You are on page 1of 10

International Journal of Engineering Science 49 (2011) 1141–1150

Contents lists available at ScienceDirect

International Journal of Engineering Science


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

A damaged visco-plasticity model for pressure and temperature


sensitive geomaterials
A. Karrech a,⇑, K. Regenauer-Lieb a,b, T. Poulet a,b
a
CSIRO Earth Science and Resource Engineering, 26 Dick Perry Ave., Kensington, WA 6151, Australia
b
Western Australian Geothermal Centre of Excellence, Earth and Environment, School of Mechanical Engineering University of Western
Australia, WA 6009, Australia

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

Article history: This study is primarily intended to describe the behaviour of non-associated geomaterials
Received 14 October 2009 within the framework of thermodynamics. By postulating a free energy and a dissipation
Accepted 5 May 2011 function, a pressure dependent constitutive model is derived. The model is then used to
solve a coupled temperature-displacement problem where shear heating feedback, multi-
ple creep mechanisms and their induced continuum damage are included. The results show
Keywords: that the frictional behaviour of geomaterials in the crust has a high influence on faults ori-
Return-mapping
entations. In addition, the model reveals that damage nucleates in the viscous tempera-
Visco-plasticity
Damage
ture-sensitive lower crust.
Temperature Ó 2011 Elsevier Ltd. All rights reserved.
Geomaterials

1. Introduction

Non-equilibrium thermodynamics has been recognised as a successful tool to understand the behaviour of randomly per-
turbed systems. Several extremal principles such as the reciprocity of Onsager (1931) have been put forward to quantita-
tively describe the evolution of the thermodynamic forces with respect to their dual fluxes. These descriptions allowed a
better understanding of ‘‘self organised’’ processes such as thermal convection, transport with reactive feedbacks or simply
oscillatory mechanical systems. This framework did not attract the attention of the mechanical engineering community until
the contributions of Ziegler (1963) who discussed the principle of maximum dissipation. Constraints on the entropy produc-
tion are still being actively developed (Rajagopal & Srinivasa, 2004), as they offer powerful tools to deduce flow rules in
accordance with the principles of thermodynamics.
Modern constitutive models are generally developed within the framework of thermodynamics which ensures enough
consistency to avoid creation of energy in non reactive solids subjected to external loading. One of the pioneering contribu-
tions, which emphasises the thermodynamics foundation of inelastic constitutive laws, was introduced by Rice (1971). It
represents a progression of the classic plasticity approaches, which assume the existence of a yield function and a flow rule.
This approach extended the normality condition to finite deformation and rate-dependent cases (Hill, 1950). It also extended
the stability postulate of Drucker (1959). Using the same background, Lubliner (1975) showed that a yield function is math-
ematically necessary if the dissipation function is positive and homogeneous of degree one in the internal-variable rates. In a
more general formulation, Lubliner (1978) showed that a viscoplastic potential derives in a similar way if the behaviour is
rate dependent. By defining the dissipation as a function of the internal-variable rates, Ziegler (1977) proposed the orthog-

⇑ Corresponding author.
E-mail address: kar10b@csiro.au (A. Karrech).

0020-7225/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijengsci.2011.05.005
1142 A. Karrech et al. / International Journal of Engineering Science 49 (2011) 1141–1150

onality principle and showed that it is equivalent to the principle of maximum dissipation. Based on this principle, Halphen
and Nguyen (1975) derived the generalised thermodynamic theory of associated materials by justifying the existence of a
plastic potential from which inelastic strain can be derived. The optimisation of entropy to deduce governing equations of
state is more generic than the concept of plasticity.
The above mentioned contributions and many others showed that thermodynamic principles represent a rigorous frame-
work in which constitutive models can be derived. However, these approaches were limited to ‘‘standard generalised mate-
rials’’. Experimental evidence shows that geomaterials, which are characterised by a frictional behaviour, exhibit non-normal
flows. In the absence of a solution similar to the standard thermodynamic approach, researchers generally assume the exis-
tence of inelastic potentials which are different than the selected yield functions to derive flow rules. Recently, a new theory
was introduced by Houlsby and co-authors and (Collins & Houlsby, 1997; Houlsby & Puzrin, 2000; Nguyen & Houlsby, 2007)
to relax this assumption and describe the constitutive behaviour of elasto-plastic non-associated materials (sand, rocks, con-
crete), within the principles of thermodynamics. The main features of the theory are (i) the assumption of a dissipation func-
tion which depends on stress and (ii) the orthogonality principle of Ziegler (1977). The flowchart (1) summarises the main
differences between the framework adapted in the current approach and the classical plasticity theory. Note that Ziegler’s
assumption of maximum entropy production can be relaxed to obtain upper and lower bounds from the thermodynamic
extremal principles (Regenauer-Lieb, Karrech, Chua, Horowitz, & Yuen, 2010). Treating large time and length scales allows
to continue using the second principle of thermodynamics which is becoming a subject of controversy especially in case
of out of equilibrium processes taking place in finite volumes at short time intervals (Carberry et al., 2004; Evans, Cohen,
& Morriss, 1993).

The above two classes of approaches offer a wide range of choices to describe several geomaterials that can be encoun-
tered in nature. In the lithosphere, for instance, the generalised standard approach is suitable to describe the inelastic re-
sponses of materials. Several studies (Regenauer-Lieb, Rosenbaum, & Weinberg, 2008; Regenauer-Lieb, Weinberg, &
Rosenbaum, 2006; Karrech, Regenauer-Lieb, & Poulet, 2010, 2011; Karrech, Seibi, & Duhamel, 2010) show that pressure-
independent yielding combined with dislocation and diffusion creep mechanisms can give appropriate estimation of the
material response. The approach is adequate since olivine, which has a metallic-like behaviour, is a predominant component
especially in the lower lithosphere/mantel layers. However, frictional behaviour prevails in the upper-crust and non-associ-
ated types of flow become more suitable. To the knowledge of the authors, there is no approach in the literature combining
the associated and non-associated behaviour in the same framework, although this type of description can be useful in geo-
sciences. It allows predicting the response of multi-scale structures without the artificial effects of localisation due to explicit
material properties transition.
In this paper, a new pressure dependant model is introduced within the thermodynamic framework of non-associated
materials (Collins & Houlsby, 1997; Houlsby & Puzrin, 2000; Nguyen & Houlsby, 2007) to ensure a combination of the
shear-dominated deformation regime of the lower crust and the volumetric-dominated deformation regime of the upper
crust. Rate dependency is introduced using the ‘‘continuous’’ formulation of persistent yielding (Carosio, Willam, & Etse,
2000; Ponthot, 1995; Voyiadjis & Abed, 2006). In addition, this study includes continuum damage in accordance with the
formulation of Karrech, Regenauer-Lieb, and Poulet (2011).
A. Karrech et al. / International Journal of Engineering Science 49 (2011) 1141–1150 1143

2. Governing equations

2.1. Energy functions

The model introduced in this paper describes the coupled thermo-mechanical behaviour of damaged elastic–viscoplastic
geomaterials. Considering a linear damaged elastic behaviour in case of small perturbations, a classic form of Helmholtz spe-
cific free energy w can be used as follows:
 
1 cx
qwðij ; aij ; T; DÞ ¼ ðij  aij Þð1  DÞC ijkl ðkl  akl Þ  3ð1  DÞK a~ ðii  aii ÞðT  T 0 Þ  ðT  T 0 Þ2 ð1Þ
2 2T 0
where C ijkl ¼ ðK  23 GÞdij dkl þ Gðdik djl þ dil djk Þ; K is the bulk modulus, G is a shear modulus,  is the total strain tensor, a is the
inelastic strain tensor, q is the material density, D is a damage parameter, T0 is a reference temperature, a ~ is the coefficient of
thermal expansion, c is the specific heat capacity, and x ¼ 1  9ð1  DÞK a ~ 2 T 0 =c. The summation of indices is used in this pa-
per and the notation ()0 indicates the deviatoric parts of tensors. The damage parameter D considered above describes the
isotropic degradation of materials. According to Lemaitre (1985) and Chaboche (1987), this parameter can be seen as a sur-
face density of defects. It varies from 0 (virgin material) to 1 (unreachable level representing the fully damaged surface).
Using the first and second laws of thermodynamics, it is possible to identify the Cauchy stress and Lemaitre damage force
as follows:
rij ¼ ð1  DÞKðii  aii Þdij þ 2ð1  DÞGð0ij  a0ij Þ  3ð1  DÞK a~ ðT  T 0 Þdij ð2Þ
"  2 #
r2eq 1 1 rii
Y¼ þ ð3Þ
ð1  DÞ 2 6G 18K req
qffiffiffiffiffiffiffiffiffiffiffiffiffiffi
where req ¼ 32 r0ij r0ij . In classic approaches, the above thermodynamic forces define the generalised stresses by virtue of the
orthogonality postulate (Ziegler, 1977). This postulate will be used in this study only as an ad hoc relationship. The dissipa-
tion function proposed in this paper has the following form:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffi   rffiffiffiffiffiffiffiffiffiffiffiffiffiffi! rffiffiffiffiffiffiffiffiffiffiffiffiffiffi!
2a 2 0 0 jðrii Þ  a a
_ ii 1 2 2 0 0 _ _ c 2 0 0
D¼ a_ a_ þ þ rrii þ / a_ a_ þ FðDÞD þ K #  a_ a_ ð4Þ
1  ~e 3 ij ij bðrii Þ 3r 1  ~e 3 ij ij 1  ~e 3 ij ij

This dissipation function is a combination of five positive terms. The first, which represents the shear dissipation, is similar to
the dissipation function suggested by Lubliner (1975) to derive the Von-Mises criterion where the parameter a plays the role
of a yielding threshold. However, it contains an additional parameter ~e  1 which allows obtaining a hyperbolic behaviour at
small shear stresses. The second term, which represents the dissipation due to volumetric change, is similar to the dissipa-
tion function suggested by Nguyen and Houlsby (2007) except that the involved parameters b and j are dependent on
hydrostatic pressure as will be shown at a later stage. The third term is the dissipation due to rate sensitivity where the po-
tential depends on the equivalent inelastic dissipation. The fourth term, which represents the dissipation due to damage, can
be encountered in several models such as Karrech et al. (2011) and Nguyen and Houlsby (2007). The last term produces no
work but it is commonly used to ensure the compatibility of the volumetric and equivalent shear strains. This type of term
was also used by Lubliner (1975) or Collins and Houlsby (1997) and it represents a constraint to the optimisation problem of
maximum dissipation. The corresponding Lagrange multiplier K will be eliminated by identification of the hydrostatic and
deviatoric stresses. It is important to note that adding a constraint to an optimisation problem certainly restricts the poten-
tial solutions to a subset, however, it does not replace the requirement for an objective function to be maximised. From Eq.
(4), it can be noticed that the dissipation function is by definition homogeneous of order one with respect to its arguments a_
and D._ Therefore, applying Euler’s theorem results in the following equation:

@D @D
D¼ a_ ij þ _ D_ ð5Þ
@ a_ ij @D
_ are thermodynamic velocities by definition, the generalised stresses can be identified from Eq. (5) as follows
Since a_ and D
(see also Houlsby & Puzrin, 2006):
@D @D
vij ¼ _ and vD ¼ _ ð6Þ
@ aij @D
Therefore, deriving the dissipation function (4) with respect to a_ results in the deviatoric and volumetric generalised
stresses:
rffiffiffi
2a þ 2/  Kc 2 a_ ij
0
0
vij ð1 þ ~eÞ ¼ qffiffiffiffiffiffiffiffiffiffi ð7Þ
2 3 a_ 0 a_ 0
ij ij
 
ja 1
vii ¼ þ r rii þ 3K ð8Þ
b r
1144 A. Karrech et al. / International Journal of Engineering Science 49 (2011) 1141–1150
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
By selecting e ¼ ð~e2 þ 2~eÞ=v0ij v0ij , it can be noticed that the norm of deviatoric stress can be written as: vq ¼ 3v0ij v0ij =2 þ e.
Hence, the Lagrange multiplier can be deduced from Eq. (7) as follows:
2ða þ /  vq Þ
K¼ ð9Þ
c
Substituting Eq. (9) into Eq. (8) results in the following expression:
6br
rbðvii  rii Þ ¼ j  a þ ða þ /  vq Þ ð10Þ
c
By selecting 6br = c in the dissipation function, the following result can be obtained:
fp ðvij Þ ¼ vq þ rbðvii  rii Þ  j  / ¼ 0 ð11Þ
_ results in the damage generalised force:
In addition, deriving the dissipation function (4) with respect to D
vD  FðDÞ ¼ 0 ð12Þ
_ trðrÞ and Y (the two latter variables do not
From Eq. (4) it can be seen that the dissipation function is dependent on a_ ; D;
affect the transformation since they are passive in the sense of Legendre–Fenchel formalism). Noticing that D is convex and
homogeneous of degree one in its active arguments, Legendre–Fenchel transformation can be applied to obtain the dual of
the dissipation function:

D ðvij ; vD Þ ¼ sup½vij a_ ij þ vD D_   Dða_ ij ; D_  Þ ¼ Ivij 2S;vD 2D ð13Þ


a_ ij ;D_ 

D ðvij ; vD Þ is identically zero if ðvij ; vD Þ 2 S  D and infinite otherwise. The above mentioned domains are defined as
S  D ¼ fðvij ; vD Þ where vij a_ ij þ vD D_ 6 Dða_ ij ; DÞ;
_ 8ða_ ij ; DÞ
_ 2 R2 g. Maximizing the dissipation under the equality constraints
(11) and (12), results in the following flow rules:

@fp 3 v0ij
a_ ij ¼ kp ¼ kp þ k rbd ð14Þ
@ vij 2 vq p ij
@fD
a_ D ¼ kD ¼ kp ð15Þ
@ vD
where kp and kD are the Lagrange multipliers of the optimisation procedure. These multipliers should be positive so that the
hydrostatic pressure and volume change as well as the thermodynamic force and damage which evolve in the same respec-
tive directions.

2.2. Validity of the dissipation function

The selection of the parameters and functions involved in Eq. (4) play a key role in guaranteeing the validity constitutive
model, but also in predicting the phenomenological aspects that can be encountered in nature. The positiveness of the dis-
sipation function is a necessary and sufficient condition for the non-violation of the second law of thermodynamics (this con-
dition is ensured through the positiveness of all the terms in the dissipation function). The validity of the first terms of Eq. (4)
can be verified by using a yielding threshold of the form a = (1  D)r0. As for the volumetric dissipation term, the parameters
are defined as follows:
 lrr 
ii
jðrii Þ ¼ ð1  DÞ r1  ðr1  r0 Þ exp
lr r 
s ð16Þ
ii
bðrii Þ ¼ exp
s
where s = (r1  r0) and r = (1  D)tan(/d)/3. As a link to the classic plasticity, j(rii) can be seen as a pressure dependent
elasticity envelop, where r1 is the limit of elasticity at high pressure, r0 is the limit of elasticity at low pressure, and rb
can be interpreted as a pressure dependent dilatation angle. Under the above considerations, it can be seen that the volu-
metric dissipation is valid if and only if expðxÞ  1 þ bx ~ is positive 8x 2 R, where b ~ ¼ 1=ðl  lDÞ and x ¼ lrrii . More pre-
s
~ ~
cisely, the volumetric dissipation is positive iff bð1  logðbÞÞ P ðn ~  2Þ=ðn~  1Þ where n ~ ¼ r1 =r0 . For simplicity, the
~ ¼ 1 which guarantees that a positive dissipation is considered in this study.
particular value b
The viscous (third) term of Eq. (4) is quadratic in /, therefore, it results in a positive dissipation. / represents an overstress
which can be calculated by inverting the creep expression:
     
req Q d þ rii V d  kr DV OH Q p þ rH V p  kr DV OH
U½1 ðreq Þ ¼ Ad exp  þ Ap r m
eq exp  ð17Þ
gd RT RT
where the first term represents the diffusion creep mechanism and the second denotes the dislocation creep mechanism. Aq
(q = p, d) are the pre-coefficients of the mechanisms, gd refers to the grains size, Qq are the activation energies, Vq are the
A. Karrech et al. / International Journal of Engineering Science 49 (2011) 1141–1150 1145

activation volumes, VOH is the weakening term for increasing pressure, and m is the exponent of the dislocation creep
mechanism, r and k are fitting parameters of the order 1. These regimes act jointly and induce a multi-mechanism void
growth. Using the upper bound theory, Karrech et al. (2011) showed that the resulting damage growth can be expressed
by D_ ¼ k_p ðð1  DÞm1 þ NðYÞ  1Þ where N(Y) describes the nucleation of defects. Therefore, the dissipation due to damage
can be identified as follows:

FðDÞD_ ¼ Y k_p ðð1  DÞm1 þ NðYÞ  1Þ ð18Þ


This expression already takes advantage of the orthogonality condition of Ziegler (1977) which states that the generalised
thermodynamic force deriving from the dissipation function coincides with the thermodynamic force deriving from Helm-
holtz free energy (vD = Y). Since all the terms of Eq. (18) are positive, the dissipation due to damage is in accordance with the
second principle. Unlike other potentials that can be found in the literature, the proposed function shows that in the absence
of voids nucleation, no damage can increase in a representative volume element. This important detail is the subject of a
controversy raised by Cocks and Ashby (1982) in a review on the limitations of continuum damage mechanics.

3. Numerical implementation

3.1. Prediction correction scheme of stress

The numerical implementation requires an incremental form of the constitutive model. The proposed procedure consists
in starting from a converged state (stress, strain, damage and temperature) and predicting a trial reversible solution as
follows:
rij ¼ r0 þ ð1  DÞC ijkl : Dkl  3ð1  DÞK a~ DTdij ð19Þ

If Dkl is purely elastic then r is the desired solution. However, a correction of the trial stress is necessary if an inelastic

ij
strain takes place. By using the flow rule (14) and taking into account the second order terms (see also Karrech et al.,
2010), the stress increment can be written as follows:
!
 @fp @ 2 fp
Dr ¼ ð1  D  DDÞC ijkl :
ij Dkl  Dkp  kp Drij  3ð1  D  DDÞK a
~ DTdij ð20Þ
@ vkl @ vkl @ rij

By calculating the incremental stress with respect to the updated state in terms of damage and temperature and using the
orthogonality condition of Ziegler (1977) which shows that a converged generalised thermodynamic stress deriving from the
dissipation function coincides with the thermodynamic stress deriving from Helmholtz free energy (vij = rij), it can be de-
duced that
Drij ¼ ð1  DÞC ijkl : ðDkl  Dkp Nkl  kp M ijkl Drij Þ ð21Þ
where D is the current damage parameter, Nkl is the flow orientation @fp/@ vkl, and Mijkl is the partial the derivative of Nkl: @ 2fp/
@ vkl @ rij. The orthogonality condition also applies on Eq. (11) which can be rewritten as follows:
f ðr Þ ¼ req  jðr Þ  / ¼ 0 ð22Þ
ij ii

The above equation is valid as long as an inelastic deformation is taking place. Therefore, if a_ ij – 0, Eq. (22) satisfies the con-
dition kpf_ ¼ 0.

3.2. Consistent tangent modulus with return mapping algorithm

The numerical algorithm used to integrate the constitutive behaviour presented in the last paragraph is based on the re-
turn mapping procedure suggested by Simo and Taylor (1985, 1993). The integration procedure involves the same major
steps as those described by Karrech et al. (2010): (i) The predictor–corrector algorithm by starting from an intermediate con-
verged configuration Cn ðrn ; vn p ; T nþ1 ; Dnþ1 Þ to predict a new one Cnþ1 ðrnþ1 ; vnþ1
p
; T nþ1 ; Dnþ1 Þ, (ii) Newton’s method to evaluate
the Lagrange multiplier D kp and (iii) evaluation of the consistent tangent modulus. The main focus of this paragraph is on
the derivation of the consistent tangent modulus which requires further attention. Based on Eq. (21), the incremental rela-
tionship between stress and strain can be expressed as follows:
Dðrij Þnþ1 ¼ Dijkl ðDkl  Dkp Nij Þ ð23Þ
1
where Dijkl ¼ ð1  DÞðC 1
ijkl þ Dkp M ijkl Þ . The following remark is particularly useful to calculate consistent tangent moduli.

Remark. If Aijkl = a(dikdjl + dildjk) + b dijdkl + cvijvkl, vii = 0 and a – (0, 3b/2, c/2) then Aijkl is invertible and

4aA1 ~
ijkl ¼ ðdik djl þ dil djk Þ þ bdij dkl þ c v ij v kl
~
~ ¼ 2b=ð3b þ 2aÞ and ~c ¼ 2c=ðcv ij v ij þ 2aÞ.
where b
1146 A. Karrech et al. / International Journal of Engineering Science 49 (2011) 1141–1150

Using twice the above remark and the definition of Mijkl, the tensor Dijkl can be calculated as follows:
 
Dijkl ¼ e 2G
K e dij dkl þ Gðd
e ik djl þ dil djk Þ  e
F Nij Nkl ð24Þ
3

where Ke ¼ ð1  DÞK; G e ¼ ð1  DÞG=ð1 þ ð3Dkp =re ÞÞ, and e


F ¼ Dkp ð2GÞ2 =ðreeq þ 2Dkp GÞ. The multiplier Dkp can be obtained
eq
using the inelastic strain flow rule and the condition of persistent yielding ðkf_ ¼ 0Þ as follows:

@ f
@ rij
Dijkl Dkl
Dkp ¼ @ f  ½1
ð25Þ
@ rij
Dijkl @@rgij þ D1t @@Uv p
eq

Noticing that 3GDkp  reeq and Dkpj00 K  1, the Jacobian describing the incremental relationship between stress and total
strain can be expressed as follows:
 
Bij Bkl
Dðrij Þnþ1 ¼ Dijkl  dDkl ð26Þ
Hr

The different terms of Eq. (26) can be expressed as:

Bij ¼ ð1  DÞK j0 ðpÞdij þ 2cGNij


Bkl ¼ ð1  DÞKrb0 ðpÞdkl þ 2cGNkl
ð27Þ
1 @/~ ½1
Hr ¼ ð1  DÞKrb0 ðpÞj0 ðpÞ þ 3cG þ vp
Dt @ eq

where c ¼ ð1  DÞð1  3GDkp =reeq Þ. In the following section, this incremental technique will be applied on a particular case
study in order to verify its applicability.

Fig. 1. Finite element model of the continental cross-section. (a) Dimensions, boundary conditions and loading. (b) Equivalent stress showing the effect of
the random distribution of material properties on the elastic behaviour of the structure.

Table 1
1
Simulation parameters. For both materials, the universal gas constant is R (J mol1 K1) = 8.3144, the weakening term, kr DV OH ¼ 10:6  106 m3 mol and the
activation volume power = 2.0  105 m3 mol1.

Quartz Feldspar
Density, q (kg m3) 3000 2750
Young modulus, E (GPa) 58 71
Poisson’s ratio, E (GPa) 0.3 0.26
Low pressure yield limit, r0 (MPa) 200 300
High pressure yield limit, r1 (MPa) 800 1200
Thermal conductivity, k (W m1 K1) 3.4 3.4
Specific heat, cp (J kg1 K1) 1200 1200
Expansion coefficient, a (K1) 1.2  105 1.2  105
Inelastic heat fraction, n 0.9 0.9
Mechanism’s Exponent, nqa np = 2.5, nd = 1.1 np = 3.2, nd = 1.1
Prefactor Ap (lm3b MPan s1 K1) 1  105 3.3  103
Activation energy Qp (kJ mol1) 167  103 238  103
Prefactor Ad (lm3b MPan s1 K1) 4.8  104 4.8  104
Activation energy Qd (kJ mol1) 150  103 150  103
Critical damage Dcr 0.85 0.85
Minimum strain (before nucleation) m 0.01 0.01
a
The subscripts q = p, d refer to power law and diffusion, respectively. The same order is valid for the following two rows
b
lm3 is included only if q = d.
A. Karrech et al. / International Journal of Engineering Science 49 (2011) 1141–1150 1147

4. Numerical application

4.1. Geometry and boundary conditions

The numerical application considered in this paper represents a plane strain continental cross-section of length 60 km
and depth 25 km where the upper and lower layers are dominated by quartz and feldspar, respectively (see Fig. 1a). The ma-
trix contains perturbations which are distributed randomly in space according to the uniform distribution: Un[u1, u2] where
u1 and u2 are the minimum and maximum values. A stochastic realisation consists in selecting elements as:
e e ¼ maxðNÞU n ½0; 1 where max(N) represents the total number of elements. The perturbations are applied by random
N
reduction of the elastic properties and yield parameters as follows: ð e E; r
~ 0; r
~ 1 Þ ¼ ðE; r0 ; r1 ÞU n ½0:35; 0:5. As they are uni-
formly distributed in space, these perturbations do not privilege directions of anisotropy or diffuse failure. The cross-section
covers a wide range of temperature distribution resulting in rate dependent behaviour and damage weakening feedback. The
rheologies of quartz and feldspar under these conditions are characterised by multiple creep mechanisms dominated by dif-
fusion and dislocation regimes. These regimes govern simultaneously the growth of nucleated voids as described in detail by
Karrech et al. (2011). The resulting inelastic energy is included in a close loop as a thermal feedback since the cross-section is

Fig. 2. Equivalent inelastic deformation in the continental cross-section: a comparative study between pressure dependent and shear-dominated
responses.
1148 A. Karrech et al. / International Journal of Engineering Science 49 (2011) 1141–1150

subjected to fully coupled thermo-mechanical loading (Regenauer-Lieb et al., 2008; Regenauer-Lieb et al., 2006). Table 1
summarises the parameters of simulation used to benchmark the numerical model.
The conducted analysis is based on a comparative study between a frictional and associated cases pointing out the effect
of pressure sensitivity. In both cases, the displacement boundary conditions consist of a free slip on the bottom and right
edges and an extension velocity v0 = 20 mm/y at the right side of the model (see Fig. 1a). An initial surface temperature
of 28 °C on the top and a gradient of 28 °C km1 are applied through the continental cross-section. In addition, a body force
is considered to include the effect of gravity.

4.2. Results and discussion

The equivalent stress distribution shows the effect of the random distribution of material properties on the response of
the cross-section as can be shown in Fig. 1b. A diffuse distribution of the equivalent stress can be noticed at the beginning of
the yielding process. The equivalent stress reduces when damage nucleates in the lower crust and propagates depending on
level of material degradation. As the nucleated weakness propagates through the cross-section, the equivalent stress reduces
progressively in both loading scenarios.
Fig. 2a and b shows the distribution of the inelastic deformation at the early stage of yielding. It can be seen that the
inelastic deformation mainly takes place in the lower crust, despite the random distribution of the material properties. This

Fig. 3. Damage propagation in the continental cross-section: a comparative study between pressure dependent and shear-dominated responses.
A. Karrech et al. / International Journal of Engineering Science 49 (2011) 1141–1150 1149

result is attributed to the effect of viscoplasticity which is predominant in the lower part of the continental cross-section.
Fig. 2c and d represents the progression of the inelastic deformation at an early stage of damage (the threshold of damage
corresponds to a critical inelastic deformation of 0.01). The loading scenarios produce different features because of the pres-
sure dependency. In case of a metallic-like behaviour (described by a Von-Mises type of flow), the created shear bands are
oriented by 45° with respect to the vertical irrespectively of the depth. However, introducing an exponential dependence of
the behaviour on pressure induces deviation of the shear zones with respect to the depth. It can be noticed for instance that
the trace of inelastic deformation is almost vertical at the surface but converges towards 45% at the bottom end. This re-
sponse is explained by the asymptotic yielding that can be encountered in the lower crust and the conical yielding that is
predominant in the upper crust. When damage increases further the above weakening process is accelerated and faults local-
ise in the continental cross-section Fig. 2e and d. A high necking effect is produced later as can be shown in Fig. 2g and h. The
two scenarios show that the damage process follows the initial paths obtained by inelastic deformation but in a more accel-
erated manner. They also prove that the frictional behaviour of geomaterials has a high influence on faults orientations.
Examination of the different maps of damage obtained from the two scenarios at various ages of deformation shows that
damage nucleates in the lower crust (see Fig. 3a and b). In addition, following the distribution of damage with respect to time
reveals that it preferentially propagates towards the weaker zones. In the absence of a dissipating material, a reflexion takes
place at the boundaries (Fig. 3). In the upper crust the degradation propagates progressively in an exponential orientation
(resp. linear orientation) if the frictional aspect is (resp. is not) taken into account, in accordance with the inelastic
deformation.

5. Conclusion

The model introduced in this paper describes the coupled thermo-mechanical behaviour of damaged geomaterials. It is
developed within the general thermodynamic framework of frictional materials as formulated by Houlsby and co-authors
(Collins & Houlsby, 1997; Houlsby & Puzrin, 2000; Nguyen & Houlsby, 2007) in terms of time-independent behaviour. How-
ever, an extension to time-dependency is introduced through the continuous approach of Ponthot (1995). By using a shear
and pressure dependent dissipation, the formulated model combines the associated and non-associated behaviour in the
same framework. As an application, a continental cross-section containing randomly distributed imperfections and sub-
jected to horizontal extension is used to compare the responses of pressure dependent and shear-dominated materials.
The results showed that pressure dependent (frictional) geomaterials exhibit depth-dependent orientations of faults which
reach 45° in the high pressure zones and tend to the vertical direction (depending on the friction angle) in the low pressure
zones. The model also showed that at the beginning of the extension process, inelastic deformations dominate the lower
crust although the material properties are randomly distributed. As consequence, faults are nucleated in the lower crust
where relatively high stresses and temperature induce yielding and viscous over-stresses. The model showed that the nucle-
ated faults propagate preferentially throughout the weaker zones and follow the pattern of equivalent inelastic deformation.

References

Carberry, D. M., Reid, J. C., Wang, G. M., Sevick, E. M., Searles, D. J., & Evans, D. J. (2004). Fluctuations and irreversibility: An experimental demonstration of a
second-law-like theorem using a colloidal particle held in an optical trap. Physical Review Letters, 92. 140601:1–4.
Carosio, A., Willam, K., & Etse, G. (2000). On the consistency of viscoplastic formulations. International Journal of Solids and Structures, 37, 7349–7369.
Chaboche, J. L. (1987). Continuum damage mechanics: Present state and future trends. Nuclear Engineering and Design, 105, 19–33.
Cocks, A. C. F., & Ashby, M. F. (1982). On creep fracture by void growth. Progress in Materials Science, 27, 189–244.
Collins, I. F., & Houlsby, G. T. (1997). Application of thermodynamical principles to the modelling of geotechnical materials. The Royal Society, 453,
1975–2001.
Drucker, D. C. (1959). A definition of a stable inelastic material. Journal of Applied Mechanics, 26, 101–106.
Evans, D. J., Cohen, E. G. D., & Morriss, G. P. (1993). Probability of second law violations in shearing steady states. Physical Review Letters, 71, 2401–2404.
Halphen, B., & Nguyen, Q. S. (1975). Sur les materiaux standards generalises. Journal de Mecanique, 14, 39–63.
Hill, R. (1950). Mathematical theory of plasticity. New York: Oxford University Press.
Hofstetter, G., Simo, J. C., & Taylor, R. L. (1993). A modified cap model: Closest point solution algorithms. Computer and Structures, 46(2), 203–214.
Houlsby, G. T., & Puzrin, A. M. (2000). A thermomechanical framework for constitutive models for rate-independent dissipative materials. International
Journal of Plasticity, 16, 1017–1047.
Houlsby, G. T., & Puzrin, A. M. (2006). Principles of hyperplasticity: An approach to plasticity theory based on thermodynamic principles. London: Springer-
Verlag.
Karrech, A., Regenauer-Lieb, K., & Poulet, T. (2010). Frame indifferent elastoplasticity of frictional materials at finite strain. International Journal of Solids and
Structures. doi:10.1016/j.ijsolstr.2010.09.026.
Karrech, A., Regenauer-Lieb, K., & Poulet, T. (2011). Continuum damage mechanics for the lithosphere. Journal of Geophysical Research, 116, Bo4205,
doi:10.1029/2010JB007501.
Karrech, A., Seibi, A., & Duhamel, D. (2010). Finite element modeling of rate-dependent ratcheting in granular materials. Computers and Geotechnics.
doi:10.1016/j.compgeo.2010.08.012.
Lemaitre, J. (1985). Coupled elasto-plasticity and damage constitutive equations. Computer Methods in Applied Mechanics and Engineering, 51, 31–49.
Lubliner, J. (1975). Non-smooth dissipation functions and yield criteria. Acta Mechanica, 22, 289–293.
Lubliner, J. (1978). A thermodynamic yield criterion in viscoplasticity. Acta Mechanica, 30, 165–174.
Nguyen, G., & Houlsby, G. T. (2007). A coupled damage-plasticity model for concrete based on thermodynamic principles: Part i: Model formulation and
parameter identification. International Journal for Numerical and Analytical Methods in Geomechanics, 32, 353–389.
Onsager, L. (1931). Reciprocal relations in irreversible processes, i. Physical Review, 37, 405–426.
Ponthot, J. P. (1995). Radial return extensions for visco-plasticity and lubricated friction. In Proceedings of the international conference on structural mechanics
and reactor technology SMIRT-13 (Vol. 2, pp. 711–722).
Rajagopal, K. R., & Srinivasa, A. R. (2004). On thermomechanical restrictions of continua. Proceedings of the Royal Society London, 460, 631–651.
1150 A. Karrech et al. / International Journal of Engineering Science 49 (2011) 1141–1150

Regenauer-Lieb, K., Karrech, A., Chua, H. T., Horowitz, F. G., & Yuen, D. (2010). Time-dependent, irreversible entropy production and geodynamics.
Transactions of the Royal Society of London: Series A, 368 (1910), 285–300.
Regenauer-Lieb, K., Rosenbaum, G., & Weinberg, R. F. (2008). Strain localisation and weakening of the lithosphere during extension. Tectonophysics, 458(1–
4), 96–104.
Regenauer-Lieb, K., Weinberg, R. F., & Rosenbaum, G. (2006). The effect of energy feedbacks on continental strength. Nature, 442, 67–70.
Rice, J. R. (1971). Inelastic constitutive relations for solids: An internal variable theory and its application to metal plasticity. Journal of Mechanics and Physics
of Solids, 19, 433–455.
Simo, J. C., & Taylor, R. L. (1985). Consistent tangent operators for rate-dependent elastoplasticity. Computer Methods in Applied Mechanics and Engineering,
48, 101–118.
Voyiadjis, G. Z., & Abed, F. H. (2006). A coupled temperature and strain rate dependent yield function for dynamic deformations of bcc metals. International
Journal of Plasticity, 22, 1398–1431.
Ziegler, H. (1963). Some extremum principles in irreversible thermodynamics with application to continuum mechanics. Progress in solid mechanics (Tome
IV). North Holland Pub.
Ziegler, H. (1977). An introduction to thermomechanics. Netherlands: North Holland.

You might also like