international journal of solids and structuresborja/pub/ijss2018(1).pdf · 2018. 11. 23. · k.c....
TRANSCRIPT
International Journal of Solids and Structures 143 (2018) 218–231
Contents lists available at ScienceDirect
International Journal of Solids and Structures
journal homepage: www.elsevier.com/locate/ijsolstr
Hyper-elastoplastic/damage modeling of rock with application to
porous limestone
K.C. Bennett a , b , R.I. Borja
a , ∗
a Department of Civil and Environmental Engineering, Stanford University, Stanford, CA 94305, USA b Fluid Dynamics and Solid Mechanics Group, Theoretical Division, Los Alamos National Laboratory, Los Alamos, NM 87545, USA
a r t i c l e i n f o
Article history:
Received 3 October 2017
Revised 31 January 2018
Available online 13 March 2018
Keywords:
Rock
Damage
Microfracture
Porosity
Eshelby stress
Drucker-Prager
a b s t r a c t
Relations between porosity, damage, and bulk plasticity are examined in the context of continuum dam-
age and hyper-elastoplasticity of porous rocks. Attention is given to a thermodynamically consistent
derivation of the damage evolution equations and their role in the constitutive equations, for which the
Eshelby stress is found to be important. The provided phenomenological framework allows for volumet-
ric damage associated with pore growth to be distinguished from the isochoric damage associated with
distributed microcracks, and a novel Drucker-Prager/cap type material model that includes damage evo-
lution is presented. The model is shown to capture well the hardening/softening behavior and pressure
dependence of the so-called brittle-ductile transition by comparison with confined triaxial compression
measurements from the literature. Non-linear finite element simulations are also provided of the predic-
tion of damage within porous limestone around a horizontal borehole wall.
© 2018 Elsevier Ltd. All rights reserved.
t
o
n
s
1
a
b
c
u
C
c
i
(
1
a
B
s
a
t
r
a
t
1. Introduction
Porous rocks exhibit microfracture and changes in porosity un-
der inelastic deformation, which affect their strength and elastic
properties. Describing their inelastic deformation by means of clas-
sical plasticity theory typically requires attributing changes in the
relative void space (porosity) to bulk plasticity, representing di-
latant and compactive volume changes (e.g., Drucker, 1957; Borja
and Tamagnini, 1998; Fossum et al., 20 0 0; Borja and Choo, 2016 ,
among others). Such models often incorporate phenomenologi-
cal hardening/softening rules (e.g., Borja and Lee, 1990; Simo and
Meschke, 1993; Borja and Tamagnini, 1998; Tamagnini and Ciantia,
2016 ). Including the evolution of elastic properties with microfrac-
ture and changes in porosity typically requires introducing either
empirical dependence of elastic moduli (and hardening moduli) on
the extent of plastic strain (cf. Borja and Lee, 1990 , among others)
or, alternatively, a ductile damage law which relates the damaged
state of the material to the extent of plastic strain (cf. Hueckel,
1976; Ortiz, 1985; Ju, 1989 , among others).
Continuum damage mechanics (CDM), generally attributed to
being first described by Kachanov (1958) , provides a description
of elastic softening ascribed to material degradation by micro-
cracking and/or void growth. The coupling of CDM with elastoplas-
∗ Corresponding author.
E-mail addresses: [email protected] (K.C. Bennett), [email protected] (R.I. Borja).
o
s
t
a
https://doi.org/10.1016/j.ijsolstr.2018.03.011
0020-7683/© 2018 Elsevier Ltd. All rights reserved.
icity, i.e., ductile damage theory, further provides a representation
f CDM as an energy-dissipative process, allowing for a thermody-
amically consistent representation of the material’s elastic con-
titutive response dependent on its damaged state (cf. Lemaitre,
985a, 1985b; Ju, 1989 ). In ductile damage theory, material dam-
ge affects both the elastic response and the material’s strength
y identifying the damage-effective stress, which provides a physi-
ally meaningful description of how a material becomes weakened
nder microcrack/void growth (cf. Lemaitre and Desmorat, 2005 ).
ombining hyper-elastoplasticity theory with a thermodynamically
onsistent description of damage for geomaterials poses longstand-
ng challenges, especially with respect to dilation and compaction
cf. Simo and Ju, 1987; Krajcinovic, 1996; Arson, 2014 ).
Relations between Eshelby’s stress tensor ( Eshelby, 1951; 1956;
975 ) and changes to a material’s free energy state during dam-
ge have long been recognized (e.g., Rice, 1968; Maugin, 1994;
rünig, 2004 ). Recently, a description of the role of the Eshelby
tress specifically in describing the hyper-elastoplasticity of materi-
ls that undergo large inelastic changes in volume (such as geoma-
erials) was provided by Bennett et al. (2016) . The implications of
ecognizing this are examined herein with respect to ductile dam-
ge theory of porous rocks, and a novel hyper-elastoplastic duc-
ile damage framework is provided. This work, contrary to that
f Bennett et al. (2016) , explores the relevance of the Eshelby
tress in formulating ductile damage constitutive models. Notably,
he constitutive theory unifies concepts of volumetric (also known
s “dilative” or “adhesive”) damage and dilation/compaction in a
K.C. Bennett, R.I. Borja / International Journal of Solids and Structures 143 (2018) 218–231 219
Fig. 1. Micro-fracture and porosity changes evident in an SEM image of post-
indented Woodford shale in the indented region. Details of experiments are pro-
vided in Bennett et al. (2015) .
t
t
b
m
a
d
2
2
u
d
R
T
2
a
J
a
t
s
2
v
(
Z
c
t
a
e
2
W
(
fi
b
e
i
d
l
b
t
b
e
o
a
u
B
2
a
r
a
m
t
i
s
d
a
(
p
a
i
t
a
T
u
w
t
I
d
T
m
d
w
i
t
d
t
d
m
d
h
o
t
b
t
n
r
V
h
h
d
t
s
t
i
s
h
t
s
L
o
r
e
2
p
t
a
o
J
a
j
f
p
hermodynamically consistent framework. The present work is mo-
ivated also by scanning electron microscope (SEM) observations
y the present authors of extensive micro-fracture and changes in
icro-void space induced by nanoindentation of a Woodford shale,
s described by Bennett et al. (2015) and shown in Fig. 1 .
Observations of microfracture in cohesive geomaterials are well
ocumented in the literature ( Cuss et al., 2003; Wong and Baud,
012; Bennett et al., 2015; Abousleiman et al., 2016; Hu et al.,
017; Semnani and Borja, 2017 ). Over the years, numerous contin-
um damage models have been proposed to model this type of
amage in rocks, both isotropic (e.g., Hajiabdolmajid et al., 2002;
icard and Bercovici, 2003; Hamiel et al., 2004; Liu et al., 2016;
ran-Manh et al., 2016 ) and anisotropic damage (e.g., Chen et al.,
010; Voyiadjis et al., 2012; Arson and Gatmiri, 2012 ), including
lso application to damage zones at faults and joints ( White, 2014;
ohri et al., 2014 ). The description of porosity changes as a dam-
ge and/or healing phenomena has, further, particularly been a
opic of recent interest ( Ju et al., 2012; Zhu and Arson, 2015; Ar-
on and Pereira, 2013; Arson and Vanorio, 2015; Le Pense et al.,
016 ). Micromechanical descriptions of damage have been pro-
ided for (non-cohesive) granular ( Zhu et al., 2010 ) and cohesive
Chazallon and Hicher, 1998; Zhu et al., 2008; Guéry et al., 2008;
hu et al., 2011; Shen et al., 2012 ) geomaterials, with microme-
hanical descriptions of porosity and damage provided in the con-
ext of Modified Cam-Clay material model ( Bignonnet et al., 2016 )
nd the degradation of cemented clays ( Ju et al., 2012; Nguyen
t al., 2014 ). The instability of granular assemblies ( Andrade et al.,
008; Chang and Meidani, 2013; Misra and Poorsolhjouy, 2015b;
ang et al., 2016 ) can also be described as a damage phenomenon
cf. Misra and Poorsolhjouy, 2015a ).
The equating of damage with pore growth appears to have been
rst proposed by Gurson (1977) , and has since found much utility
y others ( Salari et al., 2004; Arson and Pereira, 2013; Lebensohn
t al., 2013; Bronkhorst et al., 2016; Bennett et al., 2018 ). Contrar-
ly, some have ascribed damage purely to microfracture without in-
uced void space (e.g., Zhu et al., 2008; Chen et al., 2010; Buech-
er and Luscher, 2014 ). Another approach that effectually com-
ines some volumetric damage with deviatoric damage contribu-
ions was proposed by Ortiz (1985) for concrete, and has since
een widely used for describing damage of rocks (e.g., Chiarelli
t al., 2003; Parisio et al., 2015 , among others), where it is assumed
nly the positive principal stresses contribute to damage. Various
lternative approaches have been proposed to account for both vol-
metric and isochoric damage separately ( Lee and Fenves, 1998;
ažant et al., 20 0 0; Clayton, 20 06; Bakhtiary et al., 2014; White,
014; Shojaei et al., 2014; Clayton and Tonge, 2015; Grgic, 2016 ,
mong others).
Herein, a kinematic and thermodynamic description is provided
elating phenomenological damage parameters to both volumetric
nd isochoric deformation attributable to changes in porosity and
icro-fracture, respectively. The role of a spatial Eshelby-like stress
ensor in the damage-constitutive equations is described, which
s shown to be equivalent to the “Eshelby-zeta” stress tensor de-
cribed by Bennett et al. (2016) , but who made no connection with
amage mechanics. The coupling of critical state and ductile dam-
ge theory is provided in terms of a novel Drucker-Prager/Damage
DP-D) material model that consists of a smooth transition into a
ressure cap of the yield surface. Two distinct mechanisms of dam-
ge are considered: (1) volumetric damage attributed to changes
n void ratio (i.e., porosity), and (2) isochoric damage attributed
o the development of microcracks—with or without an associ-
ted void space opening (i.e., exclusive of their crack aperture).
his unifies two common but disparate descriptions of contin-
um damage, where the former description does not lend itself
ell to describing a statistical distribution of microcrack orienta-
ions, e.g., as necessary for describing anisotropic damage (c.f. Ken-
chi, 1984; Krajcinovic, 1996; Voyiadjis et al., 2012 ), and the latter
oes not account for crack openings along their length (apertures).
he two types of damage are developed with attention to ther-
odynamic restrictions, in particular the evolution of volumetric
amage is shown to arise directly from thermodynamic principles,
here damage is predicted to increase under dilation and reduce
n compression. Considering that this provides the connection be-
ween ductile damage and inelastic volumetric strain, the depen-
ence of the state of damage on isochoric plastic deformation is
hen introduced through a phenomenolgical isochoric (also called
eviatoric or shear) damage parameter, which has a clear physical
eaning as a measure of the shear induced microfracture. In or-
er to examine the ability of the proposed framework to represent
ardening/softening behavior through the evolution of damage, no
ther phenomenological hardening parameters are introduced, i.e.,
he proposed framework is not a mixed hardening/damage model
ut a purely ductile damage one.
The DP-D material model is developed for finite deforma-
ions, and the model is implemented with Abaqus non-linear fi-
ite element modeling software within a UMAT material sub-
outine. Calibration to the triaxial compression (TC) tests of
ajdova et al. (2004) on Tavel limestone are provided to show
ow the model is capable of capturing observed stress-strain and
ardening/softening behavior, in particular the observed pressure
ependence of the so-called brittle-ductile transition described in
hat work. Simulation of a horizontal wellbore in porous lime-
tone as described by Coelho et al. (2005) is also conducted, and
he model predictions of damage around the wellbore are exam-
ned. We note that issues of objectivity as they relate to mesh
ize dependence and localization of damage are not addressed
erein, as the present model is meant to address a (local) con-
inuum description of damage. The need to integrate global de-
criptions of damage and fracture (e.g., Clayton and Knap, 2015;
i et al., 2015; Zhang et al., 2016 ) with continuum damage the-
ry (cf. Chen et al., 20 0 0; Gao and Huang, 2003; Bazant, 2010 )
emains an issue of much present research effort (e.g., Verhoosel
t al., 2011; Li et al., 2015; Heyden et al., 2015; Weed et al.,
017; Tjioe and Borja, 2016 ), and is beyond the scope of this
resent work. We also note that the present work is restricted
o material isotropy, while extension to anisotropic ductile dam-
ge is part of ongoing research effort s. We refer to the works
f Nova (1980) ; Cazacu et al. (1998) ; Semnani et al. (2016) ;
iang et al. (2017) among others for a description of the
nisotropic plasticity of geomaterials and Al-Rub and Voyiad-
is (2003) ; Kondo et al. (2010) ; Chen et al. (2010) among others
or a description of anisotropic damage coupled with anisotropic
lasticity.
220 K.C. Bennett, R.I. Borja / International Journal of Solids and Structures 143 (2018) 218–231
i
p
c
s
v
a
w
m
A
v
i
a
s
c
t
a
o
b
s
t
c
s
n
t
d
a
c
w
t
t
d
e
i
i
o
i
p
t
i
J
b
W
m
i
u
s
s
t
t
b
s
i
i
i
b
J
w
d
n
i
2. Theory describing damage, porosity, and micro-fracture
In the following, a description of the kinematics and adopted
notation is first given. The description is made in as con-
cise a manner as possible since the emphasis of the work is
the development of the damage-constitutive model. We refer to
Bennett et al. (2016) for further kinematic and thermodynamic
background relevant to the current work. Throughout, tensors are
written in boldface and their components are specified by sub-
script index notation. Summation convention is assumed for re-
peated indices. Matrix notation for the inner product is assumed,
whereas dyadic products are indicated by �. The double contrac-
tion of indices are indicated by double dots (: ). For example, the
second order identity tensor is represented as 1 and its compo-
nents by the Kronecker delta δij , such that the trace of a second
order tensor is given by tr [ (•) ] := (•) : 1 = δi j (·) i j . We make use
of similar standard notion for the natural log ln ( · ), the deviatoric
operator dev [ (•) ] := (•) − 1 / 3 tr [ (•) ] 1 , and the symmetric opera-
tor sym [(•)] := 1 / 2[(•) + (•) T ] , where the T superscript denotes
the transpose.
The concept of a representative elementary volume (REV) is as-
sumed, such that the material properties at a point describe the
microstructure, including herein microcracks and also void space.
Typically, fluids may fill the voids with time dependent dissipation
giving rise to effective stresses ( Biot, 1941; Terzaghi, 1944; Borja,
2006 ). All stresses described herein are always considered to be
in this regard effective stresses, and we note that the model pre-
sented here could be applied to coupled fluid/solid problems with
the addition of fluid phase diffusion laws (e.g., Choo and Borja,
2015; Wang and Sun, 2016 ).
We adopt standard mechanics sign convention, where, for ex-
ample, positive mean stress in compressive; however, when com-
paring to measurements where opposite (geomechanics) conven-
tion is used we sometimes report negative values in order to
match those figures and also adopt, for example, the convention
for positive “axial strain” to imply compression. Conventional no-
tation for the first stress invariant I (•) 1
:= tr [ (•) ] and the second
deviatoric stress invariant J (•) 2
:= 1 / 2 dev [ (•) ] : dev [ (•) ] is adopted,
where here ( •) denotes a stress measure. The mean stress (of any
particular stress measure) is then defined by p (•) := 1 / 3 I (•) 1
, and
the Mises stress as q (•) :=
√
3 J (•) 2
. Consistent with convention, un-
specified description of “mean stress” or “Mises stress” implies
with respect to Cauchy stress.
The deformation gradient is a function of position and time,
given by
F :=
ˆ F ( X , t) =
∂ ϕ ( X , t)
∂ X
, (1)
where x = ϕ ( X , t) describes a deformation from the reference to
current configurations of the material, i.e., ϕ t : B 0 → B t ⊂ R
3 . The
multiplicative split of the deformation gradient into elastic and
plastic parts is assumed, such that F = F e F p . The left Cauchy-Green
deformation tensor (Finger tensor) of B t is given by
b := F F T ; b i j := F iK F jK , (2)
and the logarithmic strain (Hencky strain) by
ε :=
1
2
ln [ b ] ; ε i j :=
1
2
ln [ b i j ] . (3)
The natural volume strain is then defined as e v := ε : 1 = ln [ J] =e e v + e
p v such that the natural deviatoric strain is ε := ε − 1 / 3 e v 1 ,
where J := det [ F ] = J e J p is the Jacobian determinant. This allows
for a representation of large strain elastoplasticity that follows
closely the small strain version, as has been described by Simo
(1992, 1998) and others. Note that intermediate configuration
quantities are represented with an over-bar, current configuration
n lower-case, and reference configuration in upper-case, whenever
ossible. For example infinitesimal volumes are mapped by the Ja-
obian determinant, J , by dv = J e d V and d V = J p dV, where J = J e J p .
Two scalar variables of damage, d
c and d
s , are distinguished, as-
ociated with volumetric and isochoric damage, respectively. The
olumetric part d
c , which we call also the comprehensive dam-
ge parameter, is isotropic because it is directly related to porosity,
hich cannot be given a statistically preferential direction or align-
ent according to standard REV theory (cf. Hill, 1963; Borja, 2006 ).
n advantage of the present model is that because it separates the
olumetric from the isochoric contributions to the overall damage,
t allows for future work to consider the isochoric damage to be
nisotropic. This is significant because an anisotropic damage ten-
or can then be assembled from the statistical distribution of mi-
rocracks when the cracks are considered to have no opening by
he method first proposed by Lubarda and Krajcinovic (1993) (see
lso Voyiadjis et al., 2012 ), which is analogous to the development
f the micro-mechanical fabric tensor used to represent the distri-
ution of particle contact planes in granular assemblies (e.g., as de-
cribed in Chang and Bennett, 2015 ). We emphasize, however, that
he present work considers only an isotropic distribution of micro-
racks, allowing for the isochoric damage to be represented by a
calar variable. Subsequently, it will be shown that d
s is taken as a
on-linear function of its conjugate energy release rate y s , similar
o the general form proposed by Lemaitre and Chaboche (1994) ,
s =
ˆ d
s (y s ) . The terms “deviatoric damage” and “shear damage”
re variably used in the literature for what we call here “iso-
horic damage,” and we use these terms interchangeably as
ell.
The porosity ( n ) is the ratio of the volume of the voids to the
otal volume in the current configuration, i.e, n := dv v / dv , where
he total volume is the sum of the solid and void volumes, i.e.,
v = dv s + dv v . Changes in porosity result from both plastic and
lastic deformation; however, it is convenient to define the poros-
ty relative to the plastically-deformed/elastically-unloaded state,
.e., the intermediate configuration B . In this case, the definition
f porosity becomes n := d V v /d V . This is the definition of poros-
ty used in this work, which is the assumption that change in
orosity is attributed to plastic volumetric deformation, and plas-
ic volumetric deformation is likewise attributed solely to changes
n porosity. This assumption is analogous to the assumption that
e ≈ 1, which is reasonable if the elastic volume change is negligi-
le in relation to inelastic volume change, for which case n ≈ n .
e note that this assumption is commonly justifiable for geo-
aterials. For example, consider triaxial compression tests where
nitial and final specimen porosities (before loading and after
nloading) can differ significantly relative to the elastic volume
train exhibited during the test. A further justification for this as-
umption is that the solid grains are expected to exhibit elas-
ic volumetric deformation but not plastic. Hence, all of the plas-
ic deformation is reasonably attributed to changes in porosity,
ut the separation of bulk elastic deformation into that of the
olid and that of the void fractions is not obvious without mak-
ng some other assumptions (or perhaps higher resolution model-
ng, e.g., micromechanical modeling). With this definition of poros-
ty, the plastic Jacobian determinant, J p , is related to porosity
y
p =
ρ0
ρ=
d V
dV
=
d V
s + d V
v
d V
s + d V
v =
d V − d V
v + d V
v
dV
= 1 − n 0 +
d V
v
dV
,
(4)
here because of near-incompressibility of the solid constituent,
v s = d V s = d V s , and the subscript 0 is used throughout to de-
ote initial (or reference) values. Similarly, the definition of poros-
ty provides
K.C. Bennett, R.I. Borja / International Journal of Solids and Structures 143 (2018) 218–231 221
n
S
t
i
t
c
o
e
t
e
c
f
p
t
d
o
J
T
t
s
t
p
t
�
E
t
d
J
W
c
a
p
I
t
3
f
i∫
w
s
B
b
b
t
t
b
ζ
I
i
(
D
w
a
t
ξ
i
t
w
i
w
s
i
D
w
a
e
t
p
s
G
w
G
w
G
p
t
p
t
i
t
m
l
t
G
T
d
d
:=
dv v
dv ≈ d V
v
d V
=
d V − dV
s
d V
=
d V − dV + dV
v
d V
=
d V − dV + f 0 dV
d V
= 1 − 1 − n 0
J p . (5)
ince we should expect that damage is at least in part attributable
o changes in porosity (see references in the introduction), we
dentify that there is a volumetric damage parameter proportional
o porosity by α, i.e., d
c = αn . It is noted that this relationship
ould in general be nonlinear, where, for example, α could depend
n the plastic strain history. However, we do not need to specify αxplicitly here nor make any assumptions about how d
c is related
o the total damage, only recognize the proportionality for consid-
ration in Section 3 ; the precise form will be determined by the
hoice of constitutive equations in Section 4 , where an expression
or d
c as a function of volumetric plastic deformation and model
arameters is obtained. The expression for the proportionality be-
ween volumetric damage and J p can thus be written as
c = αn = α(
1 − 1 − n 0
J p
), (6)
r in other words,
p =
1 − n 0
1 − d
c /α. (7)
he relation between J p and d
c in Eq. (7) is remarkably similar to
he relation often used in modeling spall damage of metals at large
trains, which appears to be first obtained by Davison et al. (1977) ,
he difference being in only the scaling factor α and the initial
orosity 1 The relation between changes in damage and porosity is
hen found to be given by
d
c = α�n = α(n − n 0 ) = α(1 − n 0 ) (
1 − 1
J p
). (8)
q. (8) provides an explicit expression for the plastic Jacobian de-
erminant in terms of the damage increment when volumetric
amage is related to volumetric plasticity,
p =
1
1 − �d c
α(1 −n 0 )
. (9)
e emphasize that this relationship is based only on the classi-
al assumption that there is some proportionality between damage
nd porosity and that we do not require α to be a fixed material
arameter in the derivation of our material model in what follows.
t will be shown in Section 5 that the proposed model can be used
o solve for α in Eq. (8) by providing { J p , d
c }.
. Thermodynamic consistency
The second law of thermodynamics, neglecting temperature ef-
ects, can be expressed in the form of the Clausius-Planck inequal-
ty (cf. Truesdell and Noll, 2004; Simo, 1998 ),
B t
(d : σ − ρ ˙
)dv ≥ 0 , (10)
here d = sym [ F F −1 ] is the deformation rate, σ is the Cauchy
tress, ρ is density, and is the specific Helmholtz free-energy.
ennett et al. (2016) have shown that for a material undergoing
1 Equivalence to the spall damage expression of Davison et al. (1977) is obtained
y setting α = 1 in the absence of initial porosity.
w
l
d
ulk plasticity this implies that the hyperelastic constitutive equa-
ion for the stress is given for the current configuration represen-
ation of elastoplasticty in terms of the so-called “zeta” stress, ζ,
y
:= J e σ = 2
∂( ρ)
∂ b
e b
e . (11)
n order for a material model to be thermodynamically consistent,
t was shown in Bennett et al. (2016) that the plastic dissipation
D) must satisfy
= ξ : d
p − q : ˙ z ≥ 0 , (12)
here d p is the plastic part of the deformation rate d = d p + d e , z
nd q are respectively vectors of internal state variables (ISV’s) and
heir conjugate thermodynamic forces, and
:= J e σ − ( ρ) 1 = ζ − ( ρ) 1 (13)
s an Eshelby-like stress measure we call the Eshelby-zeta stress.
In this work, the free-energy is assumed to consist of only
he elastic stored energy and internal state variables associated
ith damage (e.g., there is no additional hardening potential). It
s taken to be of the functional form,
=
ˆ ( b
e , d
c , d
s ) , (14)
here b e is the elastic Finger tensor, and d
c and d
s are ISV’s de-
cribing the damaged state of the material. The dissipation inequal-
ty hence takes the form
= ξ : d
p − y c ˙ d
c − y s ˙ d
s ≥ 0 , (15)
here y c = ∂ d c ( ρ) and y s = ∂ d s ( ρ) are the energy release rates
ssociated with comprehensive (i.e., reduction of the total free-
nergy) and purely deviatoric damage (i.e., microfractures), respec-
ively.
Under the supposition of non-associated flow, the dissipation
otential G and yield function F are taken to be of the forms, re-
pectively,
:=
ˆ G ( ξ, y c , y s ; d
c , d
s ) ; F :=
ˆ F ( ξ, y c , y s ; d
c , d
s ) , (16)
ith the dissipation potential taken to be of the additive form
: = G p ( ξ, y c ; d
c , d
s ) + G d (y c , y s ; d
c , d
s )
= G p ( ξ, y c ; d
c , d
s ) + G c (y c ; d
c ) + G s (y s ; d
s ) , (17)
here G p is the (non-associative) plastic potential and the G d =
c + G s is the damage potential, assumed to be additively decom-
osed into comprehensive and deviatoric parts, G c and G s , respec-
ively. Although additive decomposition into plastic and damage
arts is typical (cf. Lemaitre and Desmorat, 2005 ), the assumption
hat the damage potential can be further additively decomposed
s justified by recognizing the functional dependence of the plas-
ic potential on y c due to the Eshelby-zeta stress (but not y s ). This
otivates the concept adopted here that G c can be somehow re-
ated to G p . In anticipation of providing a damage evolution equa-
ion consistent with Eq. (6) , we take the form of this relation to be
c := c αG p . (18)
his provides a convenient canonical form of the comprehensive
amage evolution equation in terms of the plastic potential,
˙
c = − ˙ λ∂G c ∂y c
= − ˙ λc α∂G p ∂y c
, (19)
here ˙ λ is the plastic multiplier. Similarly, the shear damage evo-
ution equation is given by
˙
s = − ˙ λ∂G s ∂y s
, (20)
222 K.C. Bennett, R.I. Borja / International Journal of Solids and Structures 143 (2018) 218–231
Table 1
List of damage state variables and associated symbols.
Type Symbol Definition
Damage internal variables d c volumetric damage variable
d s shear damage variable
Damage functions �c volumetric damage function
�s shear damage function
ω
c := �c −1 inverse of volumetric damage function
ω
s := �s −1 inverse of shear function
Energy release rates y c volumetric energy release rate
y s shear energy release rate
Damage potentials G d = G c + G s total damage potential
G c volumetric damage potential
G s shear damage potential
T
T
o
ζ
w
ζ
N
s
ξ
T
t
y
w
p
t
a
c
E
ξ
s
ξ
w
ζ
W
4
f
F
T
e
s
t
d
t
e
and the flow rule by
d
p =
˙ λ∂G p ∂ ξ
. (21)
Along with the standard so called KKT and consistency conditions
(cf. Borja, 2013 ), this provides the complete canonical form of the
(hyper-elastoplastic/damage) constitutive and evolution equations.
4. Modeling choices and implementation
The material model presented in this section can be called
a Drucker-Prager/Damage (DP-D) model. It is loosely based on
the Drucker-Prager type model first presented by Regueiro and
Ebrahimi (2010) and further modified by Bennett et al. (2016) . This
model incorporates a two-invariant Drucker-Prager yield surface
with a smooth transition to a pressure cap (DP-C), and has been
further developed herein (incorporating damage) especially for the
purpose of modeling the constitutive behavior of rocks. It is noted,
however, that Drucker-Prager type models inclusive of a pressure
cap have found broad application in the modeling of porous ma-
terials in general, and the model as presented here may be sim-
ilarly extendable (with proper selection of model parameters). It
is emphasized that the relationship that is described in the fol-
lowing between the damage-energy release rate and the Eshelby-
zeta stress of Eq. (13) provides a useful connection between en-
forcing thermodynamic consistency and providing a description of
the hardening/softening behavior of a porous material.
4.1. Helmholtz free energy
The free-energy function is taken to depend on the state
of damage described by the scalar damage variables d
c and d
s
through the damage functions �c =
ˆ �c (d
c ) and �s =
ˆ �s (d
s ) , re-
spectively. It is considered to be additively composed of volumet-
ric and deviatoric elastic stored energy potentials in the functional
form
=
ˆ (e e v , ε
e , d
c , d
s )
= �c (d
c ) ·(
˜ v (e e v ) + �s (d
s ) · ˜ s ( ε
e ) ). (22)
The damage functions 0 < �c ≤ 1 and 0 < �s ≤ 1 range between
1 for the undamaged state and approach 0 for the completely
damaged state of the material. Consistent with the principle
of strain equivalence ( Lemaitre, 1985a, 1985b ), damage-effective
stress and the corresponding damage-effective free-energy are
identified. Damage-effective quantities are represented with tilde
notation, such that the per unit volume damage-effective volumet-
ric and deviatoric elastic stored energy potentials are given respec-
tively by
ρ ˜ v (e e v ) :=
K 0
2
e e 2
v ; ρ ˜ s ( ε
e ) := μ0 ε
e : ε
e , (23)
where K 0 and μ0 are the damage-effective bulk and shear moduli,
respectively. The damage-effective free-energy per unit volume is
then expressed by
ρ ˜ =
K 0
2
e e 2
v + μ0 ε
e : ε
e . (24)
The damage functions could in general be non-linear, but are taken
for the sake of simplicity and clarity here as linear,
�c := (1 − d
c ) ; �s := (1 − d
s ) , (25)
for 0 ≤ d
c < 1 and 0 ≤ d
s < 1. It is convenient to also define (consis-
tent with standard notation)
ω
c := �c −1 =
1
(1 − d
c ) ; ω
s := �s −1 =
1
(1 − d
s ) . (26)
he comprehensive damage-effective free-energy is defined as
˜ c := ω
c . (27)
he stress is then found from the hyperelastic constitutive relation
f Eq. (11) to be given by
= (1 − d
c ) ·(K 0 e
e v 1 + (1 − d
s )2 μ0 ε
e ), (28)
ith the damage-effective stress thus given by
˜ = ω
c ω
s dev [ ζ] + ω
c p ζ 1 = K 0 e e v 1 + 2 μ0 ε
e , (29)
ote too that the expression for the damage-effective Eshelby-zeta
tress is
˜ = ω
c ω
s dev [ξ]
+ ω
c p ξ 1
= dev [
˜ ζ]
+ ( p ζ − ρ ˜ c ) 1 . (30)
he expressions for the energy release rates are found according to
heir definitions within Eq. (15) to be
c =
∂( ρ)
∂d
c = −ρ ˜ c ; y s =
∂( ρ)
∂d
s = −(1 − d
c ) ρ ˜ s , (31)
hich completes the definition of all damage state variables and
otentials. A list of the damage state variables and their defini-
ions is provided in Table 1 . Having defined the damage state vari-
bles, the expressions for the stress and damage-effective stress
an be identified. Making use of the definition in Eq. (13) with
qs. (27) and (31) , the Eshelby-zeta stress is expressed as
= ζ + (1 − d
c ) y c 1 , (32)
uch that
˜ =
˜ ζ + y c 1 , (33)
hich implies
˜ =
˜ ξ − y c 1 . (34)
ithin the described framework, we note the following relations:
˜ p ζ =
˜ p ξ − y c ; ˜ q ζ ≡ ˜ q ξ . (35)
.2. Yield function and plastic potential
The Drucker-Prager/cap (DP-C) type yield function is taken as a
unction of the hyperelastic damage-effective stress ˜ ζ = J e ˜ σ, i.e.,
:=
ˆ F ( ζ) . (36)
his form is convenient because it both provides a bound for the
lastic domain E , i.e., E := { ζ ∈ S | F( ζ) ≤ 0 } , and is tenable in the
ense that its parameters can be associated with material proper-
ies obtained from measurements (e.g., friction angle, preconsoli-
ation stress, etc.). However, the yield function is required to be of
he functional form described by Eq. (16) in order to find the nec-
ssary derivatives described within Section 3 . Fortuitously, we can
K.C. Bennett, R.I. Borja / International Journal of Solids and Structures 143 (2018) 218–231 223
Fig. 2. DP-D yield surface in ˜ q vs. −3 p damage-effective stress invariant space,
showing intersection with first stress invariant axis at X φ .
s
i
o
o
(
w
A
F
〈
X
(
h
n
t
t
o
t
s
t
e
C
n
b
f
t
l
E
G
Table 2
List of DP-D model parameters.
Type Parameter Definition
Elastic K 0 damage-effective bulk modulus
ν0 damage-effective Poisson’s ratio
Yield surface ˜ β1 damage-effective cohesion stress parameter ˜ β2 damage-effective preconsolidation stress parameter ˜ φ damage-effective friction angle ˜ ψ damage-effective dilation angle
R cap ellipticity parameter
Damage c t dilative damage parameter
c c compressive damage parameter
c s shear damage parameter
Table 3
Model parameters calibrated to measurements of Vajdova et al. (2004) .
Type Parameter Value Units Maximum variation
Elastic K 0 29.6 GPa ± 7.1%
ν0 0.12 - ± 3.0%
Yield surface ˜ β1 140.0 MPa ± 14.2% ˜ β2 -5.5 MPa ± 7.1% ˜ φ 0.6 - ± 7.1% ˜ ψ 0.16 - ± 7.1%
R 5.7 - ± 0%
Damage c t 10 0 0 - ± 15%
c c 55 - ± 20%
c s 55 - ± 36%
w
E
e
G
T
m
m
m
e
a
w
f
p
t
e
I
t
b
S
S
w
s
a
c
w
d
e
b
fi
m
ee by examining Eqs. (34) and (35) that this expression is read-
ly available, since F( ζ) ≡ F( ξ − y c 1 ) . Performing the substitution
f p ξ − y c for p ζ provides the alternative but equivalent expression
f the DP-C yield criteria, which we call a Drucker-Prager/Damage
DP-D) yield criteria:
F := ‖ dev [
˜ ζ]‖
2 − F φ
cap
(A
φ ˜ β1 − B
φ ˜ p ζ)2
= ‖ dev
[ ˜ ξ] ‖
2 − F φ
cap
(A
φ ˜ β1 − B
φ( p ξ − y c ) )2 ≤ 0 ,
(37)
here,
φ :=
2
√
6 cos ˜ φ
3 + r sin
˜ φ, B
φ :=
2
√
6 sin
˜ φ
3 + r sin
˜ φ, −1 ≤ r ≤ 1 , (38a)
φcap := 1 − 〈 β2 − 3( p ξ − y c ) 〉 ˜ β2 − 3( p ξ − y c )
(X
φ − ˜ β2 ) 2 , (38b)
β2 − 3( p ξ − y c ) 〉 :=
1
2
[| β2 − 3( p ξ − y c ) | + ( β2 − 3( p ξ − y c )) ],
(38c)
φ :=
˜ β2 − R
(A
φ ˜ β1 − B
φ ˜ β2
). (38d)
The yield function is defined in damage-effective stress space
see Fig. 2 ). The material parameters of the yield function are
ence damage-effective parameters, and are so denoted with tilde
otation. The parameters ˜ φ and
˜ β1 are the (damage-effective) fric-
ion angle and the cohesion parameter, respectively. The parame-
er ˜ β2 is associated with the preconsolidation stress, the position
f the cap along the ˜ I ζ1
= 3 p ζ axis being given by X
φ . The parame-
er R controls the ellipticity of the cap, and the shape of the yield
urface on the octahedral plane is controlled by −1 ≤ r ≤ 1 , such
hat r = 1 and r = −1 coincide with the intersection of the triaxial
xtension (TE) and triaxial compression (TC) corners of the Mohr–
oulomb yield surface, respectively. The Frobenius norm is de-
oted by ‖ ( •) ‖ , |( · )| is the absolute value, and 〈 · 〉 is the Macaulay
racket. For non-associative plasticity, we have similar functional
orm for the plastic potential function G p as for the yield func-
ion F , the only difference being that the damage-effective di-
atation angle ˜ ψ replaces the damage-effective friction angle ˜ φ in
q. (37) such that
p := ‖ dev
[ ˜ ξ] ‖
2 − F ψ
cap
(A
ψ ˜ β1 − B
ψ ( p ξ − y c ) )2
, (39)
here the superscripts denote ˜ ψ is used in place of ˜ φ in
qs. (38a) –(38d) .
The shear damage potential is chosen to be similar to the gen-
ral form proposed by Lemaitre and Chaboche (1994) ,
s :=
Sω
s
c α(s + 1)
(−c αω
c y s
S
)s +1
. (40)
he positive material parameters S and s can generally be deter-
ined from calibration with measurements for various types of
aterials (e.g., as described in Lemaitre and Chaboche, 1994 ). The
aterial parameter c α is added here to the expression in consid-
ration of keeping proportionality with the comprehensive dam-
ge rate ( Eq. (19 )). The form of Eq. (40) is convenient because it is
idely used and has been shown to be applicable to many dif-
erent types of materials under varying loading conditions with
roper choice of the material parameters. In adapting this equa-
ion for the modeling of damage in porous rock, it is necessary to
xamine the choice of material parameters carefully here as well.
n particular, two distinctive damage material behaviors of geoma-
erials are considered: (i) pressure dependence, and (ii) differences
etween dilative and compressive regimes. In order to address (i),
is taken to be pressure dependent by
:= c −p/p c s , (41)
here c s is a positive material constant, p is the mean Cauchy
tress, and p c is the (Cauchy) preconsolidation stress. In order to
ddress (ii), c α is defined as
α :=
(1 − 〈 sgn [ e p v ] 〉
)c c + 〈 sgn [ e p v ] 〉 c t , (42)
here c c and c t are positive material constants for compaction and
ilation, respectively. Table 2 lists all model parameters.
Having now explicit expressions for the potential functions, the
xpressions for the evolution equations provided in Section 3 can
e found. For expressing the flow rule, it is convenient to first de-
ne
ˆ :=
∂G p ∂ ξ
=
∂G p
∂ dev [ξ] +
1
3
∂G p ∂ p ξ
1 = 2 ω
c ω
s dev
[ ˜ ξ] + ω
c C ψ 1 , (43)
224 K.C. Bennett, R.I. Borja / International Journal of Solids and Structures 143 (2018) 218–231
h
m
T
d
w
q
A
w
5
o
c
l
b
d
V
D
a
i
e
s
p
a
T
s
p
p
a
s
g
A
n
i
i
F
a
5
t
T
m
c
t
d
h
w
where
C ψ := 2(A
ψ ˜ β1 − B
ψ ˜ p ζ )
[
F ψ
cap B
ψ
3
− (A
ψ ˜ β1 − B
ψ ˜ p ζ ) 〈 β2 − 3
p ζ 〉 (X
ψ − ˜ β2 ) 2
]
.
(44)
The flow rule of Eq. (21) is then expressed as
d
p =
˙ λ ˆ m . (45)
The deviatoric and volumetric parts of the plastic deformation rate
can be distinguished as
dev [d
p ]
= 2
λω
c ω
s dev [
˜ ζ], vol
[d
p ]
= 3
λω
c C ψ . (46)
Similarly, the damage rates of Eqs. (19) and (20) are found to be
expressed by
˙ d
c = 3 c α ˙ λC ψ , ˙ d
s =
˙ λω
s ω
c
(−c αω
c y s
S
)s
. (47)
The intermediate equations necessary for finding these solutions
are provided in Appendix A .
4.3. Implicit integration
Backward Euler implicit integration making use of the exponen-
tial map is utilized in order to obtain the integrated flow rule in
terms of the Hencky strain (cf. Simo, 1992; 1998 , among others).
This allows the classical predictor/corrector return mapping algo-
rithm of the classical small strain theory to be preserved (cf. Borja
and Tamagnini, 1998; de Souza Neto et al., 2011 , among others).
The detailed procedure involving the exponential map is not re-
peated here, as it has been well documented in the literature, and
we refer also to the details provided in Bennett et al. (2016) for a
description of the kinematics relevant to the current work.
The integrated flow rule is hence given by
ε
p n +1
= ε
p n + �λ ˆ m n +1 , (48)
which provides the elastic strain at a given increment of total
strain �ε (i.e., the predictor/corrector equation),
ε
e n +1 = ε
e,tr n +1
− �λ ˆ m n +1 , (49)
where the superscript ( •) tr denotes a trial value, and is for the elas-
tic strain taken to be at each time step as ε e,tr n +1
= ε e n + �ε . The ex-
pressions for the volumetric and deviatoric elastic strains can also
be respectively expressed by
e e v ,n +1 = e e,tr v ,n +1
− 3�λω
c n +1 C
ψ
n +1 , (50)
and
ε
e n +1 = ε
e,tr n +1 − 2�λω
c n +1 ω
s n +1 dev
[˜ ζ]. (51)
It is useful also to define
ε e s := ‖ ε
e ‖ , (52)
such that
ε e s,n +1 = ε e,tr s,n +1
− 2�λω
c n +1 ω
s n +1 ‖ dev
[˜ ζ]‖ . (53)
The update equations for the damage variables are found from
Eq. (47) , making use of Eqs. (50) and (53) to find
d
c n +1 = c α
d
c n + �e p v
1 + �e p v ; d
s n +1 = d
s n +
3
2
√
6
�ε p s
˜ q ζn +1
(−c αω
c n +1 y
s n +1
S
)s
.
(54)
Having the updated elastic strains allows the updated stress to
be calculated according to the constitutive equation (28) . We note
ere also the update equations for the mean and deviatoric stress
easures. The mean stress update is given by
p ζn +1
= (1 − d
c n +1 ) ·
(K 0 e
e v ,n +1
)= (1 − d
c n +1 ) · K 0
(tr [ε
e,tr n +1
]− 3�λω
c n +1 C
ψ
n +1
)=
˜ p ζ ,tr n +1
/ω
c n +1 − 3 K 0 �λC ψ
n +1 . (55)
he deviatoric stress update is similarly found as
ev [ξn +1
]≡ dev
[ζn +1
]= (1 − d
c n +1 ) · (1 − d
s )2 μ0 ε
e n +1
= (1 − d
c n +1 ) · (1 − d
s )2 μ0
(´ε
e,tr n +1 − 2�λω
c n +1 ω
s n +1 dev
[˜ ζn +1
]).
=
dev
[ ˜ ζ
tr
n +1
] ω
c n +1
ω
s n +1
(1 + 4 μ0 �λω
c n +1
ω
s n +1
) , (56)
hich provides (see also further details provided in Appendix A ),
ζn +1
= (1 − d
c n +1 ) · ˜ q
ζ ,tr n +1
/(1 + 4�λω
c n +1 ω
s n +1 ) . (57)
n iterative solution of the integrated equations is required. In this
ork we employ a Newton–Raphson iteration.
. Comparison of model predictions with measurements
The model is evaluated by calibration and comparison to lab-
ratory and in-situ measurements. Two different materials and
orresponding sets of measurements are considered: (1) Tavel
imestone after Vajdova et al. (2004) , and (2) a horizontal well-
ore through 30% porosity limestone off the coast of Brazil as
escribed by Coelho et al. (2005) . The experimental data of
ajdova et al. (2004) were chosen as a first example for the DP-
model calibration for a number of reasons: it consists of tri-
xial compression (TC) at a wide range of confining pressures,
t includes isotropic compression data, and it is widely refer-
nced as representative for describing the pressure dependent
o called “brittle–ductile transition” and “shear enhanced com-
action/dilation” behavior of rocks ( Rudnicki and Rice, 1975; Borja
nd Regueiro, 2001; Wong and Baud, 2012; Tjioe and Borja, 2015 ).
he wellbore study was chosen because it is accompanied by sub-
tantial site-characterization measurements and is through high
orosity limestone. The model parameters were calibrated inde-
endently to each of the materials and corresponding data sets,
nd the calibration methods along with calibrated values are pre-
ented independently in each of the following sub-sections.
It is noted that the shear stress intercept in q vs. p space is
iven by q ′ = 2 c ′ , where c ′ is the Mohr–Coulomb (MC) intercept.
lthough the MC friction angle is a parameter of the model, it
eeds to be evaluated in damage-effective space to do so. Examin-
ng Eq. (29) , it can be seen that the damage-effective friction angle
s related to the (measured) friction angle by ˜ φ = tan
−1 (ω
s tan φ) .
or each calibration procedure, the initial yield surface is provided
long with the corresponding damage-effective yield surface.
.1. Tavel limestone triaxial compression simulations
The DP-D model was calibrated to the triaxial compression (TC)
est data on Tavel limestone presented by Vajdova et al. (2004) .
he isotropic compression data allowed the damage-effective bulk
odulus K 0 , the preconsolidation stress parameter ˜ β2 and the
ompressive hardening parameter c c to be determined from the
est data (see Fig. 3 ). The TC measurements also allowed for the
etermination of the Mohr-Coulomb (MC) friction angle φ and co-
esion intercept c ′ (corresponding to the β1 parameter), which
ere directly reported by the authors as c ′ = 60 MPa and φ = 27 ◦,
K.C. Bennett, R.I. Borja / International Journal of Solids and Structures 143 (2018) 218–231 225
Fig. 3. Calibration (solid line) of bulk modulus K , preconsolidation pressure p c , and
compression damage parameter c c to isotropic compression test data (symbols) of
Vajdova et al. (2004) .
Fig. 4. Drucker-Prager/Damage initial yield surface calibrated to measurements
showing load path for 100 MPa confining pressure (solid line) and damage-effective
stress load path (dashed line), as well as damage-effective yield surface (dotted).
r
t
i
p
s
i
k
t
c
a
e
S
m
i
i
d
s
h
t
c
i
l
s
s
Table 4
Variation of parameters from those reported in Table 3 required to best fit measure-
ments of Vajdova et al. (2004) .
Confining pressure Parameter(s) Variation
10 MPa K 0 , β2 + 1.4%
ν0 , φ, ˜ ψ + 1.4%
20 MPa ν0 , φ -3.0%
30 MPa K 0 , β2 -1.4%
ν0 , φ, ˜ ψ + 3.0%
50 MPa K 0 , β2 -4.5%
ν0 , φ, ˜ ψ + 2.9%
c t -15%
c s -9.1%
100 MPa K 0 , β2 -7.1%
ν0 , φ, ˜ ψ -7.1%
150 MPa ν0 , φ, ˜ ψ -2.9%
c c + 20%
c s + 36%
200 MPa ν0 , φ, ˜ ψ + 2.9%
240 MPa ν0 , φ, ˜ ψ + 2.9%
Table 5
Least squares linear regressions of curves in Fig. 7 for back calculating α.
Confining pressure Slope Coefficient of determination α
10 MPa 635.649 0.998975 709.430
20 MPa 602.050 0.997152 671.931
30 MPa 647.793 0.997677 722.983
50 MPa 604.175 0.994839 674.302
100 MPa 46.045 0.999199 51.390
150 MPa 60.165 0.994861 67.148
200 MPa 50.458 0.995565 56.315
240 MPa 52.437 0.994642 58.523
Table 6
DP-D model parameters established from mea-
surements of Coelho et al. (2005) .
Type Parameter Value Units
Elastic K 0 571.0 MPa
ν0 0.15 -
Yield Surface ˜ β1 25.0 MPa ˜ β2 -30.0 MPa ˜ φ 0.42 - ˜ ψ 0.1 -
r 0 -
R 2 -
Damage c t 10 -
c c 10 -
c s 40 -
1
t
a
a
c
f
q
u
u
d
d
i
(
V
o
u
d
s
espectively. The damage parameter c t was calibrated to best fit
he TC measurements that were on the dilative side (at low confin-
ng pressure), and the c s parameter calibrated to the high confining
ressure curves. Fig. 4 shows the initial DP-D yield surface along-
ide the damage-effective yield surface. A critical aspect of calibrat-
ng the model was to adjust the ellipticity parameter R in order to
eep the brittle–ductile transition point (peak of the plastic po-
ential surface in p –q space) near the intersection of the 100 MPa
onfining pressure stress paths, consistent with the measurements
nd description provided by the authors (op. cit.). A low damage-
ffective dilation angle was chosen as typical of rock ˜ ψ = 0 . 16 .
ome minor adjustments of parameters were needed to best fit the
easurements, while attempting to keep as many fixed as possible.
A comparison of the model simulations with the measurements
s provided in Fig. 5 . All calibrated model parameters are reported
n Tables 3 , and 4 reports variations of model parameters for in-
ividual tests necessary to best fit the measurements. The results
how that the model is able to simulate well both the distinctive
ardening and softening evident in the measurements, as well as
he pressure dependent transition between these regimes, the so-
alled brittle-ductile transition. We emphasize that this behavior
s modeled as changes in porosity and microfracture through evo-
ution of the damage state variables in a thermodynamically con-
istent way made possible by recognizing the role of the Eshelby
tress tensor as described in Sections 2 and 3 .
Evolution of damage parameters is provided for the 50 MPa and
00 MPa confining pressure simulations in Fig. 6 . These tests mark
he transition from brittle to ductile behavior as described by the
uthors ( op. cit. ) and apparent in Fig. 5 . The comprehensive dam-
ge parameter, d
c , in the 50 MPa simulation transitions from de-
reasing to increasing, which roughly corresponds to the transition
rom hardening to softening apparent in the 50 MPa stress ratio
/ p curve on the right side ( Fig. 6 b); whereas, the 100 MPa sim-
lation predicts only increasing d
c . Note that negative total vol-
me strain is plotted, showing that the 50 MPa specimen is pre-
icted to dilate during shear and the 100 MPa specimen is pre-
icted to contract. Recalling that d
c is proportional to the poros-
ty through the plastic part of volume strain only (see Eqs. (6) and
54) ), the 50 MPa d
c curve exhibits the phenomenon described by
ajdova et al. (2004) (see also Lubarda et al., 1996 , among others)
f an initial decrease in porosity followed by a subsequent increase
nder dilation. The figure shows that the shear damage parameter,
s , is predicted to monotonically increase in both simulations, but
ignificantly more in the 50 MPa simulation, where softening is ev-
226 K.C. Bennett, R.I. Borja / International Journal of Solids and Structures 143 (2018) 218–231
Fig. 5. Comparison of model simulations (solid lines) with measurements (symbols) of Tavel limestone in triaxial compression ( Vajdova et al., 2004 ) with confining pressures
indicated on the figures.
Fig. 6. Model predictions of change in damage during simulated TC at confining pressures of 50 MPa and 100 MPa: (a) change in comprehensive damage, �d c , alongside
(negative) volume strain, e v , and (b) change in shear damage, d s , alongside (negative) stress ratio, q / p .
Fig. 7. Model predictions with parameters calibrated to measurements of
Vajdova et al. (2004) for relation between changes in d c and plastic volumetric de-
formation as described in Eq. (8) . The slope of each line corresponds to α(1 − n 0 ) =
0 . 896 α in Eq. (8) , and linear regression values are reported in Table 5 .
Fig. 8. Plane strain finite element model mesh and boundary conditions for bore-
hole simulation.
ident. We emphasize that both damage parameters contribute to
the overall amount of damage as apparent for example in Eq. (28) .
The evolution of d
c for all simulations is shown in Fig. 7 plotted
against plastic volumetric deformation as 1 − 1 /J p . This plot corre-
K.C. Bennett, R.I. Borja / International Journal of Solids and Structures 143 (2018) 218–231 227
Fig. 9. Model predictions of stress distribution around borehole.
s
α
a
i
d
T
d
t
A
n
t
c
r
a
t
d
o
t
c
e
p
f
m
v
t
c
r
d
y
r
t
c
d
i
s
a
i
w
s
p
5
ponds to the relation of Eq. (8) , where the slope of each curve is
(1 − n 0 ) . The average initial porosity was reported by the authors
s 10.4%, allowing for the model predictions of the proportional-
ty factor α at each confining pressure to be calculated. The pre-
icted linearized values for α calculated in this way are reported in
able 5 . Notably, α is predicted to increase approximately by an or-
er of magnitude on the dilative (brittle) side of the brittle-ductile
ransition (i.e, for the 10, 20, 30, and 50 MPa confining pressures).
s can be seen in Fig. 7 and also by the coefficient of determi-
ation for the least squares linear regression reported in Table 5 ,
he model predicts a near-linear relationship in Eq. (8) (i.e., near-
onstant alpha) at any specific confining pressure. Furthermore, the
esults suggest that the proportionality between volumetric dam-
ge and porosity ( α) is binary, with approximately constant dis-
inct values for dilative and compactive regimes.
It should be emphasized that the material model considers only
uctile damage and does not address the onset of brittle fracture
bserved by the authors at low confining pressures and evident in
he measurements of Fig. 5 (a) as the cataclysmic failure of the low
onfining pressure (below 100 MPa in this case) specimens. How-
ver, the dilative volumetric damage predicted at low confining
ressures may be considered to be indicative of such a cataclysmic
ailure, i.e., comparison of the model predictions to the measure-
rents suggests that the threshold for an increase in d
c may be
ery low for this material. As is evidenced by the comparison with
he measurements and is perhaps what we may intuitively expect,
onsistent with a description of the rock as being able to sustain
elatively little shear induced dilation in comparison to shear in-
uced compaction.
It should also be pointed out that shear band bifurcation anal-
sis was not conducted on the specimen responses. The softening
esponses predicted by the constitutive model resulted from ma-
erial softening, and not from localized deformation. Evidently, lo-
alized deformation in the form of a deformation band could have
ominated most of the softening responses exhibited by the spec-
mens. However, material and geometric imperfections on these
pecimens were so pervasive that any bifurcation analysis from
n initially homogeneous response would not be meaningful, since
nitial specimen responses were not truly homogeneous to begin
ith. Nevertheless, we have checked the constitutive tangent ten-
or for the 10 MPa simulation and observed that this tensor lost
ositive-definiteness right at the onset of plasticity.
.2. Simulated borehole through limestone
The model was implemented in an Abaqus UMAT material sub-
outine. A horizontal borehole in a porous limestone reservoir off
228 K.C. Bennett, R.I. Borja / International Journal of Solids and Structures 143 (2018) 218–231
Fig. 10. Model predictions of damage around borehole.
c
S
a
t
c
p
a
t
fl
6
c
i
c
n
t
s
r
t
B
f
d
t
d
h
p
m
m
f
d
p
c
t
d
t
t
n
T
b
the coast of Brazil was simulated in order to examine the model
predictions of stress and damage around the borehole. The ma-
terial properties from laboratory measurements and the far-field
stress field from in situ measurements, along with a more detailed
description of the reservoir are reported by Coelho et al. (2005) .
The elastic material properties were taken from those reported
by Coelho et al. (2005) , as were the MC friction angle, cohesion
intercept, and yield point on the hydrostatic axis ( X
φ determin-
ing ˜ β1 and
˜ β2 ). The intermediate principal stress model param-
eter r was taken as the median value of zero, and the elliptic-
ity parameter R was assigned a typical value of 2. The only re-
maining model parameters that needed to be calibrated were the
damage parameters, c t , c c , and c s , which were kept equivalent and
calibrated to best mimic the Drucker-Prager/Cap model simulated
stress field results of Spiezia et al. (2016) , while reaching final
damage states within reasonable threshold values (i.e., to not pre-
dict breakout at the walls, which is consistent with the observa-
tions). The mean stress value for evaluating Eq. (41) was taken as
the average mean stress value of 18 MPa. All model parameters are
reported in Table 6 .
The borehole radius is 4.25 in. = 107.95 mm. The in-situ geo-
static stresses were reported by Coelho et al. (2005) as vertical
stress σv = 32.1 MPa and major and minor horizontal stresses, re-
spectively, σH = σh = 9.0 MPa. The finite element model was con-
structed as plane strain, making use of the two-fold symmetry (see
Fig. 8 ). The analysis simulated the drilling of the borehole by ap-
plying the far-field stresses and pressure at the borehole wall si-
multaneously.
Fig. 9 shows the predicted stress distribution around the bore-
hole with (excess) pressure on the borehole wall absent. Fig. 10
shows the corresponding model predictions of damage distribu-
tion. Fig. 10 (a) plots the volumetric damage parameter d
c , and
Fig. 10 (b) plots the shear damage parameter d
s . Localization of
both volumetric and shear damage occur at the borehole wall
in the orientation coinciding with the direction of least principal
stress, but with opposite trends, i.e., shear damage accompanied by
volumetric hardening (healing). In other words, the simulation pre-
dicts shear induced compaction in this region, where overall hard-
ening is accompanied by microfracture. This is indicative of the
well known tendency for “borehole breakout,” where stress con-
centrations at the borehole wall cause microfracture and eventually
collapse of sections of the wall, effectively elongating the borehole
in that direction (cf. Read and Martin, 1996; Zoback, 2010 ). This is
uonsistent also with the Drucker-Prager/cap model predictions of
piezia et al. (2016) , who predicted similar trends of volumetric
nd shear strain localization around the borehole wall (although
heir model did not include damage). We emphasize that the in-
lusion of volumetric and shear damage in the model not only
rovides for a prediction of damage around the borehole wall, but
lso provides for a prediction of the stress field associated with
he state of damage, i.e., the stress field prediction of Fig. 9 is in-
uenced by the damage predicted in Fig. 10 .
. Conclusion
Relations between inelastic volume change (bulk plasticity),
hanges in porosity, and evolution of damage have been exam-
ned and incorporated into a unified hyper-elastoplasticity and
ontinuum damage framework. This has allowed for a thermody-
amically consistent description of the damage of porous rocks
hat separates volumetric and isochoric contributions to damage,
uch that the isochoric damage part describes distributed microc-
acks exclusive of their associated apertures. An Eshelby-like stress
ensor we call the Eshelby-zeta stress, which was identified by
ennett et al. (2016) as being energy-conjugate to the plastic de-
ormation rate, has been shown herein to play a valuable role in
istinguishing between volumetric and isochoric contributions to
he overall damaged state of the material while ensuring thermo-
ynamic consistency.
This theory has been developed and implemented in a novel
yper-elastoplastic damage constitutive model appropriate for
orous rocks, what we call a Drucker-Prager/Damage (DP-D)
odel. The model is based on a Drucker-Prager/Cap plasticity
odel previously presented by Regueiro and Ebrahimi (2010) and
urther modified by Bennett et al. (2016) , but the addition of
amage and associated damage potentials is novel. Furthermore,
reviously phenomenological hardening/softening rules are ex-
luded, such that hardening/softening behavior is modeled en-
irely through the described damage evolution making use of the
amage-effective stress concept. The DP-D model is provided in
erms of a current configuration representation of finite deforma-
ion hyper-elastoplasticity, and an implicit integration scheme for
umerical implementation has also been provided.
The DP-D model has been applied to simulation of confined
C measurements on Tavel limestone exemplifying the so called
rittle-ductile transition behavior ( Vajdova et al., 2004 ). The sim-
lations show that the model is capable of simulating the pressure
K.C. Bennett, R.I. Borja / International Journal of Solids and Structures 143 (2018) 218–231 229
d
s
c
p
s
a
l
f
a
fi
g
e
a
w
m
a
T
b
a
s
s
s
d
f
r
t
(
t
t
f
p
d
d
p
m
o
s
c
c
c
p
A
m
G
0
w
n
G
A
a
T
s
w
s
T
G
s
m
T
f
A
R
A
A
A
A
A
ependent hardening/softening behavior exhibited in these mea-
urements entirely through damage and damage-effective stress
oncepts related to porosity and microfracture changes and made
ossible by recognizing the role of the Eshelby stress tensor in en-
uring thermodynamic consistency. In particular, marked softening
t low confining pressures is captured well by modeling the evo-
ution of the damaged state with dilative strain. The model results
urther allow for evaluation of the relationship between damage
nd porosity established on kinematic grounds, where an identi-
ed proportionality factor ( α) has been examined. The results sug-
est that for the measurements considered, such proportionality is
ssentially binary, with the factor α increasing by approximately
n order of magnitude between compressive and dilative regimes,
hile remaining approximately constant in each.
The model has been implemented within an Abaqus UMAT
aterial subroutine, and nonlinear finite element simulations of
reservoir borehole through limestone have also been provided.
hese simulations were carried out in order to demonstrate the ro-
ustness of the model and to investigate the predictions of stress
nd damage distribution around the borehole wall. They demon-
trate the usefulness of the model for the analysis of borehole
tability and the prediction of both damaged zones and resulting
tress fields around the borehole wall.
The separation of volumetric and isochoric damage variables
escribed in this work is especially significant because it allows
or the possibility of a thermodynamically consistent tensorial rep-
esentation of isochoric damage to be constructed from the con-
inuum distribution of microcracks without the need to disregard
or otherwise make any simplifying assumptions about) their aper-
ures. Incorporating a second order isochoric damage tensor into
he DP-D model presented here is currently underway as part of
uture work with the goal of providing a description of elasto-
lastic anisotropy that ascribes the relative weakness of the bed-
ing plane observed in many geomaterials to the anisotropy pro-
uced by the microcracks. We note that further applications of the
roposed model could also include extension to coupled hydro-
echanical problems where hydraulic conductivity is dependent
n the damaged state. The phenomenological distinction empha-
ized in this work between the contribution of porosity and mi-
rofracture to damaged could be advantageous in this regard, espe-
ially in distinguishing changes (possibly anisotropic) in hydraulic
onductivity attributed to microfracture from those attributed to
orosity.
cknowledgments
This material is based upon work supported by the U.S. Depart-
ent of Energy , Office of Science, Office of Basic Energy Sciences,
eosciences Research Program, under Award number DE-FG02-
3ER15454 . Support for materials and additional student hours
ere provided by the National Science Foundation under Award
umber CMMI-1462231 , and by the Stanford University James M.
ere Research Fellowship.
ppendix A
The evolution of the damage variables are found from
∂G c ∂y c
=2 c α(A
ψ ˜ β1 − B
ψ ˜ p ζ)[3 〈 β2 − 3
p ζ 〉 (X
ψ − ˜ β2 ) 2
(A
ψ ˜ β1 − B
ψ ˜ p ζ)− F
ψ
cap B
ψ
]= −3 c αC ψ ,
(58)
nd
∂G s ∂y s
=
∂G d ∂y s
= −ω
s ω
c
(−c αω
c y s
S
)s
.
(59)
he detailed solution of q n +1 is provided by first noting that
:= dev [ ζ] =
√
2
3
q n = 2 με
e , (60)
here ˆ n := ε e / ‖ ε e ‖ ≡ ε e,tr / ‖ ε e,tr ‖ =: ˆ n
tr . Eq. (60) also reveals that
/ ‖ s ‖ ≡ ˆ n by,
‖
s ‖ =
√
2
3
˜ q = 2 μ‖ ε
e ‖
‖
s ‖
ε
e
‖ ε
e ‖
=
√
2
3
˜ q ε
e
‖ ε
e ‖
= 2 με
e
‖
s ‖
n =
√
2
3
˜ q n = 2 με
e
⇒ ‖
s ‖
n =
s
⇒
ˆ n =
˜ s
‖
s ‖
=
s
‖ s ‖
.
(61)
he plastic potential function can then be written as
p :=
2
3
˜ q 2 − F ψ
cap
(A
ψ ˜ β1 − B
ψ ˜ p ζ)2 ≤ 0 , (62)
uch that
ˆ :=
∂G p ∂ ξ
=
√
3
2
∂G p ∂q
ˆ n +
1
3
∂G p ∂ p ξ
1 =
2
√
6
3
ω
c ω
s ˜ q n + ω
c C ψ 1 . (63)
he evolution and update equation for ε e s,n +1
and q n +1 are then
ound by noticing
ε
e n +1 = ε e s,n +1 n = ε e,tr
s,n +1 n − �λ
√
3
2
∂G p ∂q
∣∣∣n +1
ˆ n
⇒ ε e s,n +1 = ε e,tr s,n +1
− �λ
√
3
2
∂G p ∂q
∣∣∣n +1
ε e s,n +1 = ε e,tr s,n +1
− �λ2
√
6
3
˜ q n +1 ω
c n +1 ω
s n +1 .
(64)
nd then from
˜ q n +1 = 2
√
3
2
με e s,n +1
=
√
6 μ
(
ε e,tr s,n +1
− �λ
√
3
2
∂G p ∂q
)
=
√
6 με e,tr s,n +1
− 4�λω
c n +1 ω
s n +1 q n +1
⇒
˜ q n +1 =
˜ q tr n +1
/(1 + 4�λω
c n +1 ω
s n +1 ) .
(65)
eferences
bousleiman, Y.N. , Hull, K.L. , Han, Y. , Al-Muntasheri, G. , Hosemann, P. , Parker, S. ,
Howard, C.B. , 2016. The granular and polymer composite nature of kerogen-richshale. Acta Geotech. 11 (3), 573–594 .
l-Rub, R.K.A. , Voyiadjis, G.Z. , 2003. On the coupling of anisotropic damage andplasticity models for ductile materials. Int. J. Solids Struct. 40 (11), 2611–2643 .
ndrade, J.E. , Baker, J.W. , Ellison, K.C. , 2008. Random porosity fields and their influ-ence on the stability of granular media. Int. J. Numer. Anal. Methods Geomech.
32 (10), 1147–1172 .
rson, C. , 2014. Generalized stress variables in continuum damage mechanics. Mech.Res. Commun. 60, 81–84 .
rson, C. , Gatmiri, B. , 2012. Thermo-hydro-mechanical modeling of damage in un-saturated porous media: theoretical framework and numerical study of the EDZ.
Int. J. Numer. Anal. Methods Geomech. 36 (3), 272–306 .
230 K.C. Bennett, R.I. Borja / International Journal of Solids and Structures 143 (2018) 218–231
D
D
E
E
G
G
G
G
H
H
H
H
J
J
J
J
K
K
KL
L
L
L
L
L
L
L
L
Arson, C. , Pereira, J.-M. , 2013. Influence of damage on pore size distribution andpermeability of rocks. Int. J. Numer. Anal. Methods Geomech. 37 (8), 810–831 .
Arson, C. , Vanorio, T. , 2015. Chemomechanical evolution of pore space in carbon-ate microstructures upon dissolution: linking pore geometry to bulk elasticity.
J. Geophys. Res. 120 (10), 6 878–6 894 . Bakhtiary, E. , Xu, H. , Arson, C. , 2014. Probabilistic optimization of a continuum me-
chanics model to predict differential stress-induced damage in claystone. Int. J.Rock Mech. Min. Sci. 68, 136–149 .
Bažant, Z.P. , Caner, F.C. , Carol, I. , Adley, M.D. , Akers, S.A. , 20 0 0. Microplane model m4
for concrete. I: formulation with work-conjugate deviatoric stress. J. Eng. Mech.126 (9), 944–953 .
Bazant, Z.P. , 2010. Can multiscale-multiphysics methods predict softening damageand structural failure? Int. J. Multiscale Comput. Eng. 8 (1) .
Bennett, K.C. , Berla, L.A. , Nix, W.D. , Borja, R.I. , 2015. Instrumented nanoindentationand 3d mechanistic modeling of a shale at multiple scales. Acta Geotech. 10 (1),
1–14 .
Bennett, K.C., Luscher, D.J., Buechler, M.A., Yeager, J.D., 2018. A micromechanicalframework and modified self-consistent homogenization scheme for the ther-
moelasticity of porous bonded-particle assemblies. Int. J. Solids Struct. doi: 10.1016/j.ijsolstr.2018.02.001 .
Bennett, K.C. , Regueiro, R.A. , Borja, R.I. , 2016. Finite strain elastoplasticity consider-ing the Eshelby stress for materials undergoing plastic volume change. Int. J.
Plast. 77, 214–245 .
Bignonnet, F. , Dormieux, L. , Kondo, D. , 2016. A micro-mechanical model for the plas-ticity of porous granular media and link with the cam clay model. Int. J. Plast.
79, 259–274 . Biot, M.A. , 1941. General theory of three-dimensional consolidation. J. Appl. Phys. 12
(2), 155–164 . Borja, R.I. , 2006. On the mechanical energy and effective stress in saturated and
unsaturated porous continua. Int. J. Solids Struct. 43 (6), 1764–1786 .
Borja, R.I. , 2013. Plasticity Modeling & Computation. Springer, Berlin-Heidelberg . Borja, R.I. , Choo, J. , 2016. Cam-Clay plasticity. Part VIII: A constitutive framework
for porous materials with evolving internal structure. Comput. Methods Appl.Mech. Eng. 309, 653–679 .
Borja, R.I. , Lee, S.R. , 1990. Cam-Clay plasticity, Part 1: Implicit integration of elasto–plastic constitutive relations. Comput. Methods Appl. Mech. Eng. 78 (1), 49–72 .
Borja, R.I. , Regueiro, R.A. , 2001. Strain localization in frictional materials exhibit-
ing displacement jumps. Comput. Methods Appl. Mech. Eng. 190 (20–21),2555–2580 .
Borja, R.I. , Tamagnini, C. , 1998. Cam-Clay plasticity Part III: Extension of the in-finitesimal model to include finite strains. Comput. Methods Appl. Mech. Eng.
155 (1–2), 73–95 . Bronkhorst, C.A. , Gray III, G.T. , Addessio, F.L. , Livescu, V. , Bourne, N.K. , McDon-
ald, S.A. , Withers, P.J. , 2016. Response and representation of ductile damage un-
der varying shock loading conditions in tantalum. J. Appl. Phys. 119 (8), 085103 .Brünig, M. , 2004. Eshelby stress tensor in large strain anisotropic damage mechan-
ics. Int. J. Mech. Sci. 46 (12), 1763–1782 . Buechler, M.A. , Luscher, D.J. , 2014. A semi-implicit integration scheme for a com-
bined viscoelastic-damage model of plastic bonded explosives. Int. J. Numer.Methods Eng. 99 (1), 54–78 .
Cazacu, O. , Cristescu, N.D. , Shao, J.F. , Henry, J.P. , 1998. A new anisotropic failure cri-terion for transversely isotropic solids. Mechanics of Cohesive Frict. Mater. 3 (1),
89–103 .
Chang, C.S. , Bennett, K.C. , 2015. Micromechanical modeling for the deformation ofsand with noncoaxiality between the stress and material axes. J. Eng. Mech. 143
(1), C4015001 . Chang, C.S. , Meidani, M. , 2013. Dominant grains network and behavior of sand–silt
mixtures: stress–strain modeling. Int J Numer Anal Methods Geomech 37 (15),2563–2589 .
Chazallon, C. , Hicher, P.Y. , 1998. A constitutive model coupling elastoplasticity and
damage for cohesive-frictional materials. Mech. Cohesive-frictional Materials 3(1), 41–63 .
Chen, J.-S. , Wu, C.-T. , Belytschko, T. , 20 0 0. Regularization of material instabilities bymeshfree approximations with intrinsic length scales. Int. J. Numer. Methods
Eng. 47 (7), 1303–1322 . Chen, L. , Shao, J.-F. , Huang, H.W. , 2010. Coupled elastoplastic damage modeling of
anisotropic rocks. Comput. Geotech. 37 (1–2), 187–194 .
Chiarelli, A.-S. , Shao, J.-F. , Hoteit, N. , 2003. Modeling of elastoplastic damage behav-ior of a claystone. Int. J. Plast. 19 (1), 23–45 .
Choo, J. , Borja, R.I. , 2015. Stabilized mixed finite elements for deformable porousmedia with double porosity. Comput. Methods Appl. Mech. Eng. 293, 131–154 .
Clayton, J.D. , 2006. Continuum multiscale modeling of finite deformation plastic-ity and anisotropic damage in polycrystals. Theor. Appl. Fract. Mech. 45 (3),
163–185 .
Clayton, J.D. , Knap, J. , 2015. Phase field modeling of directional fracture inanisotropic polycrystals. Comput. Mater. Sci 98, 158–169 .
Clayton, J.D. , Tonge, A.L. , 2015. A nonlinear anisotropic elastic–inelastic constitutivemodel for polycrystalline ceramics and minerals with application to boron car-
bide. Int. J. Solids Struct. 64, 191–207 . Coelho, L.C. , Soares, A.C. , Ebecken, N.F.F. , Alves, J.L.D. , Landau, L. , 2005. The impact
of constitutive modeling of porous rocks on 2-d wellbore stability analysis. J.
Petrol. Sci. Eng. 46 (1–2), 81–100 . Cuss, R.J. , Rutter, E.H. , Holloway, R.F. , 2003. The application of critical state soil me-
chanics to the mechanical behaviour of porous sandstones. Int. J. Rock Mech.Min. Sci. 40 (6), 847–862 .
avison, L. , Stevens, A.L. , Kipp, M.E. , 1977. Theory of spall damage accumulation inductile metals. J. Mech. Phys. Solids 25 (1), 11–28 .
rucker, D.C. , 1957. Soil mechanics and work-hardening theories of plasticity. Trans.ASCE 122, 338–346 .
shelby, J.D. , 1951. The force on an elastic singularity. Philos. Trans. R. Soc. Lond. A244 (877), 87–112 .
shelby, J.D. , 1956. The continuum theory of lattice defects. In: Solid State Physics,3. Elsevier, pp. 79–144 .
Eshelby, J.D. , 1975. The elastic energy-momentum tensor. J. Elast. 5 (3–4), 321–335 .
Fossum, A.F. , Fredrich, J.T. , et al. , 20 0 0. Cap plasticity models and compactive anddilatant pre-failure deformation. 4th North American Rock Mechanics Sympo-
sium. American Rock Mechanics Association . ao, H. , Huang, Y. , 2003. Geometrically necessary dislocation and size-dependent
plasticity. Scr. Mater. 48 (2), 113–118 . rgic, D. , 2016. Constitutive modelling of the elastic–plastic, viscoplastic and dam-
age behaviour of hard porous rocks within the unified theory of inelastic flow.
Acta Geotech. 11 (1), 95–126 . uéry, A .A .-C. , Cormery, F. , Shao, J.-F. , Kondo, D. , 2008. A micromechanical model of
elastoplastic and damage behavior of a cohesive geomaterial. Int. J. Solids Struct.45 (5), 1406–1429 .
urson, A.L. , 1977. Continuum theory of ductile rupture by void nucleation andgrowth: part i?yield criteria and flow rules for porous ductile media. J. Eng.
Mater. Technol. 99 (1), 2–15 .
ajiabdolmajid, V. , Kaiser, P.K. , Martin, C.D. , 2002. Modelling brittle failure of rock.Int. J. Rock Mech. Min. Sci. 39 (6), 731–741 .
Hamiel, Y. , Lyakhovsky, V. , Agnon, A. , 2004. Coupled evolution of damage and poros-ity in poroelastic media: theory and applications to deformation of porous
rocks. Geophys. J. Int. 156 (3), 701–713 . eyden, S. , Li, B. , Weinberg, K. , Conti, S. , Ortiz, M. , 2015. A micromechanical damage
and fracture model for polymers based on fractional strain-gradient elasticity. J.
Mech. Phys. Solids 74, 175–195 . Hill, R. , 1963. Elastic properties of reinforced solids: some theoretical principles. J.
Mech. Phys. Solids 11 (5), 357–372 . u, M. , Wang, Y. , Rutqvist, J. , 2017. Fully coupled hydro-mechanical numerical man-
ifold modeling of porous rock with dominant fractures. Acta Geotech. 12 (2),231–252 .
ueckel, T. , 1976. Coupling of elastic and plastic deformations of bulk solids. Mec-
canica 11 (4), 227–235 . iang, J. , Ling, H.I. , Kaliakin, V.N. , Zeng, X. , Hung, C. , 2017. Evaluation of an
anisotropic elastoplastic–viscoplastic bounding surface model for clays. ActaGeotech. 12 (2), 335–348 .
ohri, M. , Dunham, E.M. , Zoback, M.D. , Fang, Z. , 2014. Predicting fault damage zonesby modeling dynamic rupture propagation and comparison with field observa-
tions. J. Geophys. Res. 119 (2), 1251–1272 .
u, J.W. , 1989. On energy-based coupled elastoplastic damage theories: constitutivemodeling and computational aspects. Int. J. Solids Struct. 25 (7), 803–833 .
u, J.W. , Yuan, K.Y. , Kuo, A.W. , 2012. Novel strain energy based coupled elastoplasticdamage and healing models for geomaterials–part i: formulations. Int. J. Dam-
age Mech. 21 (4), 525–549 . Kachanov, L.M. , 1958. Time of the rupture process under creep conditions, izy akad.
sz. Akad. Nauk. SSR Otd. Tech. Nauk. 8, 26–31 . en-Ichi, K. , 1984. Distribution of directional data and fabric tensors. Int. J. Eng. Sci.
22 (2), 149–164 .
ondo, D. , Zhu, Q. , Monchiet, V. , Shao, J.-F. , 2010. Non-linear micro-cracked geoma-terials: anisotropic damage and coupling with plasticity. In: Multiscale Model-
ing of Heterogenous Materials: From Microstructure to Macro-scale P’roperties,pp. 177–201 .
rajcinovic, D. , 1996. Damage Mechanics, 41. Elsevier . e Pense, S. , Arson, C. , Pouya, A. , 2016. A fully coupled damage-plasticity model for
unsaturated geomaterials accounting for the ductile–brittle transition in drying
clayey soils. Int. J. Solids Struct. 91, 102–114 . ebensohn, R.A. , Escobedo, J.P. , Cerreta, E.K. , Dennis-Koller, D. , Bronkhorst, C.A. ,
Bingert, J.F. , 2013. Modeling void growth in polycrystalline materials. Acta Mater.61 (18), 6918–6932 .
ee, J. , Fenves, G.L. , 1998. Plastic-damage model for cyclic loading of concrete struc-tures. J. Eng. Mech. 124 (8), 892–900 .
emaitre, J. , 1985. A continuous damage mechanics model for ductile fracture. J.
Eng. Mater. Technol. 107 (1), 83–89 . emaitre, J. , 1985. Coupled elasto-plasticity and damage constitutive equations.
Comput. Methods Appl. Mech. Eng. 51 (1–3), 31–49 . Lemaitre, J. , Chaboche, J.-L. , 1994. Mechanics of Solid Materials. Cambridge Univer-
sity Press . emaitre, J. , Desmorat, R. , 2005. Engineering Damage Mechanics: Ductile, Creep, Fa-
tigue and Brittle failures. Springer Science & Business Media .
i, B. , Pandolfi, A. , Ortiz, M. , 2015. Material-point erosion simulation of dynamicfragmentation of metals. Mech. Mater. 80, 288–297 .
iu, X.S. , Ning, J.G. , Tan, Y.L. , Gu, Q.H. , 2016. Damage constitutive model based onenergy dissipation for intact rock subjected to cyclic loading. Int. J. Rock Mech.
Min. Sci. 85, 27–32 . Lubarda, V.A. , Krajcinovic, D. , 1993. Damage tensors and the crack density distribu-
tion. Int. J. Solids Struct. 30 (20), 2859–2877 .
ubarda, V.A. , Mastilovic, S. , Knap, J. , 1996. Brittle-ductile transition in porous rocksby cap model. J. Eng. Mech. 122 (7), 633–642 .
Maugin, G.A. , 1994. Eshelby stress in elastoplasticity and ductile fracture. Int. J. Plast.10 (4), 393–408 .
K.C. Bennett, R.I. Borja / International Journal of Solids and Structures 143 (2018) 218–231 231
M
M
N
N
O
P
R
R
R
R
R
S
S
S
S
S
S
S
S
S
d
S
T
T
T
T
T
T
V
V
V
W
W
W
W
W
Z
Z
Z
Z
Z
Z
isra, A. , Poorsolhjouy, P. , 2015. Granular micromechanics model for damage andplasticity of cementitious materials based upon thermomechanics. Math. Mech.
Solids . 1081286515576821 isra, A. , Poorsolhjouy, P. , 2015. Micro-macro scale instability in 2d regular granular
assemblies. Continuum. Mech. Thermodyn. 27 (1–2), 63–82 . guyen, L.D. , Fatahi, B. , Khabbaz, H. , 2014. A constitutive model for cemented clays
capturing cementation degradation. Int. J. Plast. 56, 1–18 . ova, R. , 1980. The failure of transversely isotropic rocks in triaxial compression.
Int. J. Rock Mech. Min. Sci. Geomech. Abstr. 17, 325–332 . Elsevier
rtiz, M. , 1985. A constitutive theory for the inelastic behavior of concrete. Mech.Mater. 4 (1), 67–93 .
arisio, F. , Samat, S. , Laloui, L. , 2015. Constitutive analysis of shale: a coupled dam-age plasticity approach. Int. J. Solids Struct. 75, 88–98 .
ead, R.S. , Martin, C.D. , 1996. Technical Summary of AECL’s Mine-by ExperimentPhase I: Excavation Response. Technical Report. Atomic Energy of Canada Ltd. .
egueiro, R.A. , Ebrahimi, D. , 2010. Implicit dynamic three-dimensional finite ele-
ment analysis of an inelastic biphasic mixture at finite strain: part 1: appli-cation to a simple geomaterial. Comput. Methods Appl. Mech. Eng. 199 (29–32),
2024–2049 . icard, Y. , Bercovici, D. , 2003. Two-phase damage theory and crustal rock failure:
the theoretical ‘void’ limit, and the prediction of experimental data. Geophys. J.Int. 155 (3), 1057–1064 .
ice, J.R. , 1968. A path independent integral and the approximate analysis of strain
concentration by notches and cracks. J. Appl. Mech. 35 (2), 379–386 . udnicki, J.W. , Rice, J.R. , 1975. Conditions for the localization of deformation in pres-
sure-sensitive dilatant materials. J. Mech. Phys. Solids 23 (6), 371–394 . alari, M.R. , Saeb, S. , Willam, K.J. , Patchet, S.J. , Carrasco, R.C. , 2004. A coupled elasto-
plastic damage model for geomaterials. Comput. Methods Appl. Mech. Eng. 193(27–29), 2625–2643 .
emnani, S.J. , Borja, R.I. , 2017. Quantifying the heterogeneity of shale through statis-
tical combination of imaging across scales. Acta Geotech. 12 (6), 1193–1205 . emnani, S.J. , White, J.A. , Borja, R.I. , 2016. Thermoplasticity and strain localization
in transversely isotropic materials based on anisotropic critical state plasticity.Int. J. Numer. Anal. Methods Geomech. 40 (18), 2423–2449 .
hen, W.Q. , Shao, J.-F. , Kondo, D. , Gatmiri, B. , 2012. A micro–macro model for clayeyrocks with a plastic compressible porous matrix. Int. J. Plast. 36, 64–85 .
hojaei, A. , Taleghani, A.D. , Li, G. , 2014. A continuum damage failure model for hy-
draulic fracturing of porous rocks. Int. J. Plast. 59, 199–212 . imo, J.C. , 1992. Algorithms for static and dynamic multiplicative plasticity that pre-
serve the classical return mapping schemes of the infinitesimal theory. Comput.Methods Appl. Mech. Eng. 99 (1), 61–112 .
imo, J.C. , 1998. Numerical analysis and simulation of plasticity. In: Handbook ofNumerical Analysis, 6, pp. 183–499 .
imo, J.C. , Ju, J.W. , 1987. Strain-and stress-based continuum damage models?II. Com-
putational aspects. Int. J. Solids Struct. 23 (7), 841–869 . imo, J.C. , Meschke, G. , 1993. A new class of algorithms for classical plasticity
extended to finite strains. application to geomaterials. Comput. Mech. 11 (4),253–278 .
e Souza Neto, E.A. , Peric, D. , Owen, D.R.J. , 2011. Computational methods for plas-ticity: Theory and applications. John Wiley & Sons .
piezia, N. , Salomoni, V.A. , Majorana, C.E. , 2016. Plasticity and strain localizationaround a horizontal wellbore drilled through a porous rock formation. Int. J.
Plast. 78, 114–144 .
amagnini, C. , Ciantia, M.O. , 2016. Plasticity with generalized hardening: constitutivemodeling and computational aspects. Acta Geotech. 11 (3), 595–623 .
erzaghi, K. , 1944. Theoretical Soil Mechanics. Chapman And Hali, Limited JohnWiler And Sons, Inc, New York .
jioe, M. , Borja, R.I. , 2015. On the pore-scale mechanisms leading to brittle and duc-tile deformation behavior of crystalline rocks. Int. J. Numer. Anal. Methods Ge-
omech. 39 (11), 1165–1187 . jioe, M. , Borja, R.I. , 2016. Pore-scale modeling of deformation and shear band bifur-
cation in porous crystalline rocks. Int. J. Numer. Methods Eng. 108 (3), 183–212 .
ran-Manh, H. , Sulem, J. , Subrin, D. , 2016. Progressive degradation of rock propertiesand time-dependent behavior of deep tunnels. Acta Geotech. 11 (3), 693–711 .
ruesdell, C. , Noll, W. , 2004. The non-linear field theories of mechanics. In: TheNon-Linear Field Theories of Mechanics. Springer, pp. 1–579 .
ajdova, V. , Baud, P. , Wong, T.-f. , 2004. Compaction, dilatancy, and failure in porouscarbonate rocks. J. Geophys Res. 109 (B5) .
erhoosel, C.V. , Scott, M.A. , Hughes, T.J.R. , De Borst, R. , 2011. An isogeometric anal-
ysis approach to gradient damage models. Int. J. Numer. Methods Eng. 86 (1),115–134 .
oyiadjis, G.Z. , Yousef, M.A. , Kattan, P.I. , 2012. New tensors for anisotropic damagein continuum damage mechanics. J. Eng. Mater. Technol. 134 (2), 021015 .
ang, K. , Sun, W. , 2016. A semi-implicit discrete-continuum coupling method forporous media based on the effective stress principle at finite strain. Comput.
Methods Appl. Mech. Eng. 304, 546–583 .
ang, R. , Fu, P. , Zhang, J.-M. , Dafalias, Y.F. , 2016. Dem study of fabric features gov-erning undrained post-liquefaction shear deformation of sand. Acta Geotech. 11
(6), 1321–1337 . eed, D.A. , Foster, C.D. , Motamedi, M.H. , 2017. A robust numerical framework for
simulating localized failure and fracture propagation in frictional materials. ActaGeotech. 12 (2), 253–275 .
hite, J.A. , 2014. Anisotropic damage of rock joints during cyclic loading: consti-
tutive framework and numerical integration. Int. J. Numer. Anal. Methods Ge-omech. 38 (10), 1036–1057 .
ong, T.-f. , Baud, P. , 2012. The brittle-ductile transition in porous rock: a review. J.Struct. Geol. 44, 25–53 .
hang, X. , Krischok, A. , Linder, C. , 2016. A variational framework to model diffu-sion induced large plastic deformation and phase field fracture during initial
two-phase lithiation of silicon electrodes. Comput. Methods Appl. Mech. Eng.
312, 51–77 . hu, C. , Arson, C. , 2015. A model of damage and healing coupling halite ther-
mo-mechanical behavior to microstructure evolution. Geotech. Geol. Eng. 33 (2),389–410 .
hu, Q. , Kondo, D. , Shao, J. , Pensee, V. , 2008. Micromechanical modelling ofanisotropic damage in brittle rocks and application. Int. J. Rock Mech. Min. Sci.
45 (4), 467–477 .
hu, Q.-Z. , Shao, J.F. , Kondo, D. , 2011. A micromechanics-based thermodynamic for-mulation of isotropic damage with unilateral and friction effects. Eur. J. Mech.
A/Solids 30 (3), 316–325 . hu, Q.Z. , Shao, J.-F. , Mainguy, M. , 2010. A micromechanics-based elastoplastic dam-
age model for granular materials at low confining pressure. Int. J. Plast. 26 (4),586–602 .
oback, M.D. , 2010. Reservoir Geomechanics. Cambridge University Press .