International Conference on Mathematical and Statistical Modeling

in Honor of Enrique Castillo. June 28-30, 2006

Tracking 3D Crack Path
Eduardo W. V. Chaves
Department of Applied Mechanics and Engineering Design,
University of Castilla-La Mancha.
Abstract
The main objective of this article is to obtain the 3D crack path (the overall
track technic). To attain this goal, general explicit expressions for both the critical
failure direction and the critical hardening modulus are presented. Under this light,
the ellipticity condition of the constitutive tangent operator plays a determinant
role. Taking into account the general explicit expression for the critical values we
can particularize to several constitutive models (namely, continuum damage and
plasticity models).
Key Words: Crack path, Bifurcation analysis, Hardening/Softening critical, Lo-
calization direction.
1 Introduction
Some of the methods used to simulate the crack propagation (like Strong
Discontinuity Approach) are based on an a priori knowledge of the crack
path. The crack path is a continuous line in 2D problem and a continuous
surface in 3D, dividing the zone of discontinuities. When the methods in-
clude a remeshing strategy, the crack direction must be know a priori to es-
tablish the remeshing criteria. In this article we present a method to estab-
lish the crack path, mainly based in the work of Oliver et al.(2002) known
as the Overall Tracking. One of the original ingredient of this method is
the critical direction concept. This is derived from a material bifurcation
analysis, which yields a general expression for the critical values, that can
be particularized for some constitutive models.
2 Boundary Value Problem (BVP)
We start by stating the standard Boundary Value Problem (BVP) of the
solids. Let us consider a three dimensional body, that takes up an open

Correspondence to: Eduardo W.V. Chaves. Department of Applied Mechanics

and Engineering Design. University of Castilla-La Mancha, Ciudad Real, Spain. email:
Eduardo.Vieira@uclm.es
2 E.W.V. Chaves
bounded domain B R
3
with density (see Figure 1). Let B be the
boundary of B and the unit outward normal to B. The body is assumed
to be in static equilibrium under the action of the body forces density,
b(x), and surface tractions, t

(x). The boundary B consists of the portion

u
B with prescribed displacement, u(x), and a part

B, with prescribed
traction, t

(x), such that

u
B

B = and
u
B

B = B. In
addition let (x) denote the Cauchy stress tensor and (x) the innitesimal
strain tensor. The governing equations of the Boundary Value Problem for
this quasi-static problem are summarized as following:
Partial dierential equations _ (x) +b(x) = 0 x B
Constitutive equation (x) = ( (x)) x B
Kinematics equation (x) = _
sym
u x B
Dirichlets boundary conditions u = u

(x) x
u
B
Neumanns boundary conditions t = t

(x) = x

B
(2.1)


Figure 1: Three dimensional body.
2.1 Constitutive models. Tangent modulus
When we use time integration of the evolution law it is important to obtain
the incremental constitutive equation in terms of the stress and strain rates
= C
in
: where C
in
is the inelastic tangent constitutive tensor also known
as the tangent material operator whose general expression reads:
Tracking 3D Crack Path 3
C
in
= C
e
/(C
e
: mn : C
e
) (2.2)
with
damage models =
q
r
= (1 d) ; / =
q(r)H
d
r
r
3
plasticity models = 1 ; / =
1
H
p
+n:C
e
:m
(2.3)
where d is the damage parameter which ranges from 0 to 1, i.e. 0 d 1,
r is the internal variable whose value denes the elastic limit, q is the
stress-like hardening/softening variable dened as q(r) = (1 d)r, H is the
continuum hardening/softening parameter, n is the ow plastic, normal to
the Yield Function (T), dened as n =
F

, m is the ow of the plastic

potential, which is normal to the plastic potential ((), dened as m =
G

,
and C
e
is the standard fourth-order isotropic elastic modulus tensor which
can be dened as a function of Lames parameters , , as:
C
e
= 2I +(1 1) (2.4)
where 1 is the second order unit tensor and I is the fourth order unit tensor,
dened in components as I
ijkl
=
1
2
(
ik

jl
+
il

jk
).
For more detail in damage models see Kachanov (1958), Oliver et al.(1990),
Lemaitre(1996) and plasticity models see Chen (1982), Hill(1998), Lubliner
(1990).
3 Material bifurcation
In the eld of structural engineering, bifurcation theory has been exten-
sively applied to geometrical non-linearity problems, e.g. stability prob-
lems involving buckling, as well as constitutive non-linearity, i.e., material
instability phenomena (material bifurcation). In this section we will tackle
the problem of localization as an instability in material in the spirit of
Rudnicki & Rice(1975): ...Localization can be understood as an instability
in the macroscopic constitutive description of inelastic deformation of the
material.... The main aim here is to obtain general explicit expressions
for the critical failure direction and the critical hardening modulus, Willam
4 E.W.V. Chaves
(2000), corresponding to the best-known classical continuum constitutive
models (namely, continuum damage and plasticity models). To obtain the
critical values we will take as departure point the Strong Ellipticity condi-
tion.
3.1 Strong ellipticity condition
Here we present the denition of strong ellipticity that will used throughout
this work.
If for all vector N and M with NM ,= 0, the inequality
(NM) : C
in
: (NM) > 0 (3.1)
holds, then it is said that the tangential operator C
in
is strongly elliptic.
Rearranged, this inequality can be written as:
M Q(N) M > 0 (3.2)
where Q(N) = N C
in
N is a second-order tensor known in literature as
the Tangent Acoustic Tensor. By analogy with the theory of wave prop-
agation, the eigenvalue of Q(N) are the wave propagation velocities, N is
the propagation direction and M is the polarization direction.
The pair (N, M) indicates the orientation and nature of the discontinu-
ity, i.e., opening (Mode I) when N is parallel to M, shear (Mode II) when
N is orthogonal to M, or the combination of both, see Figure 2.
3.2 Ellipticity condition
If Q(N) is symmetric, condition (3.2) will be violated if, for some N, there
exists a M ,= 0 such that
Q(N) M = 0 (3.3)
that is:
Q(N) M = 0 det [Q(N)] = 0 (3.4)
Tracking 3D Crack Path 5

MODE I M N// MODE II M N
N M//
N
M
Figure 2: Nature of the discontinuity.
Condition (3.2) means that tensor Q(N) is positive denite for all M ,=
0 implying that uniqueness and stability is guaranteed.
As mentioned above, the following analysis will be performed from a
general point of view, and, later, it will be particularized for some specic
cases. As point of departure we will analyze the acoustic tensor, Q(N), in
order to obtain the value of N that minimizes det [Q(N)]. Using equation
(2.2), the acoustic tensor can be written as:
Q(N) = N C
e
N/N (C
e
: mn : C
e
) N (3.5)
Applying the denition of the standard fourth-order isotropic elastic
modulus tensor in function of Lames parameters (, ) and equation (2.4),
the above equation can be rewritten in the following way:
Q(N) =
_
Q
e

[NTr (m) + 2N m] [NTr (n) + 2N n]

_
(3.6)
where Q
e
is the elastic acoustic tensor dened as Q
e
= N C
e
N , and
Tr() stands for the trace of ().
After some mathematical manipulations we can obtain that:
det [Q(N)] = det [Q
e
] det
_
1
/

a b
_
(3.7)
6 E.W.V. Chaves
with a =[NTr (m) + 2N m] Q
e
1
and b =[NTr (n) + 2N n] . For
more detail see Chaves & Oliver (2003), Bigoni (1991).
It is possible demonstrate the following expression:
det
_
1
/

a b
_
= 1
/

a b (3.8)
Taking into account the fact that det [Q
e
] > 0, the bifurcation condi-
tion det [Q
e
] = 0, equation (3.7), is reduce to solving the following problem:
1
/

[NTr (m) + 2N m] Q
e
1
[NTr (n) + 2N n] = 0 (3.9)
Denoting by Z(N) = [NTr (m) + 2N m]Q
e
1
[NTr (n) + 2N n]
and expanding this expression we can obtain:
Z(N) =
2
Tr(m)Tr(n) (a +b) + 2Tr(n) (N m N) (3.10)
+2Tr(m) (N n N) (a +b) + 4
2
(N n N m)
+4
2
b (N m N) (N n N)
with
a =
1

; b =
(+)
(+2)
(3.11)
The problem to be solved here is to nd the critical normal direction
vector N, which can be done by maximizing function (3.10). Thus, the
problem of nding the critical normal N
crit
can be stated as the following
minimization problem:
N
crit
= arg
N=1
mindet [Q(N)] (3.12)
Remark 3.1. Note that even for elastic regime we will have critical values
once the Z(N) function is not dependent of / or .
Tracking 3D Crack Path 7
4 Critical values
In this section, we will explain how to obtain the critical values of the
normal vector N
crit
and the hardening parameters H
crit
, for the following
cases:
The colinear case, where n and m are coaxials ( the principal direc-
tions of n and m are coincident);
The associated case, a particular coaxial case when n = m.
4.1 Case of colinearity between n, m
4.1.1 Non-associated case
Consider the tensor n and m (n and m coaxials) expressed in the principal
directions components:
n
ij
=
_
_
n
1
0 0
0 n
2
0
0 0 n
3
_
_
m
ij
=
_
_
m
1
0 0
0 m
2
0
0 0 m
3
_
_
(4.1)
Developing expression (3.10) and using the restriction |N| = 1, we can
obtain the following expression for Z(N) :
Z(N) =

AN
4
1
+

BN
4
2
+

CN
4
3
+

GN
2
1
N
2
2
+

HN
2
1
N
2
3
+

IN
2
2
N
2
3
+

SN
2
1
+

TN
2
2
+

UN
2
3
+

W
(4.2)
with
8 E.W.V. Chaves

A = (n
1
m
1
) ;

B = (n
2
m
2
) ;

C = (n
3
m
3
)

G = (n
1
m
2
+n
2
m
1
) ;

H = (n
3
m
1
+n
1
m
3
) ;

I = (n
3
m
2
+n
2
m
3
) ;

S =
2
(+2)
[Tr (n) m
1
+Tr (m) n
1
] + 4(n
1
m
1
) ;

T =
2
(+2)
[Tr (n) m
2
+Tr (m) n
2
] + 4(n
2
m
2
) ;

U =
2
(+2)
[Tr (n) m
3
+Tr (m) n
3
] + 4(n
3
m
3
) ;

W =

2
Tr(m)Tr(n)
(+2)
; =
4(+)
(+2)
We can obtain the relative maximums and minimums of function (4.2)
and can be demonstrated, see Chaves&Oliver (2003). When n
1
> n
2
> n
3
and m
1
> m
2
> m
3
the maximum of this function lie in the plane direction
1-3 which value for the angle and Z are:
N
1
N
2
N
2
Angle Z(N)

G
2A
0.0
_
1 +

G
2A
tan
2
=
2A+

G

G
Z
max
= T
3


G
2
4A
(4.3)
with
A =
4(+)
(+2)
[(n
1
n
3
) (m
1
m
3
)]

G =
2
(+2)
[Tr (n) (m
1
m
3
) +Tr (m) (n
1
n
3
)]

4(+)
(+2)
[m
3
(n
1
n
3
) +n
3
(m
1
m
3
)] + 4(n
1
m
1
n
3
m
3
)
T
3
=
1
(+2)
[(Tr (m) + 2m
3
) (Tr (n) + 2n
3
)]
The critical angle also can be expressed in function of n and m compo-
nents as:
tan
2
=
[(m
3
m
1
) n
2
+ (n
3
n
1
) m
2
] + (2n
3
n
1
) m
3
m
1
n
3
[(m
1
m
3
) n
2
+ (n
1
n
3
) m
2
] + (2n
1
n
3
) m
1
m
3
n
1
(4.4)
The function Z(N), equation (4.2), is a continuous surface having maxi-
mum and minimum relative values. The shape of this surface when n ,= m,
Tracking 3D Crack Path 9



0

=
=
=

sin

cos sin

cos cos
3
2
1
N
N
N

1 = N

) (N Z

0
3
= N
1
3
= N
0
3
= N

0
90
90 180
0
2
= N
180
0
1
= N
Figure 3: Surface Z(N) - Case n ,= m.
for Drucker-Prager non associated case, is schematized in Figure 3, with its
relative maximum and minimum points.
Calculation of H
p
crit
Damage
For isotropic damage case H
d
crit
is given by:
H
d
crit
=
_
1
r
3
Z
max
(N)
_
(4.5)
Plasticity For plasticity case H
p
crit
is given by:
H
p
crit
=

G
2
4A
+P
3
(4.6)
with

G
2
4A
=
E
4(1
2
)
[(m
1
m
3
)(n
2
+n
1
)+(n
1
n
3
)(m
3
+m
1
)]
2
(m
1
m
3
)(n
1
n
3
)
P
3
=
E
(1
2
)
[ (m
2
n
1
+m
1
n
2
) + (m
2
n
2
+m
1
n
1
)]
10 E.W.V. Chaves
4.1.2 Critical Values - Associated case (n = m)
Critical Angle The critical angle could obtained by directly equation (4.4)
with n = m :
tan
2
=
m
3
m
2
m
1
m
2
(4.7)
Critical Hardening parameter (Damage)
H
d
crit
= (1 d)
_
1
(+)r
2

2
(Tr(m))
2
+[(m
1
m
3
)
2
+2Tr(m)(m
1
+m
3
)]+2
2
(m
2
1
m
2
3
)
_
(4.8)
Critical Hardening parameter (Plasticity)
H
p
crit
= Em
2
2
(4.9)
5 Critical values for several constitutive models
In solid mechanics there are several families of constitutive equations. Some
typical examples are elasticity, plasticity, viscoelasticity, continuum dam-
age, and viscoplasticity. In this section, critical values for the localization
direction and the hardening modulus for some of these models will be de-
rived. Several classic models of plasticity and damage are employed with
this purpose. A detailed study of their features is out of the scope of this
work. For further details on the constitutive models presented here, the
reader is referred to: Chen&Han(1988), Chen(1982), Potts&Zdravkic(1999),
Desai&Siriwardane(1984), Willam(2000).
5.1 One-parameter models
5.1.1 Rankine criterion
This is a model dened by a maximum-tensile-stress criterion (see Figure
4) and was formulated by Rankine in 1876. It is characterized by only
depending on a parameter. This can be sketched as:
Tracking 3D Crack Path 11


1


2


3




1


2


3

Figure 4: Rankine yield surface.
T = ( =
1

t
= 0 n = m =
_
_
1 0 0
0 0 0
0 0 0
_
_
(5.1)
where
1
>
2
>
3
(positive stress) are the principal stress and
t
is the
tensile yield stress satisfying
t
> 0.
Using the above yield criterion and substituting it into equation (4.7)
and considering (4.9), we readily obtain the critical angle and H
p
crit
:
Critical Value for Rankine Criterion
Critical angle tan
2

crit
= 0
_

1
= 0

2
= 0
Critical
hardening
modulus
H
p
crit
= 0
(5.2)
5.1.2 von Mises yield criterion
In this criterion the elastic limit is reached when the octahedral shear stress
(
oct
) reaches a critical value (yield stress in pure shear), i.e.:
oct
=
_
2
3
J
2
=
_
2
3
. We can write the yield criterion as:
T (J
2
) = J
2

2
= 0 (5.3)
12 E.W.V. Chaves


2


3


1


2

1


3



t
R =
3
2


t
R =
3
2


yield
Figure 5: von Mises yield surface.
We will consider the J
2
plasticity model, associated von Mises (Figura
5), where:
n = m =
_
3
2
s
s
=
_
3
2
s ; |s| =

s : s = 1 (5.4)
and s is the deviatoric stress tensor, which components are given by:
s
ij
=
_
_
2
1

3
3
0 0
0
2
2

3
3
0
0 0
2
3

2
3
_
_
(5.5)
Its norm reads: |s| =
_

2
1
+
2
2
+
2
3
.
According to equation (4.7) the critical angle is determined and to ob-
tain H
p
crit
, equation (4.9) is used, resulting in:
Critical Value for von Mises Criterion
Critical angle tan
2

crit
=
s
3
+s
2
s
1
+s
2
Critical
hardening
modulus
H
p
crit
=
3Es
2
2
2(s
2
1
+s
2
2
+s
2
3
)
(5.6)
Tracking 3D Crack Path 13
5.1.3 Tresca yield criterion
This model is also known as the maximum shearing stress criterion, which
has the following yield criterion (Figure 6):
T(,
max
) = max
_
1
2
[
i

j
[
_

max
= 0 (5.7)
From the condition
1
>
2
>
3
(positive stress), equation (5.7) results
in max
_
1
2
[
i

j
[

= max
_
1
2
[
1

3
[

, thus:
n = m =
_
_
1
2
0 0
0 0 0
0 0
1
2
_
_
(5.8)
The critical angle and the critical hardening modulus are summarized
as follows:
Critical Value for Tresca Criterion
Critical angle tan
2

crit
= 1
_

1
= +45
o

2
= 45
o
Critical
hardening
modulus
H
p
crit
= 0
(5.9)
5.2 Two-parameter models
5.2.1 Mohr-Coulomb criterion
The simplest form of the Mohr envelope is the straight line (see Figure 7).
The equation of this line is:
[[ = c tan (5.10)
where c is the cohesion and is the angle of internal friction, and both are
material constants determined by experiments.
14 E.W.V. Chaves


2


1


3

octahedrical plane
0
1
= I (plane )

II


I


III


III II I
= =


6

+


pure shear state
Figure 6: Tresca yield surface.


Figure 7: Relationship between principal stresses for the Mohr-Coulomb criterion.
Considering Figure 7 and equation (5.10), and taking into account that

1

2

3
, and considering the soil mechanics for compression positive
and for traction negative, i.e.:
1
=
3
;
3
=
1
, the Mohr-Coulomb
surface equation is described as:
Tracking 3D Crack Path 15
T(
1
,
3
) =
1

3
sin
_

1
+
3
+
2c
tan
_
(5.11)
Considering that:
n =
_
_
N

1
0 0
0 0 0
0 0 N

3
_
_
; m =
_
_
M

1
0 0
0 0 0
0 0 M

3
_
_
(5.12)
where N

1
and N

3
are functions of (angle of internal friction) and M

1
and
M

3
are function of (dilatancy angle) with
N

1
=
1sin
q
2(1+sin
2
)
; N

3
=
1+sin
q
2(1+sin
2
)
M

1
=
1sin
q
2(1+sin
2
)
; M

3
=
1+sin
q
2(1+sin
2
)
(5.13)
The critical values are given by:
Critical Value for Mohr-Coulomb Criterion
Critical angle tan
2

crit
=
2 sin sin+sin+sin
2 sin sinsinsin
Critical
hardening
modulus
H
p
crit
=
E
8(1
2
)
_
(sin sin )
2
q
(1+sin
2
)(1+sin
2
)
_
(5.14)
5.2.2 Non-associated Mohr criterion- Particular case
In this particular case we have:
n =
_
_
1 0 0
0 0 0
0 0 N

_
_
; m =
_
_
1 0 0
0 0 0
0 0 M

_
_
(5.15)
where
16 E.W.V. Chaves
N

=
1sin
1+sin
; M

=
1sin
1+sin
with N

0 and M

0
(5.16)
Substituting the values of n and m, given by the expressions (5.15),
into equations (4.4) and (4.6) for non-associated problems, we can obtain
the critical angle and the critical value of the hardening modulus:
Critical Value for Mohr Criterion (associated case)
Critical angle tan
2

crit
=
2sinsin
2+sin+sin
Critical
hardening
modulus
H
p
crit
=
E
4(1
2
)
_
(N

)
2
(1+M

)(1+N

)
_
(5.17)
The result (5.17) are consistent with the results presented by Ottosen
& Runesson(1991).
5.2.3 Drucker-Prager criterion - Two-invariant plasticity formu-
lation
In this special case, see Figure 8, we can express n and m, which are
coaxials, as:
_
n =
F

=
1
s +
3
1
m =
G

=
2
s +
4
1
(5.18)
The parameters
1
and
3
describe the pressure sensitive, and
3
and

4
are frictional material parameters.
Using equations (4.4) and (4.6) with n and m given by equation (5.18),
we can obtain the critical angle and the critical hardening parameter:
Critical Value for the Drucker-Prager (associated case)
Critical angle tan
2

crit
=
(1+)(
1

4
+
2

3
)2(s
3
+s
2
)
1

2
(1+)(
1

4
+
2

3
)2(s
1
+s
2
)
1

2
Critical
hardening
modulus
H
p
crit
=
E
(1
2
)
[A
1
+A
2
+A
3
]
(5.19)
Tracking 3D Crack Path 17


1


2


3


3 2 1
= =
(a) Two-invariants

1


2


3


3 2 1
= =
(b) Particular case

Figure 8: Drucker-Prager yield surface.
where
A
1
=
[(1 2) s
3

2
+ (1 +)
4
] [(1 2) s
3

1
+ (1 +)
3
]
(1 2)
(5.20)
A
2
=
[2 (s
2
+s
1
)
1

2
+ (1 +) (
1

4
+
2

3
)]
2
4
1

2
A
3
= 3 (1 )
(1 2)
2
oct

2
+ (1 +)
3

4
(1 2)

2
oct
=
1
9
_
(
1

2
)
2
+ (
2

3
)
2
+ (
3

1
)
2
_
Particular case
1
=
2
= 1
In the particular case of the Drucker-Prager model with two invariants
take place that
1
=
2
= 1. In this case the model only depends on the
friction parameter of the Drucker-Prager parabolic model:
_
n = s +
3
1
m = s +
4
1
(5.21)
The values for the critical angle and the hardening parameter are:
18 E.W.V. Chaves
Critical Value for the Ducker-Prager (associated case)
Particular case (
1
=
2
= 1)
Critical angle tan
2

crit
=
r(12)(
c

3
)(1+)

4
+
3
2

r+(12)(
c

3
)+(1+)

4
+
3
2

Critical
hardening
modulus
H
p
crit
=
E
(1
2
)
[A
1
+A
2
+A
3
]
(5.22)
where:
c
=

1
+
3
2
; r =

1

3
2
; I

= Tr(), the rst invariant of the stress

tensor. And the values of A
1
, A
2
, A
3
are the presented in the expressions
(5.20). The expression for the critical angle obtained here is the same the
one obtained by Willam(2000).
6 The overall tracking
To establish a continuous failure surface a new method has been proposed
by Oliver et al. (2002). This idea starts from the principal that from
the normal (N) obtained from the bifurcation analysis a family of surfaces
(level surfaces) in 3D enveloping the propagation direction case can be
constructed. The following analogy with heat conduction problem (heat
conduction-like problems) to obtain that family of surfaces was presented
in Oliver et al. (2002).
6.1 Heat conduction-like problem
Let N(x, t) be a family of unity vectors, dened at every point in the
domain B at a given time t, determining the direction normal to the plane
of propagation of the discontinuity. Then let S(x, t) and T(x, t) be any
couple of unity vector orthogonal to N, so that:
S N = T N = 0 (6.1)
thus dening the plane (tangent to them) of propagation of the discon-
tinuity. The family of surfaces, enveloping both vectors, S and T, can
Tracking 3D Crack Path 19
be described by a scalar (temperature like) function (x) such that the
isothermal surfaces:
o
j
=
_
x B ; (x) =
ref
j
_
(6.2)
for all meaningful value of
ref
j
are tangent at each point x B to vectors
S and T. Therefore:
S _ = _S =

S
= 0
T _ = _T =

T
= 0
_
_
_
in B (6.3)
Solutions of the problem (6.3) are also solutions of the following heat
conduction problem:
FIND (x) such that
_

_
_ q = 0 in B
q =K
C
_ = S

S
= T

T
= 0 in B
q =(S )

S
+(T )

T
= 0 on
q
B
=

in

B
(6.4)
where is the outward normal to the boundary B and
q
B and

B
_

q
B

B = B
_
, stand, respectively, for the parts of the boundary B
where the Neumann and Dirichlet conditions are prescribed. K
C
is an
anisotropic conductivity tensor and given by:
K
C
= S S +TT (6.5)
K
C
ij
=
_
_
S
2
1
+T
2
1
S
1
S
2
+T
1
T
2
S
1
S
3
+T
1
T
3
S
1
S
2
+T
1
T
2
S
2
2
+T
2
2
S
2
S
3
+T
2
T
3
S
1
S
3
+T
1
T
3
S
2
S
3
+T
2
T
3
S
2
3
+T
2
3
_
_
(6.6)
6.1.1 Flowchart of the coupled problem
Figure 9 shows the strategy for the entire analysis of the mechanical prob-
lem. The rst phase of the process requires mechanical analysis and the
second requires thermal-like analysis.
20 E.W.V. Chaves




Figure 9: Flowchart of coupled problem.
7 Some examples of the failure surface
In this section we show some examples to describe the eciency of the
method used in this work to obtain a continuous failure surface. Numer-
Tracking 3D Crack Path 21


a) b)
Figure 10: Testing apparatus, Bocca et al. (1990).
ically we modelled all problems using the Finite Element Method (FEM)
for 3D non-linear problems. The constitutive model employed to obtain
the critical direction (and the critical surfaces) was the isotropic damage
model.
7.1 Single-notched shear beam
The specimen used is made of plain concrete. The testing apparatus is
shown in Figure 10(a). Originally it was proposed by Arrea & Ingraea
(1982). The experimental failure surface is shown in Figure 10(b). The
mesh employed and geometry in the simulate is shown in Figure 11.
Figure 12(a) shows some family of envelops of the propagation direc-
tions. And in Figure 12(b) we can see part of the beam where the continuous
failure surface is shown.
7.2 Double-notched shear beam
This test was performed by Bocca et al. (1990). The geometrical features
and boundary conditions are shown in Figure 13. The mechanical proper-
ties are: E = 27000MPa, Poissons ratio = 0.18. Figure 14 shows the
normal vector eld only for the elements intersected by two surfaces. In
Figure 15(b) we show part of the beam without the elements intersected by
22 E.W.V. Chaves

20.3

P
14
53.7 39.7
134
20.3

2
2
.
4


8
.
2


3
0
.
6



1
5
.6

3
0
.
6

Figure 11: Single-notched shear beam - geometry (centimeter).


continuous failure surface
a)
b)
Figure 12: Some family of envelopes of the critical direction, and continuous failure
surface.
Tracking 3D Crack Path 23


4
16
20
80
F
8
cm t 10 =
Figure 13: Four point shear specimen (dimensions in cm).


Figure 14: Elements intersected by failure surfaces.
the discontinuity surfaces so that it can compared with the crack pattern
obtained by Bocca et al. (1990), see Figure 15(a).
24 E.W.V. Chaves



a) Bocca et al. (1990)
continuous discontinuity surfaces
b)
c)
Figure 15: Four point shear specimen - initial failure surface.
7.3 Anchorage structure
This example was simulated numerically by Rots (1998) and de Borst (1986)
using 2D axi-symmetric nite elements. The geometry used is shown in
Figure 16. The structure consists of a steel plate embedded in a massive
concrete block. The plate is pulled out of the concrete by a vertical load F
which is applied via an anchor bolt. The material parameters taken were
E = 30000N/mm
2
, Poissons ratio = 0.2. Figure 17 shows the continuous
failure surface which was obtained using overall tracking.
References
[1] J. Oliver, A.E. Huespe, E. Samaniego, and E.W.V. Chaves, On strate-
gies for tracking strong discontinuities in computational failure mechan-
ics.WCCM V - Fifth World Congress on Computacional Mechanics,
July 7-12, Vienne, Austria H.A. Mang et al. (Eds.), 2002.
Tracking 3D Crack Path 25




F
0 =
Figure 16: Detail of the anchorage structure.
[2] R. Hill, The Mathematical Theory of Plasticity, Oxford University
Press, USA, 1998, ISBN 0-19-850367-9.
[3] J. Lubliner, Plasticity theory, Macmillan Publishing, New York, 1990
ISBN 0-02-372161-8.
[4] J. Lemaitre, A course on damage mechanics, MSpring-Verlag Berlin
Heidelberg., Germany, 1996.
[5] W.F. Chen, Plasticity in reinforced concrete, McGraw-Hill, Inc. USA,
1982.
[6] J.W. Rudnicki, and J.R. Rice, OCondition for the localization of
the deformation in pressure-sensitive dilatant material.J. Mech. Phys.
Solids, 23, 371-394, 1975.
[7] E.W.V. Chaves, and J. Oliver, A three dimensional setting for strong
discontinuities modelling in failure mechanics, International Center for
Numerical Methods in Engineering, Monograph CIMNE M73. Spain,
2003.
26 E.W.V. Chaves



Figure 17: Continuous failure surface for the anchorage structure.
[8] K. Willam, Constitutive models for materials, Encyclopedia of Physical
Science and Technology, 3rd edition. Academic Press, 2000.
[9] W.F. Chen, and J.D. Han, Plasticity for Structural Engineers, Springer-
Verlag New Yor Inc, 1988.
[10] D.M. Potts, and L. Zdravkovic, Finite element analysis in geotechnical
engineering, Thomas Telford Publishing, London, 1999.
[11] C.S. Desai, and H.J. Siriwardane, Constitutive laws for engineering
materials with emphasis on geologic materials, Prentice-Hall, Inc.USA,
1984.
[12] N.S. Ottosen, and K. Runesson, Discontinuous bifurcation in a nonas-
sociated Mohr material.Mech. Mater., 12, 255-256, Elsevier, 1991.
[13] M. Arrea, and A.R. Ingraea, Mixed-mode crack propagation in mortar
and concrete, Report N
o
81-13, Department of Structural Engineering,
Cornell University, Ithaca, New York, 1982.
[14] P. Bocca, A. Carpinteri, and S. Valente, Size eects in the mixed mode
Tracking 3D Crack Path 27
crack propagation: Softening and snap-back analysis.Eng. Fracture
Mech., 35:159-170, 1990.
[15] J.G. Rots, Computational modeling of concrete fracture, Dissertation,
Delft University of Technology, Delft, Netherlands, 1988.
[16] R. de Borst, Non-linear analysis of frictional materials., Dissertation,
Delft University of Technology, Delft, Netherlands, 1986.
[17] D. Bigoni, and T. Hueckel, Uniqueness and localization - I. Asso-
ciative and non-associative elastoplasticityInt. J. Solids Struct., 28(2),
pp. 197-213, 1991.
[18] D. Bigoni, and D. Zaccaria, On strain localization analysis of elasto-
plastic materials at nite strainsInt. J. Plasticity, 9 N
o
1, pp. 21-33,
1993.