx is larger than σyy but the fixed strain εyy can be either contractional (positive) or
extensional (negative) according to the far-field stresses and elastic moduli. In the existing model, only the
effect of microcrack damage is considered, and the outermost layer usually has a relatively small differen-
tial stress and high elastic moduli (derived from intact laboratory samples). Hence, the fixed strain εyy is
mostly contractional (positive). As revealed by the cyclic loading experiments on different rock types (Heap
et al., 2010), the elastic moduli evolution with increasing microcrack damage manifests a reduction of Eeff
by 11%–32% and a drastic increase in veff by 72%–600%. Applying such measurement results, the coefficients
of the two right-hand terms in Equation 25 both increase with veff regardless of the decreasing Eeff. This
plausibly leads to a value of  yy larger than σxx, especially when veff is close to 0.5, that is, veff/(1−veff ) ≈ 1.
Consequently, the stress can rotate over the angle of 45°, where the differential stress reaches the minimum
and starts to increase. However, we argue that the increase of differential stress is incompatible with the
minimum energy principle under this circumstance.
The analysis also holds for the case with θ < 45°, which has not been extensively studied previously. In this
case, σyy is always larger than σxx and the fixed strain εyy is thus contractional (positive). With the laborato-
ry-derived elastic moduli changes,  yy could further increase, contributing to an increase of mean stress and
a decreasing angle θ (Faulkner et al., 2006). Also note that, since the second right-hand term monotonically
increases with veff,  yy could decrease to some extent if the first right-hand term further decreases. This
trend was originally presented by Casey (1980), in which a lower value of Young's modulus in the weak
zone is set while maintaining constant Poisson's ratio constant. More generally, a substantial decrease of
Young's modulus can be expected in this study when considering macroscopic fractures. In particular,  yy
can decrease even to the value of σxx when veff ≈ 0.5 and Eeff ≈ 0 GPa, corresponding to the limit of θ = 45°.
Due to the strain boundary in the model, it is reasonable to expect that the effective elastic properties and
their changes through layers could be significantly different from those computed under stress boundary
conditions. For example, a fractured rock mass could feature a bulk Poisson's ratio well above 0.5 under
stress boundary conditions (Min & Jing, 2004), which is not the case in the multilayer model. In other
Figure 11. Rotations of the maximum horizontal principal stress SHmax near the San Andreas Fault system (SAF) and the S3 Fault zone at the Grimsel Test
Site (GTS). Vertical error bars are inferred from focal mechanism inversions (Hardebeck & Hauksson, 1999; Provost & Houston, 2001). These gray and black
bars delineate the 95% and 80% confidence limit on the inverted stress orientations, respectively. The shaded area indicates more elaborate constraints on
the regional stresses at a scale of 10 km near the SAF (Townend & Zoback, 2004). Blue triangles show that SHmax rotates away from the fault plane at the GTS
(Krietsch et al., 2019). The inset illustrations correspond to the two cases of stress rotations presented.
words, a direct application of laboratory-measured elastic moduli, measured under conventional stress
boundary conditions, to the multilayer model should be cautious and warrant further study. In addition, the
predicted limiting angle of 45° is equivalent to that in previous studies on fault gouge by assuming Coulomb
plasticity (Byerlee & Savage, 1992; Lockner & Byerlee, 1993) or material softening (Rice, 1992). The plastic
layer is stressed under constant compression σxx, and a small shear stress τxy suffices to induce plastic flow.
Assuming that the plastic flow is volume-preserving, the continuous loading leads to extrusion in the fault
strike direction, which renders the constant strain boundary condition inapplicable in this limiting case.
In Figure 11, the stress rotations associated with the SAF and the fault zone at the GTS are compared, de-
the SAF system, focal mechanism inversions indicate that S1 rotates from ∼90° at far-field to ∼45° inside the
spite the fact that the tectonic fault SAF is much larger in scale and more complicated than the latter. For
fault zone (Hardebeck & Hauksson, 1999; Provost & Houston, 2001). Since the earthquakes far away from
the fault zone are relatively scarce, the far-field stress orientations inverted by focal mechanisms are subject
to high uncertainties, as indicated by some angles larger than 90° and by the large error bars in Figure 11.
Therefore, we further plot the results of the integrated stress measurements (Townend & Zoback, 2004),
which also show high angles (65°–85°) between the far-field S1 and the fault strike. By contrast, the inver-
sion results near the fault zone are of less uncertainties due to the concentration of earthquakes on the fault
plane. In particular, S1 close to the creeping section is found to be oriented at 43°–56° to the fault strike
(Provost & Houston, 2001). Considering the inversion uncertainties, this may point to a plastic layer within
which a limiting angle of 45° exists for S1. To our knowledge, there is no field evidence showing a lower
angle than 45° in the case. Based on experimental observations, Lockner et al. (2011) pointed out that the
plasticity of the fault gouge recovered from the SAF creeping section is associated with the abundance of
compliant minerals and extremely low frictional strength. The correlation between the plastic fault gouge
and the limiting angle of 45° is well captured by our model. In addition, the measured stress variations
around the ductile fault zone at the GTS also provide independent field evidence to justify this correlation,
as shown in Figure 11.
5.3. Spatial-Temporal Variations of the In Situ Stress Within Fault Zones
Our model is capable to predict continuous variations of local stresses and effective elastic properties with-
in the damage zones of a mature fault. In particular, this model presents the off-fault variations through
varying fracture densities within one side of a fault damage zone. As suggested by the field observations,
fracture density within fault damage zones generally decreases with the distance away from the fault core,
for example, following an exponential (Chester & Logan, 1986; Vermilye & Scholz, 1998; Wilson et al., 2003)
or a power law (Johri et al., 2014) distribution. In other words, fractures are typically localized in the vicinity
of the fault core. As revealed by our model, drastic changes in local stresses and effective elastic properties
are restricted to the regions near the fault core. In other words, the weakness of a poorly oriented fault,
tending to be active, mainly lies in the core zone and the adjacent limited damage zones. Considering the
subsidiary faults that are commonly observed in the field (e.g., Ostermeijer et al., 2020), the assumption of
monotonic changes in fracture density in our model may be an oversimplification, let alone the asymmetry
on both sides of a fault. Nevertheless, the relationship between stress variations and fracture density (e.g.,
high fracture density promotes stress rotation) revealed by this study may still hold.
We also acknowledge the complexities of natural faults, such as the asymmetry between both sides of a fault
(Rempe et al., 2018; Rice et al., 2005) and the along-strike heterogeneities (e.g., Perrin et al., 2016). These
spatial heterogeneities are closely related to the fault growth process and rupture dynamics that are difficult
to quantify. But again, our model can be extended along the strike direction so that along-strike variations/
heterogeneities can be explored, given that the fracture patterns are explicitly specified. Furthermore, the
spatial variations in our model are purely caused by the layer-to-layer contrasts within the fault structure,
which can account for static weakening mechanisms for the unfavorably oriented but active faults. As re-
cently revealed by Fialko and Jin (2021), by contrast, fault rotation as well as far-field stress rotation due
to the long-term tectonic loading can also explain the activation of unfavorably oriented faults but in a
dynamic manner.
In essence, stress variations within the fault zone rock mass are controlled jointly by fracture deformation and
specific boundary conditions. The fracture constitutive behavior in this study is independent of time, which
ignores the significant evolution of fracture deformation with time (Brantut et al., 2013; Scholz, 1968). If
time-dependent fracture behavior is considered, it is expected to further promote stress rotation and weaken
the rock mass frictional strength in a temporal sense (Zhang & Ma, 2021). In addition, the stress variations
and effective elastic property changes in our model can be more pronounced, when further energy dissipa-
tion mechanisms are introduced to both the fractures and the matrix. For example, the chemically active
fluids in fault zones are able to weaken stiff mineral phase constituents (Collettini et al., 2019; Schleicher
et al., 2010; Wibberley, 1999). Such fluid-assisted reaction-softening could, on one hand, weaken the pro-
toliths and fractures, aiding fracture deformation and contributing to time-dependent stress rotations over
longer time scales. On the other hand, some other fluid-assisted mechanisms (e.g., mineral deposition and
cementation) can lead to healing and strengthening of fractures (Chester & Logan, 1986). The transition of
mineral phases could affect fracture frictional strength and rock mass frictional stability, as implied by labo-
ratory friction experiments (Fang et al., 2018; Zhang et al., 2019). In addition, fluid-assisted fracture healing
can generate low-permeability barriers (Chester et al., 1993; Faulkner et al., 2006). Hence the trapped fluids
can result in sustained overpressure, resulting the weakening of the fault rock mass in a time-dependent
manner until the next earthquake takes place (Sleep & Blanpied, 1992).
6. Conclusions
In this study, we numerically investigate the changes in stresses and effective elastic properties within the
fault zone rock mass by explicitly considering the deformation of fractures in a multilayer model. Due to the
synergistic effects of the specific boundary conditions and fracture deformation, it is found that the stress
variations, as well as effective elastic property changes, within the fault zone rock mass are the results of a
self-regulated process. We illustrate such stress variations in a strike-slip fault setting.
Frictional fractures are considered under plane strain condition, featuring elastic and plastic shear defor-
mation and shear-induced dilatancy in the normal direction. Based on the unified fracture constitutive
relationship, a parametric study is systematically conducted to understand the effects of fracture properties
(density, shear stiffness, friction) and pore pressure on stress variations. It is found that the increase of frac-
ture density and pore pressure and the decrease of fracture shear stiffness and the frictional coefficient can
promote the stress variations to different extents. On the one hand, the damage-induced stress variations
through layers manifest themselves as principal stress rotations and magnitude changes. The simulation
results reveal that the major principal stress tends to rotate toward the angle of 45° as the fault core is
approached, accompanied by diminishing differential stress. However, the change of mean stress depends
closely on the fault orientation. The mean stress increases with the damage degree when the far-field major
principal stress is oriented at an angle higher than 45° with respect to the fault plane, and on the contrary,
decreases when the angle is less than 45°. On the other hand, we also show that the stress variations and
effective elastic property changes accompany each other. As the damage intensifies, the effective Young's
modulus continues to decrease while the effective Poisson's ratio increases monotonically. The degrees of
the stress variations and changes in effective elastic properties are essentially proportional to the fracture
deformation. It is demonstrated that fracture properties, pore pressure and fracture density are able to mod-
ulate the elastic and plastic components of the fracture deformation to various extents.
It is also worth emphasizing that we correlate a limiting condition with the plastic rheology upon which
the pure shear state emerges. A limiting angle of θ = 45° is attained for the major principal stress rotation,
when the effective Young's modulus and effective Poisson's ratio reach their limiting values, that is, 0 and
0.5, respectively. In other words, our model is able to capture continuous variations of local stresses and ef-
fective elastic properties within the fault zone rock mass. We note that this model is not limited to strike-slip
faults but can be applicable to dip-slip faults, given that the fault plane parallels the regional intermediate
principal stress. This study has important implications for inferring the state of stress and effective elastic
moduli within the fault zone rock mass, especially in the absence of direct stress measurements.
Appendix A: Relation of Local Fracture Deformation and Global Strain

Instead of directly varying elastic properties through layers, we initially assign the resolved stress tensor σ
of the outermost layer to all layers of the multilayer model. For one inner layer with certain fracture densi-
ty, the effective normal and shear stresses (σn and τ) acting on a fracture with normal unit vector n can be
related to the stress tensor σ by:
 n nT    n  PP
(A1)

(A2)
 nT    s
in which Pp is the pore pressure and the in-plane shear unit vector s defined in Figure 1a is adopted when
the relative shear motion is right lateral, otherwise the opposite direction is used.
Based on the stress-displacement constitutive relationship, we can calculate the displacement vector u of
the fracture, which is further quantitatively related to the global strain of the rock mass volume representa-
tive of the layer according to effective medium theory (Kachanov, 1992):



(A3)
A
u  n  n  u
with A the cross-section area of the rock mass volume and δ the fracture half-length. Considering all frac-
tures in the rock mass volume, the total global strain can be obtained by summing the elastic matrix strain
and the additional strain contributed by fractures:

 tot   m   i i  ui  ni  ni  ui 
(A4) A
where the subscript i of the second right-hand term denotes the ith fracture under consideration, and εm is
the elastic matrix strain given by Hooke's law under plane strain:
 xx v / 1  v 
m
 1 0   
  1 v   xx 
 m

(A5)   m
yy     v / 1  v  1 0   yy 
 m 2G 
 xy   0 0 1/    xy 
1  v
with the superscript m representing the purely elastic (intact) matrix.
As indicated by Equations 9 and 10, the shear displacement of any fracture is a function of the average
matrix resistance and thus of the effective properties of the model (see Equation 4). Generally, there are two
theoretical methods to calculate the effective properties, depending on whether the fracture interactions
are considered or not (Davy et al., 2018; Kachanov, 1992). One method is the Non-Interaction Approxima-
tion (NIA), which states that each fracture can be regarded as isolated and that its surrounding medium is
maintained intact as the initial elastic matrix. The NIA allows a simple summation to calculate the effective
properties while keeping high accuracy. Grechka and Kachanov (2006) showed that the NIA still holds up
even at high fracture densities, as long as fracture locations are random. The second method, for example,
Self-Consistent Scheme (O'Connell & Budiansky, 1974) and Differential Scheme (Hashin, 1988), suggests
that the impact of fracture interactions can be approximated by reduced effective properties of the sur-
rounding medium. For the ith fracture, therefore, the average matrix resistance can be calculated as:
2Gi 1
km ,i 
(A6)
 i 1 1  vi 1 
in which Gi− and vi−1 are the effective shear modulus and effective Poisson's ratio, respectively, of the rock
mass volume resulting from the previous (i−1) fractures. In Section 4, randomly distributed fractures are
generated within the elastic matrix. In order to investigate the influences of fracture properties on the ef-
fective properties and stress variations, we use the NIA for simplicity to quantify the fracture contribution.
Appendix B: Bisection Method for Stress Updates

The bisection method is utilized to iteratively determine the stress change Δσyy, before which a continuous
function with variable Δσyy is established according to the constant strain boundary:
m
f Δ yy, j   yy
(B1)
m frac
,0   yy , j   yy, j    
m m m
where
E  yy ,0   yy is the fixed reference strain in the y-direction with “0” being the initial input.
E  yyE, j and  yyfrac
,j
in the parenthesis represent the strain components in the y-direction of the elastic matrix and contributed
by fractures, respectively, at the jth iterative step. With the trial Δσyy,j and a trial stress tensor,
 xx  xy 
σi 
(B2) 

 yx  yy  Δ yy , j 
  
m
E  yy , j is calculated by Equation A5. Note that, the updated normal stress difference, that is, differential stress
|σxx − σyy|, invariably decreases. Therefore, the displacements of all fractures should be re-calculated under
the trial stress tensor according to the unloading constitutive law in Section 2.2. Based on Equation A4, the
new elastic matrix strain and new fracture displacements are further combined to update the global strain
of each layer. This iteration will continue until Equation B1 reaches zero or a sufficiently small value.
More specifically, Figure B1 shows the numerical algorithm searching for the root of f (Δσyy) = 0. At the be-
ginning of the iteration, it is necessary to define an interval [m, n] allowing f (m)·f (n) < 0. In the context of
stress adjustment, we simply let m = 0, which is one end-member case with no constraint of strain bound-
ary, and let n = |σxx - σyy|, which is the other end-member case where the representative volume is under an
isotropic state of stress. Then each iteration executes these steps:
Figure B1. Schematic diagram of the bisection method used to solve the stress change Δσyy due to the fracture-induced
strain increments
E ( yyfrac
, j ) and the global strain boundary in the slip direction.
Step-1: Calculate the midpoint of the interval, Δσyy,j = (m + n)/2;

Step-2: Calculate the function value at the midpoint, f (Δσyy,j);
Step-3: If |Δσyy,j – m| is sufficiently small or f (Δσyy,j) = = 0, return Δσyy,j and stop iterating;
Step-4: If Step-3 is not satisfied, examine the sign of f (Δσyy,j). Replace either m or n with Δσyy,j in order to
ensure the root is within the new interval. Then Set j = j + 1 and return to Step-1.
Data Availability Statement

This is a theoretical study and contains no collected data. The MATLAB scripts used to produce the results
are publicly available at https://doi.org/10.17632/g8byrmhp4s.1.
Acknowledgments References
This study is supported by the Swiss
National Science Foundation (Grant no. Bauer, J. F., Meier, S., & Philipp, S. L. (2015). Architecture, fracture system, mechanical properties and permeability structure of a fault
182150). The authors thank Kai Bröker zone in Lower Triassic sandstone, Upper Rhine Graben. Tectonophysics, 647, 132–145. https://doi.org/10.1016/j.tecto.2015.02.014
for his comments on an earlier version Bonnet, E., Bour, O., Odling, N. E., Davy, P., Main, I., Cowie, P., & Berkowitz, B. (2001). Scaling of fracture systems in geological media.
of the manuscript. The authors are Reviews of Geophysics, 39(3), 347–383. https://doi.org/10.1029/1999RG000074
grateful to the Editor (Dr. Isabelle Man- Bour, O., Davy, P., Darcel, C., & Odling, N. (2002). A statistical scaling model for fracture network geometry, with validation on a
ighetti) and both reviewers (Dr. David multiscale mapping of a joint network (Hornelen Basin, Norway). Journal of Geophysical Research, 107(B6). ETG-4. https://doi.
Healy and Dr. Mateo Acosta) for their org/10.1029/2001JB000176
insightful comments and constructive Brantut, N., Heap, M. J., Meredith, P. G., & Baud, P. (2013). Time-dependent cracking and brittle creep in crustal rocks: A review. Journal
suggestions to improve this manuscript. of Structural Geology, 52, 17–43. https://doi.org/10.1016/j.jsg.2013.03.007
Open access funding enabled and Byerlee, J. (1978). Friction of rocks. In Rock friction and earthquake prediction (pp. 615–626). Birkhäuser. https://doi.
organized by ETH Zürich. org/10.1007/978-3-0348-7182-2_4
Byerlee, J. D., & Savage, J. C. (1992). Coulomb plasticity within the fault zone. Geophysical Research Letters, 19(23), 2341–2344. https://doi.
org/10.1029/92GL02370
Caine, J. S., Evans, J. P., & Forster, C. B. (1996). Fault zone architecture and permeability structure. Geology, 24(11), 1025–1028. https://doi.
org/10.1130/0091-7613(1996)024<1025:FZAAPS>2.3.CO;2
Cappa, F., Guglielmi, Y., & Virieux, J. (2007). Stress and fluid transfer in a fault zone due to overpressures in the seismogenic crust. Geo-
physical Research Letters, 34(5). https://doi.org/10.1029/2006GL028980
Carpenter, B. M., Kitajima, H., Sutherland, R., Townend, J., Toy, V. G., & Saffer, D. M. (2014). Hydraulic and acoustic properties of the active
Alpine Fault, New Zealand: Laboratory measurements on DFDP-1 drill core. Earth and Planetary Science Letters, 390, 45–51. https://
doi.org/10.1016/j.epsl.2013.12.023
Casey, M. (1980). Mechanics of shear zones in isotropic dilatant materials. Journal of Structural Geology, 2(1–2), 143–147. https://doi.
org/10.1016/0191-8141(80)90044-9
Cheng, C., Li, X., Xu, N., & Zheng, B. (2019). Direct shear experimental study on the mobilized dilation behavior of granite in ALXA can-
didate area for high-level radioactive waste disposal. Energies, 13(1). https://doi.org/10.3390/en13010122
Chester, F. M., Evans, J. P., & Biegel, R. L. (1993). Internal structure and weakening mechanisms of the San Andreas fault. Journal of Geo-
physical Research, 98(B1), 771–786. https://doi.org/10.1029/92JB01866
Chester, F. M., & Logan, J. M. (1986). Implications for mechanical properties of brittle faults from observations of the Punchbowl fault
zone, California. Pure and Applied Geophysics, 124(1), 79–106. https://doi.org/10.1007/BF00875720
Collettini, C., Tesei, T., Scuderi, M. M., Carpenter, B. M., & Viti, C. (2019). Beyond Byerlee friction, weak faults and implications for slip
behavior. Earth and Planetary Science Letters, 519, 245–263. https://doi.org/10.1016/j.epsl.2019.05.011
Cornet, F. H., Bérard, T., & Bourouis, S. (2007). How close to failure is a granite rock mass at a 5 km depth? International Journal of Rock
Mechanics and Mining Sciences, 44(1), 47–66. https://doi.org/10.1016/j.ijrmms.2006.04.008
Cowie, P. A., Knipe, R. J., & Main, I. G. (1996). Scaling laws for fault and fracture populations: Analyses and applications Introduction.
Journal of Structural Geology, 18(1996), R5–R11. https://doi.org/10.1016/s0191-8141(96)80038-1
Darcel, C., Bour, O., Davy, P., & De Dreuzy, J. R. (2003). Connectivity properties of two-dimensional fracture networks with stochastic
fractal correlation. Water Resources Research, 39(10). https://doi.org/10.1029/2002WR001628
David, E. C., Brantut, N., Schubnel, A., & Zimmerman, R. W. (2012). Sliding crack model for nonlinearity and hysteresis in the uni-
axial stress–strain curve of rock. International Journal of Rock Mechanics and Mining Sciences, 52, 9–17. https://doi.org/10.1016/j.
ijrmms.2012.02.001
Davy, P., Darcel, C., Le Goc, R., & Mas Ivars, D. (2018). Elastic properties of fractured rock masses with frictional properties and power
law fracture size distributions. Journal of Geophysical Research: Solid Earth, 123(8), 6521–6539. https://doi.org/10.1029/2017JB015329
Davy, P., Sornette, A., & Sornette, D. (1990). Some consequences of a proposed fractal nature of continental faulting. Nature, 348(6296),
56–58. https://doi.org/10.1038/348056a0
Fang, Y., Elsworth, D., Wang, C., & Jia, Y. (2018). Mineralogical controls on frictional strength, stability, and shear per-meability evolution
of fractures. Journal of Geophysical Research: Solid Earth, 123, 3549–3563. https://doi.org/10.1029/2017jb015338
Farahmand, K., Vazaios, I., Diederichs, M. S., & Vlachopoulos, N. (2018). Investigating the scale-dependency of the geometrical and me-
chanical properties of a moderately jointed rock using a synthetic rock mass (SRM) approach. Computers and Geotechnics, 95, 162–179.
https://doi.org/10.1016/j.compgeo.2017.10.002
Faulkner, D. R., Jackson, C. A. L., Lunn, R. J., Schlische, R. W., Shipton, Z. K., Wibberley, C. A. J., & Withjack, M. O. (2010). A review of
recent developments concerning the structure, mechanics and fluid flow properties of fault zones. Journal of Structural Geology, 32(11),
1557–1575. https://doi.org/10.1016/j.jsg.2010.06.009
Faulkner, D. R., Lewis, A. C., & Rutter, E. H. (2003). On the internal structure and mechanics of large strike-slip fault zones: Field observa-
tions of the Carboneras fault in southeastern Spain. Tectonophysics, 367(3–4), 235–251. https://doi.org/10.1016/S0040-1951(03)00134-3
Faulkner, D. R., Mitchell, T. M., Healy, D., & Heap, M. J. (2006). Slip on'weak'faults by the rotation of regional stress in the fracture damage
zone. Nature, 444(7121), 922–925. https://doi.org/10.1038/nature05353
Fialko, Y., & Jin, Z. (2021). Simple shear origin of the cross-faults ruptured in the 2019 Ridgecrest earthquake sequence. Nature Geoscience,
14, 513–518. https://doi.org/10.1038/s41561-021-00758-5
Funato, A., & Ito, T. (2017). A new method of diametrical core deformation analysis for in-situ stress measurements. International Journal
of Rock Mechanics and Mining Sciences, 91, 112–118. https://doi.org/10.1016/j.ijrmms.2016.11.002
Grasselli, G., & Egger, P. (2003). Constitutive law for the shear strength of rock joints based on three-dimensional surface parameters.
International Journal of Rock Mechanics and Mining Sciences, 40(1), 25–40. https://doi.org/10.1016/S1365-1609(02)00101-6
Grechka, V., & Kachanov, M. (2006). Effective elasticity of fractured rocks: A snapshot of the work in progress. Geophysics, 71(6), W45–
W58. https://doi.org/10.1190/1.2360212
Haimson, B. C., & Cornet, F. H. (2003). ISRM suggested methods for rock stress estimation—Part 3: Hydraulic fracturing (HF) and/or
hydraulic testing of pre-existing fractures (HTPF). International Journal of Rock Mechanics and Mining Sciences, 40(7–8), 1011–1020.
https://doi.org/10.1016/j.ijrmms.2003.08.002
Hardebeck, J. L., & Hauksson, E. (1999). Role of fluids in faulting inferred from stress field signatures. Science, 285(5425), 236–239. https://
doi.org/10.1126/science.285.5425.236
Hashin, Z. (1988). The differential scheme and its application to cracked materials. Journal of the Mechanics and Physics of Solids, 36(6),
719–734. https://doi.org/10.1016/0022-5096(88)90005-1
Healy, D. (2008). Damage patterns, stress rotations and pore fluid pressures in strike-slip fault zones. Journal of Geophysical Research,
113(B12). https://doi.org/10.1029/2008JB005655
Heap, M. J., Faulkner, D. R., Meredith, P. G., & Vinciguerra, S. (2010). Elastic moduli evolution and accompanying stress changes with
increasing crack damage: Implications for stress changes around fault zones and volcanoes during deformation. Geophysical Journal
International, 183(1), 225–236. https://doi.org/10.1111/j.1365-246X.2010.04726.x
Heesakkers, V., Murphy, S., Lockner, D. A., & Reches, Z. (2011). Earthquake rupture at focal depth, Part II: Mechanics of the 2004 M2.2
earthquake along the pretorius fault, TauTona Mine, South Africa. Pure and Applied Geophysics, 168(12), 2427–2449. https://doi.
org/10.1007/s00024-011-0355-6
Isaacs, A. J., Evans, J. P., Kolesar, P. T., & Nohara, T. (2008). Composition, microstructures, and petrophysics of the Mozumi fault, Japan:
In situ analyses of fault zone properties and structure in sedimentary rocks from shallow crustal levels. Journal of Geophysical Research,
113(B12). https://doi.org/10.1029/2007JB005314
Jeanne, P., Guglielmi, Y., Lamarche, J., Cappa, F., & Marié, L. (2012). Architectural characteristics and petrophysical properties evolution
of a strike-slip fault zone in a fractured porous carbonate reservoir. Journal of Structural Geology, 44, 93–109. https://doi.org/10.1016/j.
jsg.2012.08.016
Jeanne, P., Guglielmi, Y., Rutqvist, J., Nussbaum, C., & Birkholzer, J. (2017). Field characterization of elastic properties across a fault zone
reactivated by fluid injection. Journal of Geophysical Research: Solid Earth, 122(8), 6583–6598. https://doi.org/10.1002/2017JB014384
Johri, M., Zoback, M. D., & Hennings, P. (2014). A scaling law to characterize fault-damage zones at reservoir depths. AAPG Bulletin,
98(10), 2057–2079. https://doi.org/10.1306/05061413173
Kachanov, M. (1992). Effective elastic properties of cracked solids: Critical review of some basic concepts. Applied Mechanics Reviews,
45(8), 304–335. https://doi.org/10.1115/1.3119761
Kim, T. H. (2007). Fracture characterization and estimation of fracture porosity of naturally fractured reservoirs with no matrix porosity using
stochastic fractal models. Texas A&M University.
Knott, S. D., Beach, A., Brockbank, P. J., Brown, J. L., McCallum, J. E., & Welbon, A. I. (1996). Spatial and mechanical controls on normal
fault populations. Journal of Structural Geology, 18(2–3), 359–372. https://doi.org/10.1016/S0191-8141(96)80056-3
Krietsch, H., Gischig, V., Evans, K., Doetsch, J., Dutler, N. O., Valley, B., & Amann, F. (2019). Stress measurements for an in situ stimulation
experiment in crystalline rock: Integration of induced seismicity, stress relief and hydraulic methods. Rock Mechanics and Rock Engi-
neering, 52(2), 517–542. https://doi.org/10.1007/s00603-018-1597-8
Leclère, H., Cappa, F., Faulkner, D., Fabbri, O., Armitage, P., & Blake, O. (2015). Development and maintenance of fluid overpressures in
crustal fault zones by elastic compaction and implications for earthquake swarms. Journal of Geophysical Research: Solid Earth, 120(6),
4450–4473. https://doi.org/10.1002/2014JB011759
Lin, W., Yeh, E. C., Hung, J. H., Haimson, B., & Hirono, T. (2010). Localized rotation of principal stress around faults and fractures de-
termined from borehole breakouts in hole B of the Taiwan Chelungpu-fault Drilling Project (TCDP). Tectonophysics, 482(1–4), 82–91.
https://doi.org/10.1016/j.tecto.2009.06.020
Lockner, D. A., & Byerlee, J. D. (1993). How geometrical constraints contribute to the weakness of mature faults. Nature, 363(6426),
250–252. https://doi.org/10.1038/363250a0
Lockner, D. A., Morrow, C., Moore, D., & Hickman, S. (2011). Low strength of deep San Andreas Fault gouge from SAFOD core. Nature,
472(7341), 82–85. https://doi.org/10.1038/nature09927
Lockner, D. A., Tanaka, H., Ito, H., Ikeda, R., Omura, K., & Naka, H. (2009). Geometry of the Nojima Fault at Nojima-Hirabayashi, Ja-
pan—I. A simple damage structure inferred from borehole core permeability. Pure and Applied Geophysics, 166(10), 1649–1667. https://
doi.org/10.1007/s00024-009-0515-0
Ma, X., & Zoback, M. D. (2017). Lithology-controlled stress variations and pad-scale faults: A case study of hydraulic fracturing in the
Woodford Shale, OklahomaWoodford Shale case study. Geophysics, 82(6), ID35–ID44. https://doi.org/10.1190/geo2017-0044.1
Mello, M., Bhat, H. S., Rosakis, A. J., & Kanamori, H. (2010). Identifying the unique ground motion signatures of supershear earthquakes:
Theory and experiments. Tectonophysics, 493(3–4), 297–326. https://doi.org/10.1016/j.tecto.2010.07.003
Michael, A. J. (1984). Determination of stress from slip data: Faults and folds. Journal of Geophysical Research, 89(B13), 11517–11526.
https://doi.org/10.1029/JB089iB13p11517
Min, K. B., & Jing, L. (2004). Stress-dependent mechanical properties and bounds of Poisson’s ratio for fractured rock masses investigated
by a DFN-DEM technique. International Journal of Rock Mechanics and Mining Sciences, 3(41), 390–395. https://doi.org/10.1016/j.
ijrmms.2004.03.072
Min, K. B., Rutqvist, J., Tsang, C. F., & Jing, L. (2004). Stress-dependent permeability of fractured rock masses: A numerical study. Interna-
tional Journal of Rock Mechanics and Mining Sciences, 41(7), 1191–1210. https://doi.org/10.1016/j.ijrmms.2004.05.005
Mitchell, T. M., & Faulkner, D. R. (2009). The nature and origin of off-fault damage surrounding strike-slip fault zones with a wide range
of displacements: A field study from the Atacama fault system, northern Chile. Journal of Structural Geology, 31(8), 802–816. https://
doi.org/10.1016/j.jsg.2009.05.002
Moein, M. J. A., Valley, B., & Evans, K. F. (2019). Scaling of fracture patterns in three deep boreholes and implications for constraining fractal
discrete fracture network models. Rock Mechanics and Rock Engineering, 52(6), 1723–1743. https://doi.org/10.1007/s00603-019-1739-7
Obara, Y., & Sugawara, K. (2003). Updating the use of the CCBO cell in Japan: Overcoring case studies. International Journal of Rock Me-
chanics and Mining Sciences, 40(7–8), 1189–1203. https://doi.org/10.1016/j.ijrmms.2003.07.007
O'Connell, R. J., & Budiansky, B. (1974). Seismic velocities in dry and saturated cracked solids. Journal of Geophysical Research, 79(35),
5412–5426. https://doi.org/10.1029/JB079i035p05412
Odling, N. E., Harris, S. D., & Knipe, R. J. (2004). Permeability scaling properties of fault damage zones in siliclastic rocks. Journal of
Structural Geology, 26(9), 1727–1747. https://doi.org/10.1016/j.jsg.2004.02.005
Ostermeijer, G. A., Mitchell, T. M., Aben, F. M., Dorsey, M. T., Browning, J., Rockwell, T. K., et al. (2020). Damage zone heterogeneity on
seismogenic faults in crystalline rock; a field study of the Borrego Fault, Baja California. Journal of Structural Geology, 137, 104016.
https://doi.org/10.1016/j.jsg.2020.104016
Perrin, C., Manighetti, I., Ampuero, J. P., Cappa, F., & Gaudemer, Y. (2016). Location of largest earthquake slip and fast rupture controlled
by along-strike change in fault structural maturity due to fault growth. Journal of Geophysical Research: Solid Earth, 121(5), 3666–3685.
https://doi.org/10.1002/2015jb012671
Pierdominici, S., & Heidbach, O. (2012). Stress field of Italy—Mean stress orientation at different depths and wave-length of the stress
pattern. Tectonophysics, 532, 301–311. https://doi.org/10.1016/j.tecto.2012.02.018
Pollard, D. D., & Segall, P. (1987). Theoretical displacements and stresses near fractures in rock: With applications to faults, joints, veins,
dikes, and solution surfaces. In Fracture mechanics of rock (pp. 277–349). https://doi.org/10.1016/b978-0-12-066266-1.50013-2
Provost, A. S., & Houston, H. (2001). Orientation of the stress field surrounding the creeping section of the San Andreas Fault: Evidence for
a narrow mechanically weak fault zone. Journal of Geophysical Research, 106(B6), 11373–11386. https://doi.org/10.1029/2001JB900007
Rempe, M., Mitchell, T. M., Renner, J., Smith, S. A., Bistacchi, A., & Di Toro, G. (2018). The relationship between microfracture damage
and the physical properties of fault-related rocks: The Gole Larghe Fault Zone, Italian Southern Alps. Journal of Geophysical Research:
Solid Earth, 123(9), 7661–7687. https://doi.org/10.1029/2018JB015900
Renshaw, C. E. (1999). Connectivity of joint networks with power law length distributions. Water Resources Research, 35(9), 2661–2670.
https://doi.org/10.1029/1999WR900170
Rice, J. R. (1992). Fault stress states, pore pressure distributions, and the weakness of the San Andreas fault. International Geophysics, 51,
475–503. https://doi.org/10.1016/s0074-6142(08)62835-1
Rice, J. R., Sammis, C. G., & Parsons, R. (2005). Off-fault secondary failure induced by a dynamic slip pulse. Bulletin of the Seismological
Society of America, 95(1), 109–134. https://doi.org/10.1785/0120030166
Rizzo, R. E., Healy, D., & De Siena, L. (2017). Benefits of maximum likelihood estimators for fracture attribute analysis: Implications for
permeability and up-scaling. Journal of Structural Geology, 95, 17–31. https://doi.org/10.1016/j.jsg.2016.12.005
Savage, H. M., & Brodsky, E. E. (2011). Collateral damage: Evolution with displacement of fracture distribution and secondary fault strands
in fault damage zones. Journal of Geophysical Research, 116(B3). https://doi.org/10.1029/2010JB007665
Schleicher, A. M., van der Pluijm, B. A., & Warr, L. N. (2010). Nanocoatings of clay and creep of the San Andreas fault at Parkfield, Califor-
nia. Geology, 38(7), 667–670. https://doi.org/10.1130/G31091.1
Schmalzle, G., Dixon, T., Malservisi, R., & Govers, R. (2006). Strain accumulation across the Carrizo segment of the San Andreas Fault, Cal-
ifornia: Impact of laterally varying crustal properties. Journal of Geophysical Research, 111(B5). https://doi.org/10.1029/2005JB003843
Scholz, C. H. (1968). Mechanism of creep in brittle rock. Journal of Geophysical Research, 73(10), 3295–3302. https://doi.org/10.1029/
JB073i010p03295
Scholz, C. H. (2019). The mechanics of earthquakes and faulting. Cambridge University Press.
Sjöberg, J., Christiansson, R., & Hudson, J. A. (2003). ISRM suggested methods for rock stress estimation: Part 2: Overcoring methods.
International Journal of Rock Mechanics and Mining Sciences, 40(7–8), 999–1010. https://doi.org/10.1016/j.ijrmms.2003.07.012
Sleep, N. H., & Blanpied, M. L. (1992). Creep, compaction and the weak rheology of major faults. Nature, 359(6397), 687–692. https://doi.
org/10.1038/359687a0
Stephansson, O., & Zang, A. (2012). ISRM suggested methods for rock stress estimation—Part 5: Establishing a model for the in situ stress
at a given site. Rock Mechanics and Rock Engineering, 45(6), 955–969. https://doi.org/10.1007/s00603-012-0270-x
Tamagawa, T., & Pollard, D. D. (2008). Fracture permeability created by perturbed stress fields around active faults in a fractured basement
reservoir. AAPG Bulletin, 92(6), 743–764. https://doi.org/10.1306/02050807013
Townend, J., & Zoback, M. D. (2004). Regional tectonic stress near the San Andreas Fault in central and southern California. Geophysical
Research Letters, 31(15). https://doi.org/10.1029/2003GL018918
Vavryčuk, V. (2014). Iterative joint inversion for stress and fault orientations from focal mechanisms. Geophysical Journal International,
199(1), 69–77. https://doi.org/10.1093/gji/ggu224
Vermilye, J. M., & Scholz, C. H. (1998). The process zone: A microstructural view of fault growth. Journal of Geophysical Research, 103(B6),
12223–12237. https://doi.org/10.1029/98JB00957
Wibberley, C. (1999). Are feldspar-to-mica reactions necessarily reaction-softening processes in fault zones? Journal of Structural Geology,
21(8–9), 1219–1227. https://doi.org/10.1016/S0191-8141(99)00019-X
Wilson, J. E., Chester, J. S., & Chester, F. M. (2003). Microfracture analysis of fault growth and wear processes, Punchbowl Fault, San An-
dreas system, California. Journal of Structural Geology, 25(11), 1855–1873. https://doi.org/10.1016/S0191-8141(03)00036-1
Yale, D. P. (2003). Fault and stress magnitude controls on variations in the orientation of in situ stress. Geological Society, London, Special
Publications, 209(1), 55–64. https://doi.org/10.1144/GSL.SP.2003.209.01.06
Yoshinaka, R., & Yamabe, T. (1986). Joint stiffness and the deformation behaviour of discontinuous rock. International Journal of Rock
Mechanics and Mining Sciences, 23(1), 19–28. https://doi.org/10.1016/0148-9062(86)91663-3
Zhang, F., An, M., Zhang, L., Fang, Y., & Elsworth, D. (2019). The role of mineral composition on the frictional and stability properties
of powdered reservoir rocks. Journal of Geophysical Research: Solid Earth, 124(2), 1480–1497. https://doi.org/10.1029/2018JB016174
Zhang, S., & Ma, X. (2021). Global frictional equilibrium via stochastic, local Coulomb frictional slips. Journal of Geophysical Research:
Solid Earth, 126, e2020JB021404. https://doi.org/10.1029/2020JB021404
Zhao, X. G., & Cai, M. (2010). A mobilized dilation angle model for rocks. International Journal of Rock Mechanics and Mining Sciences,
47(3), 368–384. https://doi.org/10.1016/j.ijrmms.2009.12.007
Zoback, M. D. (2010). Reservoir geomechanics. Cambridge University Press.
Zoback, M. D., & Townend, J. (2001). Implications of hydrostatic pore pressures and high crustal strength for the deformation of intraplate
lithosphere. Tectonophysics, 336(1–4), 19–30. https://doi.org/10.1016/S0040-1951(01)00091-9
Zoback, M. D., Zoback, M. L., Mount, V. S., Suppe, J., Eaton, J. P., Healy, J. H., et al. (1987). New evidence on the state of stress of the San
Andreas Fault system. Science, 238(4830), 1105–1111. https://doi.org/10.1126/science.238.4830.1105
Zoback, M. L. (1992). First-and second-order patterns of stress in the lithosphere: The World Stress Map Project. Journal of Geophysical
Research, 97(B8), 11703–11728. https://doi.org/10.1029/92JB00132

2021JB022348 Skib BNGST

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

2021JB022348 Skib BNGST

Uploaded by

Copyright:

Available Formats

RESEARCH ARTICLE How Does In Situ Stress Rotate Within a Fault Zone?

2. Constitutive Behavior of Frictional Fractures

2.1. Stress-Displacement Relationship for Fractures Under Loading Condition

The matrix resistance to the opening of fracture walls km is given by:

Therefore, Equation 2 can be written as:

By substituting τf and τm into Equations 6 and 7, we further obtain:

2.2. Stress-Displacement Relationship for Fractures Under Unloading Condition

3. Effective Properties of Fractured Rock Mass Within a Fault Damage Zone

3.1. Boundary Conditions of a Fault Damage Zone

3.2. Updating Stress Components on the Strain Boundary

4. Model Experimentation and Results

4.1. Generation of Discrete Fracture Networks

in which fracture length is normalized by the model size. The constant

Assuming that fractures can hardly interpenetrate under compression

(S1 = 150 MPa, S3 = 75 MPa, Pp = 40 MPa). A buried depth of ∼4 km is

assumed so that the vertical stress, or the intermediate principal stress

4.2. Effect of Fracture Shear Stiffness on Stress Variations

4.3. Effect of Fracture Frictional Coefficient on Stress Variations

In this section, we further examine the effect of fracture frictional coef-

4.4. Effect of Local Pore Pressure on Stress Variations

The negligible pore pressure effect in these simulations can be account-

4.5. Influences of Fracture Deformation on the Effective Elastic Properties

Our simulations are performed based on the multilayer model, which

5.3. Spatial-Temporal Variations of the In Situ Stress Within Fault Zones

Appendix A: Relation of Local Fracture Deformation and Global Strain

with the superscript m representing the purely elastic (intact) matrix.

Appendix B: Bisection Method for Stress Updates

Step-1: Calculate the midpoint of the interval, Δσyy,j = (m + n)/2;

Data Availability Statement

You might also like