This action might not be possible to undo. Are you sure you want to continue?

A.R. Shahani

*

, M.R. Amini Fasakhodi

Faculty of Mechanical Engineering, K.N.T. University of Technology, Pardis Street, Mollasadra Avenue, Vanak Square, P.O. Box 19395-1999, Tehran, Iran

a r t i c l e i n f o

Article history:

Received 29 April 2008

Accepted 21 June 2008

Available online 2 July 2008

Keywords:

Dynamic fracture toughness

Remeshing

Arrest

RDCB

a b s t r a c t

This paper presents a ﬁnite element analysis based on the remeshing technique to predict the dynamic

crack propagation and crack arrest in a brittle material, namely Araldite-B. The dynamic fracture tough-

ness criterion is used to form the crack tip equation of motion. Plane stress condition is invoked in the

present two-dimensional fracture analysis of the RDCB (rectangular double cantilever beam) specimens

and the unknown crack tip position, and velocity is computed during the analysis by the dynamic fracture

criterion. According to the new crack tip position, a remeshing algorithm has been used to simulate the

dynamic crack growth and arrest. The obtained results including kinetic energy and strain energy, crack

tip velocity, dynamic stress intensity factor during crack growth and also crack arrest length have been

presented. Comparison of the results with those cited in the literature has shown a good agreement.

Ó 2008 Elsevier Ltd. All rights reserved.

1. Introduction

Dynamic fracture deals with fracture under conditions where

inertia forces must be included in the problem formulation. This

occurs either under dynamic loading or in the case of static loading

as a rapidly propagating crack runs through a structure. In this pa-

per, the latter one is investigated in the RDCB (rectangular double

cantilever beam) specimens.

The double cantilever beam specimen is one of the most com-

monly used test conﬁgurations for determining crack growth

speed and fracture toughness of materials. Some theoretical analy-

sis of the RDCB specimen, via either simple beam or shear beam

theories, can be found in previous literature [1–6]. The solutions

have been given based on the energy balance for rapid crack prop-

agation and arrest. The work implemented by Shahani and Forqani

[6] includes the beam theory considering shear deformation ef-

fects, and the obtained ﬁnal intricate equations of motion for the

case of dynamic analysis were simpliﬁed by eliminating the inertia

terms and were solved as a quasi-static state. The application of

beam theory to dynamic crack propagation is particularly attrac-

tive because it is one-dimensional analysis. Although the beam

analysis cannot predict the details of crack tip stresses or strains,

it does provide an accurate account of energy quantities which

form the basis of the fracture criterion.

Parallel to the analytical research, the ﬁnite element method

provides a powerful alternative of analyzing most real crack prop-

agation in specimen conﬁgurations. In general, analytic solutions

for the prediction problems of dynamic crack propagation are

rarely available. Therefore, numerical methods are necessary to ob-

tain solutions to predict these problems. Among the numerical

methods, ﬁnite element method is the most popular numerical

method used for the analysis and predictions of dynamic fracture,

and when it is employed accuracy of the results heavily depends on

the simulation method of crack growth. Consequently, a reliable

numerical analysis method for dynamic fracture is needed.

The earliest ﬁnite element methods for crack propagation use a

node decoupling technique with a simple nodal force release

mechanism [7–9]. As discussed by Kanninen [10], these early

methods generally produce inaccurate results, and are inappropri-

ate for modeling crack propagation. Node release technique has the

drawback of requiring a priori the knowledge of the path followed

by the crack, and also singular elements could not be employed

around the crack tip in this method.

Nishioka and Atluri [11] introduced a moving singular element

procedure for dynamic crack propagation analysis. In their method,

a special singular element that follows the moving crack tip is

used, and during the simulation of crack propagation only the con-

ventional elements immediately surrounding the singular element

are distorted. Dynamic fracture analysis of RDCB specimen using

the moving ﬁnite element method was carried out by Nishioka

and Atluri [12].

A version of the mixed Eulerian–Lagrangian kinematics descrip-

tion (ELD) was developed by Koh and Haber [13] and Koh et al. [14].

In this method, a single mesh pattern changes continuously and

independently from the material motion to model the crack propa-

gation, and auxiliary regional mapping is used to relate the changes

of the nodal coordinates to the crack tip motion. A major disadvan-

tage of this methodis that the ﬁnal generalizedcoefﬁcient matrixfor

the ﬁnite element equation results in a full-sized nonsymmetrical

0261-3069/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.

doi:10.1016/j.matdes.2008.06.049

* Corresponding author. Tel.: +98 21 84063221; fax: +98 21 88674748.

E-mail address: Shahani@Kntu.ac.ir (A.R. Shahani).

Materials and Design 30 (2009) 1032–1041

Contents lists available at ScienceDirect

Materials and Design

j our nal homepage: www. el sevi er . com/ l ocat e/ mat des

matrix, and solving these equations greatly increases the required

time. Koh et al. [15] applied ELD approach in ﬁnite element method

to predict the dynamic fracture in RDCB specimens.

The element-free Galerkin (EFG) method [16,17] is another

technique which, based on the moving least-square interpolants,

aimed at the simulation of crack growth problems. It requires only

nodal data, and no element connectivity is needed. Imposition of

the geometrical boundary conditions by the EFG technique is more

complicated than for the FEM.

In elastic–plastic materials, Beissel et al. [18] introduced an ele-

ment failure algorithm for crack growth in general direction. This

algorithm does not require remeshing technique, and is achieved

by tracking the path of the crack tip and failing the elements

crossed by the path such that they can no longer sustain deviatoric

or tensile volumetric stresses. Singular elements cannot be em-

ployed in this algorithm because the model has not been re-

meshed, and analysis is performed via the single and ﬁxed mesh

pattern during the entire process.

One of the unique features of any discrete fracture propagation

ﬁnite element code is its remeshing algorithm. Crack propagation

in elastic and elastic–plastic materials can be analyzed via this

method. Also this technique is not restricted to the special

mode in fracture mechanics, and due to this ability crack propaga-

tion in problems which have no symmetry, neither in geometry nor

in loading conditions, can be simulated.

Another important consideration in any such remeshing algo-

rithm is the application of elements to adequately model the crack

tip singularity [19]. In principle, two remeshing approaches have

been reported in the literature. The ﬁrst is to remesh the entire

model in every step as advocated by Shephard et al. [20] and Sumi

[21]. The advantage of this approach is that well shaped elements

can usually be generated. However, its disadvantage is that a large

number of state variables must be transferred from the old to the

new mesh, particularly when an elasto-plastic material is involved.

The second approach employs a local remeshing technique where-

by only the region of elements within a certain distance of the

crack tip is modiﬁed. This approach has the advantage of requiring

only the transference of state variables from a small region which

is to be remeshed. However, the generated elements may not be as

well-shaped as in the ﬁrst approach since there tend to be greater

constraints on the placement of the new elements due to the exist-

ing mesh surrounding the region to be remeshed.

Some crack propagation analysis based on the remeshing tech-

nique can be found in the previous works. In the LEFM conditions,

the local approach was used to investigate the fatigue crack growth

by Wawrzynek and Ingraffea [22,23] and simulation of dynamic

crack propagation in an inﬁnite medium by Swenson and Ingraffea

[24]. Bittencourt et al. [25] presented an algorithm named quasi-

automatic simulation of stable crack propagation for two-dimen-

sional LEFM problems in order to predict crack trajectory that is

very similar to the work done by Swenson and Ingraffea [24] in

remeshing algorithm. The analysis in [25] was said to be quasi-

automatic only because the user still needs to provide a desired

crack length increment at the beginning of each simulation, and

this is for the sake of being a stable crack growth. Tradegard

et al. [26] employed a combination of entire remeshing and nodal

relaxation to study mode I stable crack propagation in an elastic–

plastic material by ABAQUS software. Rethore et al. [27] employed

a stable numerical scheme for the analysis of dynamic crack

growth with remeshing but their study focused on the stability

subject of dynamic calculations. In principle, the aim of their work

is not to develop an efﬁcient remeshing procedure. Dynamic frac-

ture mechanics has been used as a basis for their study, and the

crack growth speed has been calculated via the criterion based

on the energy release rate considering that the material toughness

is independent of the crack speed. Recently, Shahani and Seyyedian

[28] employed an entire remeshing approach in order to simulate

glass cutting with the impinging hot air jet which could be inter-

preted as a controlled crack growth due to thermal stresses caused

by the hot air jet.

In this paper, the attempt is made toward gaining a ﬁnite ele-

ment simulation based on the remeshing technique compatible

with the actual dynamic fracture process. In this regard, mode I dy-

namic crack propagation and arrest phenomena are investigated in

the RDCB specimens under ﬁxed displacement loading condition.

The ﬁnite element modeling is accomplished with the standard FE

code ANSYS 7.0. Since the dynamic crack growth is intrinsically

an unstable phenomenon, crack growth increment cannot be de-

ﬁned by the user, and a dynamic fracture criterion should be em-

ployed to ﬁnd the crack tip velocity and crack extension every

time, automatically. In this study, dynamic fracture toughness cri-

terion is used to predict the crack tip velocity. The quarter point sin-

gular elements are used around the crack tip in order to model the

singularity. Since the crack is to propagate, a remeshing algorithm

is used in each crack extension step. For this to be achieved, ANSYS

Parametric Design Language, APDL, is employed. Owing to time-

dependent nature of the problem and inertia effects, nodal data

should be transferred from the old mesh to the new mesh in each

step of the remeshing. Finally, the problem is analyzed for two dif-

ferent crack tip bluntings to investigate its effects on crack arrest

length, crack arrest time, crack growth velocity and the quantity

of strain energy and kinetic energy during the crack advancement.

The predicted results showgood agreement with those obtained via

the experimental study and the other numerical techniques.

2. Dynamic crack propagation analysis

2.1. Dynamic stress intensity factor

In general, the crack tip stress intensity factor represents the ef-

fect of the applied loading, the geometrical conﬁguration of the

body and the bulk material parameters in the crack tip region for

any motion of the crack tip. For a dynamically propagating crack

under a steady-state or under an unsteady-state condition, these

parameters depend on time, the crack length and the crack tip

velocity. For the model problem considered here, the instanta-

neous value of the crack tip stress intensity factor for an arbitrary

motion of the crack tip depends on crack motion only through the

instantaneous value of the crack tip speed, _ a, and time, t. Moreover,

this dependence is of separable form [29,30]:

K

d

I

ðt; _ aÞ ¼ k

I

ð _ aÞK

Ã

I

ðtÞ ð1Þ

where K

d

I

is the instantaneous dynamic stress intensity factor for

crack propagation, K

Ã

I

is the equilibrium stress intensity factor that

depends on the current length of the crack, the applied load, and the

history of crack extension, but not on the crack velocity. It has the

dimensions of a static stress intensity factor, but in general it is

not equal to the static stress intensity factor for a stationary crack

of the same length as the moving crack. k

I

ð _ aÞ is a function of crack

speed, and can be approximated in the form:

k

I

ð

_

aÞ %

1 À _ a=C

R

1 À0:5_ a=C

R

ð2Þ

or

k

I

ð

_

aÞ %

1 À

_

a=C

R

ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

1 À

_

a=C

D

_ ð3Þ

where

C

2

D

¼

2l

q

1Àv

1À2v

_ _

plane strain

2l

q

1

1Àv

_ _

plane stress

_

_

_

ð4Þ

A.R. Shahani, M.R. Amini Fasakhodi / Materials and Design 30 (2009) 1032–1041 1033

Here C

R

and C

D

are the Raleigh wave speed and the dilatational

wave speed, respectively. Rayleigh wave speed is a root of

equation

4a

S

a

D

À ð1 þa

2

S

Þ

2

¼ 0 ð5Þ

where

a

2

D

¼ 1 À

_

a

C

D

_ _

2

ð6Þ

a

2

S

¼ 1 À

_

a

C

S

_ _

2

ð7Þ

where C

S

is the shear wave speed given by

C

2

S

¼

l

q

ð8Þ

in which l is the shear modulus and q is the density of the material.

The crack-velocity function k

I

is a universal function for all elas-

todynamically propagating cracks. When the crack is not propagat-

ing ( _ a ¼ 0), this function takes a unit value (k

I

(0) = 1). The function

k

I

ð _ aÞ decreases monotonically with increasing crack velocity _ a

while k

I

reduces to zero for _ a ¼ C

R

. Since k

I

ð _ aÞ is the decreasing

function as explained above, the stress intensity factor drops dras-

tically for a sudden increase of the crack velocity. The details of this

discussion can be found in [30].

2.2. Dynamic fracture toughness

During unstable crack propagation, the fracture toughness of

the material is introduced as dynamic fracture toughness, and is

denoted by K

d

IC

, which is not a constant value. The quantity K

d

IC

represents the resistance of the material to crack growth. The

magnitude of K

d

IC

in a special temperature is expected to depend

on the crack speed and on the properties of the material. All iner-

tial, plasticity and rate effects are lumped into the material prop-

erty K

d

IC

.

In general, the available experimental results for dynamic crack

propagation are very limited. Results of several dynamic fracture

experiments on materials by Kobayashi and Dally [31], Rosakis

et al. [32], Zehnder and Rosakis [33] for metals and by Paxton

and Lucas [34] for polymers indicate that the material’s level of

resistance to crack advance may depend on the instantaneous

crack tip speed. These are typical of most of the data that have been

reported which suggest that K

d

IC

is roughly speed independent at

low crack speeds. But the most signiﬁcant feature of the speed-

dependency is the increasing sensitivity of dynamic fracture

toughness to crack tip speed with increasing speed.

Fig. 1 exhibits dynamic fracture toughness vs. crack tip speed

for Araldite-B polymer. In this investigation, the extrapolated data

of dynamic fracture toughness for the velocity greater than 340 m/s

are used in the analysis of fast fracture if needed.

Experimental data presented in [32], relating the dynamic frac-

ture toughness to the crack tip velocity, can be correlated by the

heuristic experimental relation suggested by Kanninen and Popelar

[35]

K

d

IC

¼

K

IA

1 À

_ a

V

l

_ _

m

ð9Þ

where K

IA

, V

l

and m are the material constants that must be deter-

mined empirically. These constants have a clear physical interpreta-

tion: V

l

corresponds to a limiting crack speed, K

IA

corresponds to the

nearly constant value in the low crack speed regime, while m is a

dimensionless shape factor.

2.3. Dynamic fracture criterion and crack tip equation of motion

In order to specify an acceptable crack tip equation of motion,

dynamic crack growth criterion is required. For crack growth pro-

cesses in materials which fail in a predominantly brittle manner, or

in which any inelastic crack tip zone is completely contained with-

in the surrounding elastic crack tip zone, the most common crack

growth criteria are the generalization of Grifﬁth’s critical energy

release rate criterion and Irwin’s critical stress intensity factor cri-

terion. According to Irwin’s criterion, a crack must grow in such a

way that the crack tip dynamic stress intensity factor is always

equal to the dynamic fracture toughness of the material which

characterizes the resistance of the material to crack advance and

must be speciﬁed:

K

d

I

½aðtÞ;

_

aðtÞ; t; load ¼ K

d

IC

ð

_

aÞ ð10Þ

where K

d

I

is the dynamic stress intensity factor, and as mentioned

earlier it depends on crack length, crack speed, time and applied

load, and K

d

IC

is the dynamic fracture toughness. In principle, K

d

I

may be determined by a pure elastodynamic analysis. In practice,

K

d

I

cannot generally be determined analytically. Thus, numerical

and optical techniques are necessary to interpret dynamic fracture

experiments.

Through the introduced equation of motion (10), crack tip posi-

tion and crack tip velocity could be found during the unstable frac-

ture process, and also crack arrest could be predicted by this

equation. According to the criterion given by (10), crack arrest oc-

curs when the stress intensity factor at the crack tip becomes smal-

ler than or equal to a critical value. This can be expressed as

K

I

6 K

d

IC

ð0Þ K

dyn

Ia

ð11Þ

where K

dyn

Ia

denotes the dynamic crack arrest toughness.

3. Physical model for dynamic crack propagation

Fig. 2 depicts a RDCB specimen geometry used in experimental

study performed by Kalthoff et al. [36], and is employed in this

study. The dimensions of the specimens are as follows: length

L = 321 mm, initial crack length a

0

= 67.8 mm, beam height

h = 63.5 mm, pin diameter d = 25 mm, distance from beam end to

pin e = 16 mm, distance from crack plane to pin b = 20 mm and

thickness B = 10 mm. They were able to deduce the dynamic stress

0

0.5

1

1.5

2

2.5

0 100 200 300 400

velocity (m/s)

d

y

n

a

m

i

c

f

r

a

c

t

u

r

e

t

o

u

g

h

n

e

s

s

(

M

P

a

.

m

^

0

.

5

)

Kalthoff et al [36]

Extrapolated data

Fig. 1. dynamic fracture toughness vs. crack tip speed for Araldite-B.

1034 A.R. Shahani, M.R. Amini Fasakhodi / Materials and Design 30 (2009) 1032–1041

intensity factor for a running crack in the polymer, Araldite-B, by

using the method of shadow patterns or caustics.

In their experiments, different values of the crack initiation

stress intensity factor K

IQ

were obtained according to the degree

of blunting of the initial crack tip. This means that the cracks were

initiated from blunted notches at initiation stress intensity factors

K

IQ

which are larger than the fracture toughness K

IC

.

Consequently, crack growth is initiated from an artiﬁcially

blunted initial crack tip so that the running crack, which propa-

gates as a sharp crack, is driven at higher speeds. But because

the load pin displacement does not change appreciably, the crack

driving force diminishes and crack arrest can occur.

In the experimental study [36], the RDCB specimens with K

IQ

values 2.32 and 1.76 MPa m

1/2

were identiﬁed as specimens 4

and 8, respectively. For convenience, the same identiﬁcation is

used in the present numerical simulation. Plane stress condition

is postulated in the analysis because of the relatively thin nature

of the specimen, and also was assumed in the experimental study

[36]. The dimensions and material properties of the wedge-loaded

RDCB specimen used in the experimental study [36] and also in

this investigation are summarized in Table 1.

Araldite-B exhibits small differences between the dynamic and

static material properties. The dynamic material properties are

used in this study, and sometimes have been compared with the

results obtained with the static ones.

As shown in Fig. 2, the RDCB specimen has geometrical symme-

try, and by applying symmetrical loading with respect to the crack

plane the conditions of mode I crack growth could be preserved in

the problem. Two common types of loading in this specimen are

ﬁxed displacement and ﬁxed load on the pins shown in Fig. 2.

The former causes the crack growth to be stopped after a speciﬁed

time, because the load pin displacement does not change and con-

sequently crack driving force diminishes, while in the latter crack

propagation will be continued until the full rupture of specimen.

4. Finite element modeling procedure for dynamic crack

propagation

This section presents a ﬁnite element analysis for modeling dy-

namic fracture problems using the remeshing technique. Quarter-

point singular isoparametric elements are used for modeling the

singular ﬁeld near the crack tip. The procedure for simulation of

dynamic crack extension is outlined in the next two parts.

4.1. Singular ﬁeld near the crack tip

Because of the symmetry, only one half of the RDCB specimen is

modeled by FEM. A ﬁnite element mesh layout of model and also

the mesh pattern generated around crack tip are illustrated in

Fig. 3. The FE standard code ANSYS 7.0 has been employed for

modeling the problem. Quadratic isoparametric triangular ele-

ments with two degrees of freedom for each node are employed

for the discretization of the model (plane 2, ANSYS 7.0). These ele-

ments are capable of constructing singular elements, which are

used for modeling the familiar square root singularity ðr

À1=2

Þ at

the crack tip in elastic fracture mechanics. This can be easily done

by shifting the mid-side nodes of the six-node element to the quar-

ter-point positions [37].

Another consideration for using the quadratic six-node triangu-

lar element is that singular form of these elements leads to far bet-

ter results (stress intensity factors) than rectangular elements for

elastic fracture [37].

4.2. Remeshing algorithm

Due to the fact that fast fracture is intrinsically a dynamic anal-

ysis and also rupture phenomenon causes the geometry of the

model to be time dependent, ANSYS Parametric Design Language,

APDL, is employed for creating the fully automatic program to sim-

ulate the problem without any user interaction through the

process.

Fig. 2. Geometry of RDCB specimen used by Kalthoff et al. [36].

Table 1

Elastic properties of material Araldite-B [36]

Static elastic modulus, E

s

3380 MN/m

2

Dynamic elastic modulus, E

D

3660 MN/m

2

Static Poisson’s ratio, m

S

0.33

Dynamic Poisson’s ratio, m

D

0.39

Density, q 1172 kg/m

3

Dynamic bar wave speed, C

0

¼

ﬃﬃﬃﬃﬃﬃﬃﬃﬃ

E=q

_

1767 m/s

Fig. 3. Finite element model of the problem: (a) the entire model of half-RDCB; (b) the meshes generated around the crack tip.

A.R. Shahani, M.R. Amini Fasakhodi / Materials and Design 30 (2009) 1032–1041 1035

Fig. 4 shows the Flow-chart of the prepared APDL code of the

problem based on the remeshing technique. Transient analysis of

the problem is carried out via the Newmark scheme by choosing

time integration parameters d = 0.5 and b = 0.25. According to the

algorithm, after initial geometrical and physical modeling of the

problem, mesh generation of the model is accomplished via the

combination of free and manual methods in the software, and then

the dynamic analysis of the problem is performed. Finally, at the

end of every time increment Dt, stress intensity factor is computed

and the crack tip velocity is predicted by the dynamic fracture cri-

terion which includes the material property K

d

IC

: In order to ﬁnd

new crack tip position, the magnitude of _ aDt is computed, where

_ a is the current crack tip velocity. At this step, remeshing is accom-

plished according to the new crack tip position. In fact, remeshing

starts from a rosette of singular elements around the crack tip and

a new mesh generation with this start point is done. Subsequently,

the necessary data are transferred from the previous mesh to the

new mesh, and is interpolated to the nodal values of the present

mesh. Since the problem is ﬁxed in time during the remeshing step

and only the mesh is being changed, the shape functions are used

to interpolate the nodal data. The algorithm is repeated through

the mentioned steps during the analysis.

5. Results and discussion

In this part, some numerical examples are considered. All the

specimens are of the same geometry and material except in the de-

gree of crack tip blunting, which have been used in the experimen-

tal study [36], numerical analysis such as moving ﬁnite element

method by Nishioka and Atluri [12] and moving ﬁnite element

Setting the geometrical and input material properties data of the problem

Start

Discretization of the model by Plane2 elements

Applying the load and boundary conditions

Structural analysis of the dynamic problem

Computing the Stress Intensity Factor, K

I

K

I

> K

ID

(0)

Remeshing is accomplished to the base of the new crack tip position

Interpolation of the previous mesh nodal data for finding the corresponding

data of the present mesh

STOP

An arrest

phenomenon

has happened.

Yes

No

Computation of dynamic stress intensity factor and then finding the crack

tip velocity and amount of movement using the dynamic fracture criterion

Setting the dynamic parameters of transient analysis

Fig. 4. Flow-chart of the FEM algorithm of the problem.

1036 A.R. Shahani, M.R. Amini Fasakhodi / Materials and Design 30 (2009) 1032–1041

0

0.4

0.8

1.2

1.6

2

0 100 200 300 400 500

Time (micro second)

d

y

n

a

m

i

c

S

I

F

(

M

P

a

.

m

^

0

.

5

)

present study with dynamic properties

experimental by Kalthoff et al [36]

Nishioka & Atluri [12]

Koh et al [15]

Kobayashi [39]

Fig. 5. Variation of dynamic stress intensity factor vs. time for RDCB-4.

0

0.5

1

1.5

2

2.5

60 100 140 180 220

Crack length (mm)

d

y

n

a

m

i

c

S

I

F

(

M

P

a

.

m

^

0

.

5

)

Present study with dynamic properties

Experimental by Kalthoff et al [36]

Gehlen et al [38] Without torsional spring

Gehlen et al [38] With torsional spring

Koh et al [15]

Fig. 6. Variation of dynamic stress intensity factor vs. crack length for RDCB-4.

0

0.4

0.8

1.2

1.6

2

0 100 200 300 400 500

Time (micro second)

d

y

n

a

m

i

c

S

I

F

(

M

P

a

.

m

^

0

.

5

)

Experimental Results by Kaltoff et al [36]

Present study with dynamic properties

Present study with static properties

Fig. 7. Dynamic stress intensity factor vs. time for RDCB-4 based on the dynamic

and static material properties.

0

100

200

300

400

0 100 200 300 400 500 600

Time (micro second)

v

e

l

o

c

i

t

y

(

m

/

s

)

Present study with dynamic properties

Experimental by Kalthoff et al [36]

Koh et al [15]

Gehlen et al [38] without torsional spring

Nishioka & Atluri [12]

Fig. 9. History of crack tip velocity for RDCB-4.

0

0.1

0.2

0.3

0.4

0 100 200 300 400 500 600 700

time (micro second)

k

i

n

e

t

i

c

&

s

t

r

a

i

n

e

n

e

r

g

y

(

J

o

u

l

e

)

Strain energy with dynamic properties, Present study

Strain energy with static properties, Present study

Kinetic energy with static properties, Present study

Kinetic energy with dynamic properties, Present study

Strain energy by Gehlen et al [38]

Kinetic energy by Gehlen et al [38]

Kinetic energy by Nishioka & Atluri [12]

Strain energy by Nishioka & Atluri [12]

Fig. 10. History of energy quantities for RDCB-4.

60

100

140

180

220

0 100 200 300 400 500 600

time (micro second)

c

r

a

c

k

l

e

n

g

t

h

(

m

m

)

Present by dynamic properties

Experimental by Kalthoff et al [36]

Koh et al [15]

Nishioka & Atluri [12]

Fig. 8. History of crack extension for RDCB-4.

A.R. Shahani, M.R. Amini Fasakhodi / Materials and Design 30 (2009) 1032–1041 1037

method based on the ELD by Koh et al. [15]. The results presented

in [12] and [15] were obtained based on the static and dynamic

material properties, respectively. Gehlen et al. [38] investigated

the rapid crack growth in these specimens, and obtained equations

of motion by theory implementation based on the one-dimen-

sional assumption. The ﬁnal achieved equations by their approach

were solved via the ﬁnite difference method. In the following, the

obtained results of this study are compared with those obtained by

the above researchers showing good agreement.

5.1. Results for RDCB-4 specimen

This specimen has a degree of crack tip blunting correspond-

ing to the initiation stress intensity factor K

Q

= 2.32 MPa m

1/2

.

Fig. 5 shows the time variations of the dynamic stress intensity

factor for RDCB-4 specimen. The results have been compared

with the previous numerical results cited in the literature, Nish-

ioka and Atluri [12], Koh et al. [15] and Kobayashi [39] as well

as the experimental results of Kalthoff et al. [36]. In principle,

the dynamic stress intensity factor predicted by this study

agrees well with the experimental results particularly in the

range of time between 70–230 ls and 370 ls to the arrest time.

It is also seen that the dynamic stress intensity factor obtained

in this study, Nishioka and Atluri [12] and Kobayashi [39], shows

nearly the same trend in behavior where they have a maximum

value at the start of crack growth, and thereafter decreasing and

again reaching to a local maximum value at the instant of about

290 ls. The behavior of dynamic stress intensity factor curve at

the early stage could not be conﬁrmed by the experimental re-

sults due to lack of experimental results in this interval. In addi-

tion, the arrest time predicted by this study shows a very good

agreement with the experimental study, and has a value about

504 ls. This value for the work done by Koh et al. [15], Nishioka

and Atluri [12] and Kobayashi [39] is 438 ls, 528 ls and 497 ls,

respectively.

Fig. 6 shows the variations of the dynamic stress intensity factor

with respect to crack length for this specimen. It should be noted

that the crack length is measured from the pin location. It is seen

Fig. 11. Variation of mesh layout during dynamic crack propagation in RDCB-4 specimen.

1038 A.R. Shahani, M.R. Amini Fasakhodi / Materials and Design 30 (2009) 1032–1041

that the best estimation for the crack arrest with reference to the

experimental study corresponds to this study.

Fig. 7 shows the results for the dynamic stress intensity factor

vs. time obtained based on the dynamic and static material proper-

ties separately. Although the results are nearly similar the predic-

tions using the static elastic properties compare generally less

favorably than those based upon the dynamic properties in this

specimen.

The crack length history and crack tip velocity history are

shown in Figs. 8 and 9, respectively. According to Fig. 9, the behav-

ior of crack tip velocity curve is similar to the dynamic stress inten-

sity factor curve, and has a local maximum at a time about 290 ls.

It is seen that prior to arrest the crack velocity decreases with an

abrupt slope from the local maximum value. It is also seen from

Fig. 9 that the predicted results via the different methods agree

well with each other with the exception of Gehlen et al.’s method

[38], which shows the maximum difference at the arrest time. In

general, the crack tip velocity predicted by this study and the

method of [12] is close to each other.

Energy quantities such as kinetic energy and strain energy dur-

ing dynamic fracture process with the ﬁxed displacement loading

condition are depicted in Fig. 10. The ﬁgure shows the results ob-

tained by Nishioka and Atluri [12] which is in good agreement with

those of this study. Fig. 10 also contains the kinetic energy and

strain energy values obtained based on the dynamic material prop-

erties by Gehlen et al. formulation without using torsional spring

in the equations [38]. As shown, their results are in the higher level

of magnitude and time process among the methods. As shown in

Fig. 10, the value of kinetic energy is zero at the beginning of prop-

agation because the model is in the static conditions on the thresh-

old of crack propagation. The magnitude of kinetic energy starts to

increase parallel to the crack advancement and reaches a maxi-

mum value at the time about 140 ls, and thereafter begins to de-

crease in the remainder path until arrival at the zero value at the

arrest time. This is the time that the model again reaches a static

condition because of crack arrest. Owing to loading condition in

this problem, the maximum value of strain energy occurs on the

threshold of crack advancement, and thereafter decreases through-

out the analysis until the arrest time where its magnitude is about

0.0287 joule at this instant for RDCB-4. For this specimen, whether

using dynamic or using static material properties the maximum ki-

netic energy is 26.4% of the initial strain energy which does not

indicate that the event occurs statically. Decrease of both types

of energy, however, is because of energy release due to the crack

advancing.

Fig. 11 shows the mesh details and the variation of mesh layout

during the dynamic fracture simulation of RDCB-4 specimen. It is

observed that because of using remeshing technique, the ﬁnite ele-

ment analysis of the problem is accomplished with the well-

shaped elements, and the optimum mesh is generated each time

during the crack propagation. Moreover, the singular elements

are always at the crack tip position.

5.2. Results for RDCB-8 specimen

This specimen has a degree of crack tip bluntness that corre-

sponds to the initiation stress intensity factor K

Q

= 1.76 MPa m

1/2

.

This specimen is analyzed based on the dynamic material proper-

ties only. Fig. 12 shows the variation of dynamic stress intensity

factor vs. crack length obtained from the present method along

with the existent results for this specimen, which includes the

experimental results [36] and numerical results of [15]. The curve

obtained for this case shows a trend similar to that presented for

RDCB-4, and comprises a maximum value at the start of crack

growth. Despite the prediction of Koh et al. [15], the predicted

crack arrest length via the present method is nearly identical with

that measured in the experimental study [36], and is about

170 mm.

Figs. 13 and 14 show the crack length and energy history,

respectively. It is seen that there is a good agreement between

crack length histories. Also, the arrest time predicted by this study

and [15] is identical, and has a value equal to 369 ls.

The time duration obtained by the experimental study [36] is

larger with respect to the value computed by this investigation.

According to Fig. 14, the maximum value of strain energy for

RDCB-8 is 0.16 joule and corresponds to the magnitude of critical

applying load (ﬁxed displacement) for initial crack advancement.

This quantity was about 0.3 joule for RDCB-4 specimen. But the

amount of strain energy at the arrest time is 0.0276 joule which

is nearly equal to the similar value of RDCB-4 at the arrest time.

For this specimen, the maximum kinetic energy is 21.6% of the ini-

tial strain energy which is 4.8% less than the ﬁrst studied type of

specimen, and does not indicate that the event occurs statically.

Fig. 15 shows the crack tip velocity vs. crack length for RDCB-8

which represents the comparison of the present results with the

experimental results.

0

0.4

0.8

1.2

1.6

60 90 120 150 180

crack length (mm)

d

y

n

a

m

i

c

S

I

F

(

M

P

a

.

m

^

0

.

5

)

Present study with dynamic properties

Experimental results by Kalthoff et al [36]

Koh et al [15]

Fig. 12. Variation of dynamic stress intensity factor vs. crack length for RDCB-8.

60

90

120

150

180

0 100 200 300 400 500 600

time (micro second)

c

r

a

c

k

l

e

n

g

t

h

(

m

m

)

Present study with dynamic properties

Experimental results by Kalthoff et al [36]

Koh et al [15]

Fig. 13. History of crack extension for RDCB-8.

A.R. Shahani, M.R. Amini Fasakhodi / Materials and Design 30 (2009) 1032–1041 1039

5.3. Crack velocity–crack length relation

A further comparison could be implemented for more assurance

of the predicted crack arrest length using the hypothesis presented

by Hahn et al. [40] and Shmuely [41]. Their theoretical calculations

predict a unique relation between the length of the crack at arrest

and the maximum constant crack velocity which depends only on

the geometry of the specimen but not on the toughness of the

material. Both theoretical curves in a properly normalized form

are shown in Fig. 16. The present results together with the exper-

imental results of Kalthoff et al. [36] for both RDCB-4 and RDCB-8

specimens are summarized in Table 2 in order to compare them

with the predicted results by Shmuely [41] and Hahn et al. [40]

arisen from Fig. 16. Here, c

0

is the bar wave speed in the material.

This study yields to the values of 0.1726 and 0.1613 for

Vmax

c

0

in

the RDCB-4 and RDCB-8 specimens, respectively, the crack arrest

lengths of which are presented in the table. The predicted crack ar-

rest lengths corresponding to the same values of

Vmax

c

0

have been ex-

tracted from Fig. 16 and presented in other rows of the same table.

It is seen that a relatively good agreement exists between the re-

sults. Nevertheless, the experimental results [36] assure the pre-

ciseness of the results of this investigation over those of Hahn

et al. [40] and Shmuely [41].

6. Conclusion

The problem of dynamic crack propagation in RDCB specimen,

made of a brittle material, has been analyzed. A ﬁnite element

analysis based on the remeshing technique has been used to sim-

ulate the crack growth during the fracture process. The remeshing

technique has been preferred among the traditional methods, and

this is due to the fact that when it combines with the singular ele-

ments it could model the crack tip singularity and conserve it

through the crack advancement. Moreover, a dynamic fracture cri-

terion, which includes the dynamic fracture toughness, has been

employed in order to extract the crack growth velocity at each time

step. In general, fast fracture mechanics analysis deals with the

parameters such as dynamic stress intensity factor, history of crack

propagation and velocity, energy quantities and crack arrest

length. The aforementioned parameters in this study have shown

good agreement with the experimental results as compared with

the other numerical results published in the literature. The follow-

ing conclusions could be drawn from the above-mentioned unsta-

ble crack growth analysis:

With decreasing crack tip blunting, the initial stored strain

energy in the specimen decreases causing the crack arrest length

and time to decrease.

The ratio of the maximum kinetic energy to the initial strain

energy in the specimen is considerable, and this implies that

the dynamic effects are dominant in the unstable crack propaga-

tion phenomenon and should be taken into account in the

analysis.

Table 2

Summarized results of different predictions for RDCB-4 and RDCB-8 specimens

RDCB-4 RDCB-8

_ amax;cons=c

0

aarrest=a

0

_ amax;cons=c

0

aarrest=a

0

Present study 0.1726 3.038 0.1613 2.51917

Predicted by Shmuely [41] 0.1726 2.9516 0.1613 2.6926

Predicted by Hahn et al. [40] 0.1726 2.7716 0.1613 2.5726

Experimental by Kalthoff et al.

[36]

0.16978 2.9522 0.155 2.51917

0

50

100

150

200

250

300

350

67 87 107 127 147 167

crack length (mm)

v

e

l

o

c

i

t

y

(

m

/

s

)

Present study with dynamic properties

Experimental results by Kalthoff et al [36]

Fig. 15. Variation of crack tip velocity vs. crack length for RDCB-8.

0

0.05

0.1

0.15

0.2

0 2 4

relative crack length, a(arrest)/a0

r

e

l

a

t

i

v

e

c

r

a

c

k

v

e

l

o

c

i

t

y

,

(

V

m

a

x

/

c

0

)

Predicted by

Shmuely [41]

Predicted by Hahn

et al [40]

1 3

Fig. 16. Relation between maximum constant velocity and crack length at arrest.

0

0.04

0.08

0.12

0.16

0 50 100 150 200 250 300 350 400

time (micro second)

s

t

r

a

i

n

&

k

i

n

e

t

i

c

e

n

e

r

g

y

(

j

o

u

l

e

)

strain energy with dynamic properties, Present study

kinetic energy with dynamic properties, Present study

Fig. 14. Variation of energy quantities vs. time for RDCB-8.

1040 A.R. Shahani, M.R. Amini Fasakhodi / Materials and Design 30 (2009) 1032–1041

The size of blunting at the crack tip inﬂuences the level of crack

growth velocity, the magnitude of crack arrest length and also

the arrest time in the ﬁxed-displacement loading condition.

The strain energy related to the arrest point for both specimens

analyzed here, i.e., RDCB-4 and RDCB-8, is nearly identical, and

this implies that the arrest occurs when the strain energy

reaches a pronounced value related to the value of crack arrest

toughness.

The accuracy of this investigation is more noticeable when it

concerns with the crack arrest length or arrest time predicted

in the specimens.

References

[1] Berry JP. Some kinetic considerations of the Grifﬁth criterion for fracture I.

Equations of motion at constant force. J Mech Phys Solids 1960;8:194–206.

[2] Berry JP. Some kinetic considerations of the Grifﬁth criterion for fracture II.

Equations of motion at constant deformation. J Mech Phys Solids

1960;8:207–16.

[3] Kanninen MF. A dynamic analysis of unstable crack propagation and arrest in

DCB test specimen. Int J Fract 1974;10(3):415–30.

[4] Freund LB. A simple model of the double cantilever beam crack propagation

specimen. J Mech Phys Solids 1977;25:69–79.

[5] Malluck JF, King WW. Simulation of fast fracture in the DCB specimen using

Kanninen’s model. Int J Fract 1977;13(5):655–65.

[6] Shahani AR, Forqani M. Static and dynamic fracture mechanics analysis of a

DCB specimen considering shear deformation effects. Int J Solids Struct

2004;41:3793–807.

[7] Yagawa G, Sakai Y, Ando Y. In: Hann GT, Kanninen MF, editors. Fast fracture

and crack arrest. ASTM STP, vol. 627; 1977. p. 109–22.

[8] Keegstra PNR, Head JL, Turner CE. In: Luxmoore AR, Owen DRJ, editors.

Numerical methods in fracture mechanics. University College, Swensea; 1978.

p. 634–47.

[9] Caldis ES, Owen DRJ, Zienkiewicz OC. Nonlinear dynamic transient methods in

crack propagation studies. In: Nonlinear and dynamic fracture mechanics.

ASME AMD, vol. 35; 1979. p. 1–17.

[10] Kanninen MF. A critical appraisal of solution techniques in dynamic fracture

mechanics. In: Luxmore AR, Owen DR, editors. Numerical methods in fracture

mechanics. Swansea (UK): Pineridge Press; 1978. p. 612–33.

[11] Nishioka T, Atluri SN. Numerical modeling of dynamic crack propagation in

ﬁnite bodies by moving singular elements. Part I. J Appl Mech

1980;47(3):570–6.

[12] Nishioka T, Atluri SN. Numerical analysis of dynamic crack propagation:

generation and prediction studies. Eng Fract Mech 1982;16(3):303–32.

[13] Koh HM, Haber RB. Elastodynamic formulation of the Eulerian–Lagrangian

kinematic description. J Appl Mech 1986;53:839–44.

[14] Koh HM, Lee SH, Haber RB. Dynamic crack propagation analysis using

Eulerian–Lagrangian kinematics descriptions. Comput Mech 1988;3:141–55.

[15] Koh HM, Lee SH, Jeong UY. An incremental formulation of the moving grid

ﬁnite element method for the prediction of dynamic crack propagation. Nucl

Eng Des 1995;158:295–309.

[16] Belytschko T, Lu YY, Gu L, Tabbara M. Element-free Galerkin methods for static

and dynamic fracture. Int J Solids Struct 1995;32(17/18):2547–70.

[17] Belytschko T, Tabbara M. Dynamic fracture using element-free Galerkin

methods. Int J Numer Meth Eng 1996;39:923–38.

[18] Beissel SR, Johnson GR, Popelar CH. An element failure algorithm for

dynamic crack propagation in general directions. Eng Fract Mech 1998;61:

407–25.

[19] Lim IL, Johnston IW, Choi SK. Application of quadratic singular distorted

isoparametric elements to linear fracture mechanics. Int J Numer Meth Eng

1993;36:2473–99.

[20] Shephard MS, Yehia NAB, Burd GS, Weidner TJ. Automatic crack propagation

tracking. Compos Struct 1985;20:211–23.

[21] Sumi Y. Computational crack path prediction. Theor Appl Fract Mech

1985;4:149–56.

[22] Wawrzynek PA, Ingraffea AR. Interactive ﬁnite element analysis of fracture

processes: an integrated approach. Theor Appl Fract Mech 1987;8:137–50.

[23] Wawrzynek PA, Ingraffea AR. An Interactive approach to local remeshing

around a propagating crack. Fin Elem Anal Des 1989;5:87–96.

[24] Swenson DV, Ingraffea AR. Modeling mixed-mode dynamic crack propagation

using ﬁnite elements, theory and applications. Comput Mech 1988;3:381–97.

[25] Bittencourt TN, Wawrzynek PA, Ingraffea AR, Sousa JL. Quasi-automatic

simulation for 2D LEFM problems. Eng Fract Mech 1996;55(2):321–34.

[26] Tradegard A, Nilsson F, Ostlund S. FEM-remeshing technique applied to crack

growth problems. Comput Meth Appl Mech Eng 1998;160:115–31.

[27] Rethore J, Gravouil A, Combescure A. A stable numerical scheme for the ﬁnite

element simulation of dynamic crack propagation with remeshing. Comput

Meth Appl Mech Eng 2004;193:4493–510.

[28] Shahani AR, Seyyedian M. Simulation of glass cutting with an impinging hot air

jet. Int J Solids Struct 2004;41:1313–29.

[29] Rose LRF. Recent theoretical and experimental results on fast brittle fracture.

Int J Fract 1975;12(6):799–813.

[30] Freund LB. Dynamic fracture mechanics. Berlin: Cambridge University Press;

1998.

[31] Kobayashi T, Dally JW. In: Hahn GT, Kanninen MF, editors. Crack arrest

methodology and applications, ASTM STP 711. Philadelphia: American Society

of Testing and Materials; 1979. p. 189–210.

[32] Rosakis AJ, Duffy J, Freund LB. The determination of dynamic fracture

toughness of AISI 4340 steel by the shadow spot method. J Mech Phys Solids

1984;3(4):443–60.

[33] Zehnder AT, Rosakis AJ. Dynamic fracture initiation and propagation in 4340-

steel under impact loading. Int J Fract 1990;43:271–85.

[34] Paxton TL, Lucas RA. In: Sih GC, editor. An experimental investigation of the

velocity characteristics of a ﬁxed boundary fracture, dynamic crack

propagation. Leiden: Noordhoff; 1973. p. 415–26.

[35] Kanninen MF, Papelar CH. Advanced fracture mechanics. New-York: Oxford

University Press; 1985.

[36] Kalthoff JF, Beinert J, Winkler S. Measurement of dynamic stress intensity

factors for fast running and arresting cracks in double-cantilever-beam

specimens. In: Hahn GT, Kanninen MF, editors. Fast fracture and crack

arrest, ASTM STP 627, ASTM; 1977. p. 161–76.

[37] Barsoum RS. Triangular quarter-point elements as elastic and perfectly-plastic

crack tip elements. Int J Numer Meth Eng 1977;11:58–98.

[38] Gehlen PC, Popelar CH, Kanninen MF. Modeling of dynamic crack

propagation, I: Validation of one dimensional analysis. Int J Fract 1979;15(3):

281–94.

[39] Kobayashi AS. Dynamic fracture analysis by dynamic ﬁnite element method

generation and propagation analysis. In: Perrone N, Atluri SN, editors.

Nonlinear and dynamic fracture mechanics, ASME, AMD, vol. 35; 1979. p.

19–37.

[40] Hahn GT, Hoagland RG, Kanninen MF, Rosenﬁeld AR, Sejnoha R. Fast fracture

resistance and crack arrest in structural steels. SSC-progress report on project

SR-201. Battelle Columbus Laboratories, Columbus, Ohio; 1973.

[41] Shmuely M. Analysis of fast fracture and crack arrest by ﬁnite differences. Int J

Fract 1977;13(4):443–54.

A.R. Shahani, M.R. Amini Fasakhodi / Materials and Design 30 (2009) 1032–1041 1041

In elastic–plastic materials. and the history of crack extension. The ﬁnite element modeling is accomplished with the standard FE code ANSYS 7. the applied load. neither in geometry nor in loading conditions. ANSYS Parametric Design Language. Since the dynamic crack growth is intrinsically an unstable phenomenon. However. Owing to timedependent nature of the problem and inertia effects. but in general it is not equal to the static stress intensity factor for a stationary crack _ of the same length as the moving crack. Another important consideration in any such remeshing algorithm is the application of elements to adequately model the crack tip singularity [19]. and is achieved by tracking the path of the crack tip and failing the elements crossed by the path such that they can no longer sustain deviatoric or tensile volumetric stresses. this dependence is of separable form [29. The element-free Galerkin (EFG) method [16. In principle. Amini Fasakhodi / Materials and Design 30 (2009) 1032–1041 1033 matrix. and time.30]: _ _ I K d ðt. the crack length and the crack tip velocity. the instantaneous value of the crack tip stress intensity factor for an arbitrary motion of the crack tip depends on crack motion only through the _ instantaneous value of the crack tip speed. Moreover. It has the dimensions of a static stress intensity factor. automatically. Since the crack is to propagate. the generated elements may not be as well-shaped as in the ﬁrst approach since there tend to be greater constraints on the placement of the new elements due to the existing mesh surrounding the region to be remeshed. Koh et al.R. [15] applied ELD approach in ﬁnite element method to predict the dynamic fracture in RDCB specimens. its disadvantage is that a large number of state variables must be transferred from the old to the new mesh. crack growth increment cannot be deﬁned by the user. Also this technique is not restricted to the special mode in fracture mechanics. [18] introduced an element failure algorithm for crack growth in general direction. The predicted results show good agreement with those obtained via the experimental study and the other numerical techniques. the aim of their work is not to develop an efﬁcient remeshing procedure. This algorithm does not require remeshing technique. and a dynamic fracture criterion should be employed to ﬁnd the crack tip velocity and crack extension every time.R. Shahani and Seyyedian [28] employed an entire remeshing approach in order to simulate glass cutting with the impinging hot air jet which could be interpreted as a controlled crack growth due to thermal stresses caused by the hot air jet. However. is employed. [27] employed a stable numerical scheme for the analysis of dynamic crack growth with remeshing but their study focused on the stability subject of dynamic calculations.23] and simulation of dynamic crack propagation in an inﬁnite medium by Swenson and Ingraffea [24]. Tradegard et al. In this regard. and can be approximated in the form: _ kI ðaÞ % or _ 1 À a=C R _ 1 À 0:5a=C R ð2Þ _ 1 À a=C R _ kI ðaÞ % pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ _ 1 À a=C D where ð3Þ C2 D 8 À Á < 2l 1Àv plane strain q 1À2v ¼ À 1 Á 2l : plane stress q 1Àv ð4Þ . [25] presented an algorithm named quasiautomatic simulation of stable crack propagation for two-dimensional LEFM problems in order to predict crack trajectory that is very similar to the work done by Swenson and Ingraffea [24] in remeshing algorithm. Imposition of the geometrical boundary conditions by the EFG technique is more complicated than for the FEM.0. The second approach employs a local remeshing technique whereby only the region of elements within a certain distance of the crack tip is modiﬁed. and this is for the sake of being a stable crack growth. kI ðaÞ is a function of crack speed. particularly when an elasto-plastic material is involved. Dynamic fracture mechanics has been used as a basis for their study. and solving these equations greatly increases the required time.17] is another technique which. but not on the crack velocity. In this study. and the crack growth speed has been calculated via the criterion based on the energy release rate considering that the material toughness is independent of the crack speed. It requires only nodal data. nodal data should be transferred from the old mesh to the new mesh in each step of the remeshing. 2. the crack tip stress intensity factor represents the effect of the applied loading. In the LEFM conditions. One of the unique features of any discrete fracture propagation ﬁnite element code is its remeshing algorithm. [26] employed a combination of entire remeshing and nodal relaxation to study mode I stable crack propagation in an elastic– plastic material by ABAQUS software. aimed at the simulation of crack growth problems. For this to be achieved. and no element connectivity is needed. and analysis is performed via the single and ﬁxed mesh pattern during the entire process. the attempt is made toward gaining a ﬁnite element simulation based on the remeshing technique compatible with the actual dynamic fracture process. can be simulated. the problem is analyzed for two different crack tip bluntings to investigate its effects on crack arrest length. This approach has the advantage of requiring only the transference of state variables from a small region which is to be remeshed. Singular elements cannot be employed in this algorithm because the model has not been remeshed. Recently. and due to this ability crack propagation in problems which have no symmetry. Some crack propagation analysis based on the remeshing technique can be found in the previous works. [20] and Sumi [21]. the geometrical conﬁguration of the body and the bulk material parameters in the crack tip region for any motion of the crack tip. In principle. these parameters depend on time. aÞ ¼ kI ðaÞK Ã ðtÞ I Kd I ð1Þ where is the instantaneous dynamic stress intensity factor for crack propagation. The advantage of this approach is that well shaped elements can usually be generated. The ﬁrst is to remesh the entire model in every step as advocated by Shephard et al. In this paper. Shahani. Beissel et al. Dynamic crack propagation analysis 2. Rethore et al. mode I dynamic crack propagation and arrest phenomena are investigated in the RDCB specimens under ﬁxed displacement loading condition. a. The analysis in [25] was said to be quasiautomatic only because the user still needs to provide a desired crack length increment at the beginning of each simulation. crack arrest time. Dynamic stress intensity factor In general. APDL. dynamic fracture toughness criterion is used to predict the crack tip velocity. Crack propagation in elastic and elastic–plastic materials can be analyzed via this method. t.A. Finally. M. For the model problem considered here. The quarter point singular elements are used around the crack tip in order to model the singularity. the local approach was used to investigate the fatigue crack growth by Wawrzynek and Ingraffea [22. two remeshing approaches have been reported in the literature. a remeshing algorithm is used in each crack extension step. For a dynamically propagating crack under a steady-state or under an unsteady-state condition. based on the moving least-square interpolants.1. Bittencourt et al. K Ã is the equilibrium stress intensity factor that I depends on the current length of the crack. crack growth velocity and the quantity of strain energy and kinetic energy during the crack advancement.

crack speed. or in which any inelastic crack tip zone is completely contained within the surrounding elastic crack tip zone. Amini Fasakhodi / Materials and Design 30 (2009) 1032–1041 dynamic fracture toughness (MPa. The function _ _ kI ðaÞ decreases monotonically with increasing crack velocity a _ _ while kI reduces to zero for a ¼ C R . According to the criterion given by (10).R. 2. dynamic fracture toughness vs.5 _ a CS 2 ð7Þ 1 where CS is the shear wave speed given by 0. pin diameter d = 25 mm. The details of this discussion can be found in [30]. and as mentioned earlier it depends on crack length. K d IC I may be determined by a pure elastodynamic analysis. These are typical of most of the data that have been reported which suggest that K d is roughly speed independent at IC low crack speeds. which is not a constant value.m^0. In practice. the fracture toughness of the material is introduced as dynamic fracture toughness. crack arrest occurs when the stress intensity factor at the crack tip becomes smaller than or equal to a critical value. time and applied load. Results of several dynamic fracture experiments on materials by Kobayashi and Dally [31]. K d cannot generally be determined analytically. Rosakis et al. the stress intensity factor drops drastically for a sudden increase of the crack velocity.5 Kalthoff et al [36] 2 Extrapolated data 4aS aD À ð1 þ a2 Þ2 ¼ 0 S where ð5Þ a2 ¼ 1 À D a2 ¼ 1 À S _ a CD 2 ð6Þ 1.5) Here C R and C D are the Raleigh wave speed and the dilatational wave speed. All inertial. plasticity and rate effects are lumped into the material property K d . the available experimental results for dynamic crack propagation are very limited.5 C2 ¼ S l q ð8Þ 0 0 100 200 300 400 in which l is the shear modulus and q is the density of the material. They were able to deduce the dynamic stress ð9Þ where KIA. the most common crack growth criteria are the generalization of Grifﬁth’s critical energy release rate criterion and Irwin’s critical stress intensity factor criterion.5 mm. Fig. Since kI ðaÞ is the decreasing function as explained above. KIA corresponds to the nearly constant value in the low crack speed regime. For crack growth processes in materials which fail in a predominantly brittle manner. Rayleigh wave speed is a root of equation 2.R. distance from crack plane to pin b = 20 mm and thickness B = 10 mm. The quantity K d IC IC represents the resistance of the material to crack growth. numerical I and optical techniques are necessary to interpret dynamic fracture experiments.8 mm.3. crack tip position and crack tip velocity could be found during the unstable fracture process. Dynamic fracture toughness During unstable crack propagation. According to Irwin’s criterion. aðtÞ. can be correlated by the heuristic experimental relation suggested by Kanninen and Popelar [35] velocity (m/s) Fig. Thus. Dynamic fracture criterion and crack tip equation of motion In order to specify an acceptable crack tip equation of motion. 1 exhibits dynamic fracture toughness vs. . Vl and m are the material constants that must be determined empirically. and also crack arrest could be predicted by this equation. But the most signiﬁcant feature of the speeddependency is the increasing sensitivity of dynamic fracture toughness to crack tip speed with increasing speed. In this investigation.2. Through the introduced equation of motion (10). The magnitude of K d in a special temperature is expected to depend IC on the crack speed and on the properties of the material. This can be expressed as K I 6 K d ð0Þ K dyn IC Ia where K dyn Ia denotes the dynamic crack arrest toughness. initial crack length a0 = 67. IC In general. [32]. 2 depicts a RDCB specimen geometry used in experimental study performed by Kalthoff et al. distance from beam end to pin e = 16 mm. Experimental data presented in [32]. The dimensions of the specimens are as follows: length L = 321 mm.1034 A. When the crack is not propagat_ ing (a ¼ 0). the extrapolated data of dynamic fracture toughness for the velocity greater than 340 m/s are used in the analysis of fast fracture if needed. These constants have a clear physical interpretation: Vl corresponds to a limiting crack speed. 1. 2. while m is a dimensionless shape factor. ð11Þ Kd ¼ IC K IA m a 1 À V_ l 3. In principle. Zehnder and Rosakis [33] for metals and by Paxton and Lucas [34] for polymers indicate that the material’s level of resistance to crack advance may depend on the instantaneous crack tip speed. and is denoted by K d . respectively. M. dynamic crack growth criterion is required. and is employed in this study. this function takes a unit value (kI(0) = 1). The crack-velocity function kI is a universal function for all elastodynamically propagating cracks. load ¼ K d ðaÞ I IC Kd I ð10Þ where is the dynamic stress intensity factor. crack tip speed for Araldite-B polymer. [36]. Physical model for dynamic crack propagation Fig. t. Shahani. and K d is the dynamic fracture toughness. beam height h = 63. relating the dynamic fracture toughness to the crack tip velocity. crack tip speed for Araldite-B. a crack must grow in such a way that the crack tip dynamic stress intensity factor is always equal to the dynamic fracture toughness of the material which characterizes the resistance of the material to crack advance and must be speciﬁed: _ _ K d ½aðtÞ.

A ﬁnite element mesh layout of model and also the mesh pattern generated around crack tip are illustrated in Fig. This means that the cracks were initiated from blunted notches at initiation stress intensity factors KIQ which are larger than the fracture toughness KIC. 3. Fig. Singular ﬁeld near the crack tip Because of the symmetry. C 0 ¼ E=q 3380 MN/m2 3660 MN/m2 0. 4. is employed for creating the fully automatic program to simulate the problem without any user interaction through the process. because the load pin displacement does not change and consequently crack driving force diminishes. mS Dynamic Poisson’s ratio. Plane stress condition is postulated in the analysis because of the relatively thin nature of the specimen. ANSYS Parametric Design Language. 3. [36]. Geometry of RDCB specimen used by Kalthoff et al. crack growth is initiated from an artiﬁcially blunted initial crack tip so that the running crack. 2. and also was assumed in the experimental study [36]. and sometimes have been compared with the results obtained with the static ones.33 0.32 and 1. In the experimental study [36]. q pﬃﬃﬃﬃﬃﬃﬃﬃﬃ Dynamic bar wave speed.R. the RDCB specimens with KIQ values 2. is driven at higher speeds. In their experiments.R. Amini Fasakhodi / Materials and Design 30 (2009) 1032–1041 1035 Fig. Araldite-B. The procedure for simulation of dynamic crack extension is outlined in the next two parts. As shown in Fig. Consequently.2. APDL. The former causes the crack growth to be stopped after a speciﬁed time. the crack driving force diminishes and crack arrest can occur. .39 1172 kg/m3 1767 m/s Due to the fact that fast fracture is intrinsically a dynamic analysis and also rupture phenomenon causes the geometry of the model to be time dependent. ED Static Poisson’s ratio. ANSYS 7. Es Dynamic elastic modulus. This can be easily done by shifting the mid-side nodes of the six-node element to the quarter-point positions [37]. 2. Remeshing algorithm intensity factor for a running crack in the polymer. The dynamic material properties are used in this study. which propagates as a sharp crack. which are used for modeling the familiar square root singularity ðrÀ1=2 Þ at the crack tip in elastic fracture mechanics. (b) the meshes generated around the crack tip. while in the latter crack propagation will be continued until the full rupture of specimen. These elements are capable of constructing singular elements. Finite element model of the problem: (a) the entire model of half-RDCB. Quarterpoint singular isoparametric elements are used for modeling the singular ﬁeld near the crack tip. Shahani. For convenience. mD Density.1. only one half of the RDCB specimen is modeled by FEM. the RDCB specimen has geometrical symmetry.76 MPa m1/2 were identiﬁed as specimens 4 and 8. Quadratic isoparametric triangular elements with two degrees of freedom for each node are employed for the discretization of the model (plane 2.0). 2. by using the method of shadow patterns or caustics.0 has been employed for modeling the problem. respectively. The dimensions and material properties of the wedge-loaded RDCB specimen used in the experimental study [36] and also in this investigation are summarized in Table 1.A. Finite element modeling procedure for dynamic crack propagation This section presents a ﬁnite element analysis for modeling dynamic fracture problems using the remeshing technique. 4. Araldite-B exhibits small differences between the dynamic and static material properties. different values of the crack initiation stress intensity factor KIQ were obtained according to the degree of blunting of the initial crack tip. But because the load pin displacement does not change appreciably. and by applying symmetrical loading with respect to the crack plane the conditions of mode I crack growth could be preserved in the problem. The FE standard code ANSYS 7. the same identiﬁcation is used in the present numerical simulation. M. Two common types of loading in this specimen are ﬁxed displacement and ﬁxed load on the pins shown in Fig. Another consideration for using the quadratic six-node triangular element is that singular form of these elements leads to far better results (stress intensity factors) than rectangular elements for elastic fracture [37]. 4. Table 1 Elastic properties of material Araldite-B [36] Static elastic modulus.

remeshing starts from a rosette of singular elements around the crack tip and a new mesh generation with this start point is done. mesh generation of the model is accomplished via the combination of free and manual methods in the software. According to the algorithm. remeshing is accomplished according to the new crack tip position. Subsequently. Amini Fasakhodi / Materials and Design 30 (2009) 1032–1041 Start Setting the geometrical and input material properties data of the problem Discretization of the model by Plane2 elements Setting the dynamic parameters of transient analysis Applying the load and boundary conditions Structural analysis of the dynamic problem Computing the Stress Intensity Factor. and is interpolated to the nodal values of the present mesh. numerical analysis such as moving ﬁnite element method by Nishioka and Atluri [12] and moving ﬁnite element . which have been used in the experimental study [36]. the magnitude of a Dt is computed. at the end of every time increment Dt. KI No KI > KID (0) Yes Computation of dynamic stress intensity factor and then finding the crack tip velocity and amount of movement using the dynamic fracture criterion An arrest phenomenon has happened. Shahani. In fact. 4. At this step. 4 shows the Flow-chart of the prepared APDL code of the problem based on the remeshing technique.R. and then the dynamic analysis of the problem is performed. after initial geometrical and physical modeling of the problem. 5. Transient analysis of the problem is carried out via the Newmark scheme by choosing time integration parameters d = 0.R. the shape functions are used to interpolate the nodal data. where _ a is the current crack tip velocity.25. Fig. Remeshing is accomplished to the base of the new crack tip position Interpolation of the previous mesh nodal data for finding the corresponding data of the present mesh STOP Fig. The algorithm is repeated through the mentioned steps during the analysis.5 and b = 0. Since the problem is ﬁxed in time during the remeshing step and only the mesh is being changed.1036 A. All the specimens are of the same geometry and material except in the degree of crack tip blunting. stress intensity factor is computed and the crack tip velocity is predicted by the dynamic fracture criterion which includes the material property K d : In order to ﬁnd IC _ new crack tip position. the necessary data are transferred from the previous mesh to the new mesh. Results and discussion In this part. Finally. some numerical examples are considered. Flow-chart of the FEM algorithm of the problem. M.

5.6 180 1. Present study Strain energy with static properties.3 1.5 Present study with dynamic properties Experimental by Kalthoff et al [36] Gehlen et al [38] Without torsional spring Gehlen et al [38] With torsional spring Koh et al [15] 400 dynamic SIF (MPa. 2.4 100 Present by dynamic properties Experimental by Kalthoff et al [36] Koh et al [15] Nishioka & Atluri [12] 0 0 100 200 300 400 500 60 0 100 200 300 400 500 600 Time (micro second) Fig.m^0.2 0.2 0.2 crack length (mm) 140 0. 9.5 Present study with dynamic properties Experimental by Kalthoff et al [36] Koh et al [15] Gehlen et al [38] without torsional spring Nishioka & Atluri [12] 0 60 100 140 180 220 0 0 100 200 300 400 500 600 Crack length (mm) Fig.R. . time for RDCB-4 based on the dynamic and static material properties.8 present study with dynamic properties experimental by Kalthoff et al [36] Nishioka & Atluri [12] Koh et al [15] Kobayashi [39] 0.R. Present study Kinetic energy with static properties.5) 2 300 1. History of energy quantities for RDCB-4. Time (micro second) Fig.4 0. time for RDCB-4. Amini Fasakhodi / Materials and Design 30 (2009) 1032–1041 1037 220 2 dynamic SIF (MPa. Present study Strain energy by Gehlen et al [38] Kinetic energy by Gehlen et al [38] Kinetic energy by Nishioka & Atluri [12] Strain energy by Nishioka & Atluri [12] dynamic SIF (MPa. 6.m^0.8 Experimental Results by Kaltoff et al [36] Present study with dynamic properties Present study with static properties 0. History of crack extension for RDCB-4.5) kinetic & strain energy (Joule) 500 0. time (micro second) Fig. 10.5 velocity (m/s) 200 1 100 0.m^0. Variation of dynamic stress intensity factor vs. time (micro second) Fig. Present study Kinetic energy with dynamic properties. crack length for RDCB-4. 2 0. 7. Variation of dynamic stress intensity factor vs. Shahani.6 Strain energy with dynamic properties. History of crack tip velocity for RDCB-4.1 0 0 100 200 300 400 0 0 100 200 300 400 500 600 700 Time (micro second) Fig. Dynamic stress intensity factor vs.5) 1. M. 8.A.4 1.

5 shows the time variations of the dynamic stress intensity factor for RDCB-4 specimen.1. 5. Gehlen et al. In principle. Koh et al. In addition. [15] and Kobayashi [39] as well as the experimental results of Kalthoff et al. Variation of mesh layout during dynamic crack propagation in RDCB-4 specimen. 528 ls and 497 ls. [38] investigated the rapid crack growth in these specimens.32 MPa m1/2. and obtained equations of motion by theory implementation based on the one-dimensional assumption. M. Nishioka and Atluri [12] and Kobayashi [39]. Fig. [15].1038 A. respectively. This value for the work done by Koh et al. [15].R. It is seen Fig. the dynamic stress intensity factor predicted by this study agrees well with the experimental results particularly in the range of time between 70–230 ls and 370 ls to the arrest time. Nishioka and Atluri [12] and Kobayashi [39] is 438 ls. Fig. shows nearly the same trend in behavior where they have a maximum value at the start of crack growth. and thereafter decreasing and again reaching to a local maximum value at the instant of about 290 ls. . In the following. the arrest time predicted by this study shows a very good agreement with the experimental study. 6 shows the variations of the dynamic stress intensity factor with respect to crack length for this specimen. The results presented in [12] and [15] were obtained based on the static and dynamic material properties. [36]. It should be noted that the crack length is measured from the pin location. respectively. Amini Fasakhodi / Materials and Design 30 (2009) 1032–1041 method based on the ELD by Koh et al. and has a value about 504 ls. Nishioka and Atluri [12]. The results have been compared with the previous numerical results cited in the literature. It is also seen that the dynamic stress intensity factor obtained in this study. The ﬁnal achieved equations by their approach were solved via the ﬁnite difference method. The behavior of dynamic stress intensity factor curve at the early stage could not be conﬁrmed by the experimental results due to lack of experimental results in this interval.R. Results for RDCB-4 specimen This specimen has a degree of crack tip blunting corresponding to the initiation stress intensity factor KQ = 2. the obtained results of this study are compared with those obtained by the above researchers showing good agreement. Shahani. 11.

0287 joule at this instant for RDCB-4. and thereafter begins to decrease in the remainder path until arrival at the zero value at the arrest time. Fig. Figs. and thereafter decreases throughout the analysis until the arrest time where its magnitude is about 0.6% of the initial strain energy which is 4.16 joule and corresponds to the magnitude of critical applying load (ﬁxed displacement) for initial crack advancement. the maximum value of strain energy occurs on the threshold of crack advancement. the maximum kinetic energy is 21. Moreover. Results for RDCB-8 specimen This specimen has a degree of crack tip bluntness that corresponds to the initiation stress intensity factor KQ = 1. Variation of dynamic stress intensity factor vs. Although the results are nearly similar the predictions using the static elastic properties compare generally less favorably than those based upon the dynamic properties in this specimen. Owing to loading condition in this problem. 9. [15]. Also. Fig. the arrest time predicted by this study and [15] is identical. This is the time that the model again reaches a static condition because of crack arrest.0276 joule which is nearly equal to the similar value of RDCB-4 at the arrest time.5) 1. As shown.A.8% less than the ﬁrst studied type of specimen. the crack tip velocity predicted by this study and the method of [12] is close to each other. The magnitude of kinetic energy starts to increase parallel to the crack advancement and reaches a maximum value at the time about 140 ls.2 0. This quantity was about 0. the behavior of crack tip velocity curve is similar to the dynamic stress intensity factor curve. M. In general. and the optimum mesh is generated each time during the crack propagation. This specimen is analyzed based on the dynamic material properties only.R. Amini Fasakhodi / Materials and Design 30 (2009) 1032–1041 1039 5. crack length obtained from the present method along with the existent results for this specimen. 14. however. History of crack extension for RDCB-8. the ﬁnite element analysis of the problem is accomplished with the wellshaped elements. According to Fig. the singular elements are always at the crack tip position. 8 and 9. and is about 170 mm. It is also seen from Fig. 10 also contains the kinetic energy and strain energy values obtained based on the dynamic material properties by Gehlen et al. 7 shows the results for the dynamic stress intensity factor vs. 9 that the predicted results via the different methods agree well with each other with the exception of Gehlen et al.4 Present study with dynamic properties Experimental results by Kalthoff et al [36] Koh et al [15] 0 60 90 120 150 180 crack length (mm) Fig. 11 shows the mesh details and the variation of mesh layout during the dynamic fracture simulation of RDCB-4 specimen. The curve obtained for this case shows a trend similar to that presented for RDCB-4.2. that measured in the experimental study [36]. .6 dynamic SIF (MPa.m^0. is because of energy release due to the crack advancing. It is seen that prior to arrest the crack velocity decreases with an abrupt slope from the local maximum value.76 MPa m1/2. 12. The ﬁgure shows the results obtained by Nishioka and Atluri [12] which is in good agreement with those of this study. and does not indicate that the event occurs statically. crack length for RDCB-8 which represents the comparison of the present results with the experimental results. 10.4% of the initial strain energy which does not indicate that the event occurs statically. Fig. their results are in the higher level of magnitude and time process among the methods. The time duration obtained by the experimental study [36] is larger with respect to the value computed by this investigation. respectively. As shown in Fig. which includes the experimental results [36] and numerical results of [15]. Decrease of both types of energy. and has a local maximum at a time about 290 ls. It is observed that because of using remeshing technique.’s method [38]. formulation without using torsional spring in the equations [38]. 13. Energy quantities such as kinetic energy and strain energy during dynamic fracture process with the ﬁxed displacement loading condition are depicted in Fig. the maximum value of strain energy for RDCB-8 is 0. Despite the prediction of Koh et al. the value of kinetic energy is zero at the beginning of propagation because the model is in the static conditions on the threshold of crack propagation. 12 shows the variation of dynamic stress intensity factor vs. Fig.R. 180 150 120 90 Present study with dynamic properties Experimental results by Kalthoff et al [36] Koh et al [15] 60 0 100 200 300 400 500 600 time (micro second) Fig. For this specimen. According to Fig.8 0. whether using dynamic or using static material properties the maximum kinetic energy is 26. For this specimen. 15 shows the crack tip velocity vs. 1. The crack length history and crack tip velocity history are shown in Figs. respectively. 13 and 14 show the crack length and energy history. and has a value equal to 369 ls. time obtained based on the dynamic and static material properties separately. 10. and comprises a maximum value at the start of crack growth.3 joule for RDCB-4 specimen. Shahani. It is seen that there is a good agreement between crack length histories. the predicted crack arrest length via the present method is nearly identical with crack length (mm) that the best estimation for the crack arrest with reference to the experimental study corresponds to this study. Fig. which shows the maximum difference at the arrest time. crack length for RDCB-8. But the amount of strain energy at the arrest time is 0.

In general. Variation of crack tip velocity vs. history of crack propagation and velocity. The following conclusions could be drawn from the above-mentioned unstable crack growth analysis: With decreasing crack tip blunting.12 relative crack velocity. The aforementioned parameters in this study have shown good agreement with the experimental results as compared with the other numerical results published in the literature. respectively. [40] arisen from Fig. The present results together with the experimental results of Kalthoff et al. Here. Amini Fasakhodi / Materials and Design 30 (2009) 1032–1041 strain & kinetic energy (joule) 0.R.05 time (micro second) Fig. c0 is the bar wave speed in the material.51917 200 150 100 50 0 67 87 107 127 147 167 Present study with dynamic properties Experimental results by Kalthoff et al [36] 6. 14.04 0. A ﬁnite element analysis based on the remeshing technique has been used to simulate the crack growth during the fracture process. The predicted crack armax rest lengths corresponding to the same values of V c0 have been extracted from Fig.16978 aarrest =a0 3.R. the crack arrest lengths of which are presented in the table. 16.08 0. It is seen that a relatively good agreement exists between the results. 5.3. Moreover.cons =c0 Present study Predicted by Shmuely [41] Predicted by Hahn et al. and this is due to the fact that when it combines with the singular elements it could model the crack tip singularity and conserve it through the crack advancement. Crack velocity–crack length relation A further comparison could be implemented for more assurance of the predicted crack arrest length using the hypothesis presented by Hahn et al.cons =c0 0. which includes the dynamic fracture toughness. 0 0 350 300 250 1 2 3 4 relative crack length. Relation between maximum constant velocity and crack length at arrest. Their theoretical calculations predict a unique relation between the length of the crack at arrest and the maximum constant crack velocity which depends only on the geometry of the specimen but not on the toughness of the material. 16 and presented in other rows of the same table.1 0 0 50 100 150 200 250 300 350 400 0. .2 Predicted by Shmuely [41] Predicted by Hahn et al [40] 0. M.1726 0.9522 RDCB-8 _ amax.1726 and 0. has been analyzed. the experimental results [36] assure the preciseness of the results of this investigation over those of Hahn et al. a(arrest)/a0 Fig. (Vmax/c0) 0.6926 2.1613 for V c0 in the RDCB-4 and RDCB-8 specimens.1613 0. 16.9516 2.51917 2. The remeshing technique has been preferred among the traditional methods. Conclusion The problem of dynamic crack propagation in RDCB specimen.1613 0. energy quantities and crack arrest length.16 strain energy with dynamic properties.1726 0. Variation of energy quantities vs. Both theoretical curves in a properly normalized form are shown in Fig. [40] and Shmuely [41]. velocity (m/s) Table 2 Summarized results of different predictions for RDCB-4 and RDCB-8 specimens RDCB-4 _ amax.15 0. Shahani. a dynamic fracture criterion.1726 0. 15. [36] 0. [40] and Shmuely [41]. Present study 0. [36] for both RDCB-4 and RDCB-8 specimens are summarized in Table 2 in order to compare them with the predicted results by Shmuely [41] and Hahn et al. The ratio of the maximum kinetic energy to the initial strain energy in the specimen is considerable. crack length (mm) Fig.038 2.1613 0.1040 A. has been employed in order to extract the crack growth velocity at each time step. fast fracture mechanics analysis deals with the parameters such as dynamic stress intensity factor. and this implies that the dynamic effects are dominant in the unstable crack propagation phenomenon and should be taken into account in the analysis. the initial stored strain energy in the specimen decreases causing the crack arrest length and time to decrease. crack length for RDCB-8. time for RDCB-8. made of a brittle material.155 aarrest =a0 2. Present study kinetic energy with dynamic properties.7716 2.5726 2. Nevertheless. [40] Experimental by Kalthoff et al. 16. max This study yields to the values of 0.

Leiden: Noordhoff. Theor Appl Fract Mech 1987. Jeong UY. Ohio. Haber RB. Int J Numer Meth Eng 1996. Int J Solids Struct 2004. Combescure A. Modeling of dynamic crack propagation. Weidner TJ. Fast fracture and crack arrest.32(17/18):2547–70. [13] Koh HM. Equations of motion at constant deformation. Owen DRJ.8:137–50. editors. Shahani. Recent theoretical and experimental results on fast brittle fracture. editors. J Appl Mech 1980. 1977. Ostlund S. Head JL.3:381–97. Yehia NAB. In: Hann GT. Dynamic crack propagation analysis using Eulerian–Lagrangian kinematics descriptions. An incremental formulation of the moving grid ﬁnite element method for the prediction of dynamic crack propagation. Eng Fract Mech 1996. An element failure algorithm for dynamic crack propagation in general directions. Atluri SN. [24] Swenson DV. Amini Fasakhodi / Materials and Design 30 (2009) 1032–1041 1041 The size of blunting at the crack tip inﬂuences the level of crack growth velocity.43:271–85. [17] Belytschko T. King WW.20:211–23. Dynamic fracture using element-free Galerkin methods. Turner CE. Sakai Y.8:194–206.25:69–79. p. Equations of motion at constant force. Dynamic fracture analysis by dynamic ﬁnite element method generation and propagation analysis. [18] Beissel SR. 1979. In: Perrone N. vol. p. The strain energy related to the arrest point for both specimens analyzed here. Nonlinear and dynamic fracture mechanics. [41] Shmuely M. editors. 35. Theor Appl Fract Mech 1985. J Mech Phys Solids 1960. Comput Meth Appl Mech Eng 2004.3:141–55. Some kinetic considerations of the Grifﬁth criterion for fracture I. Int J Fract 1979. Measurement of dynamic stress intensity factors for fast running and arresting cracks in double-cantilever-beam specimens. Freund LB.158:295–309. editors. Int J Numer Meth Eng 1993. A simple model of the double cantilever beam crack propagation specimen. 1979. Simulation of fast fracture in the DCB specimen using Kanninen’s model. [29] Rose LRF.12(6):799–813. Int J Fract 1977. i. Part I. Triangular quarter-point elements as elastic and perfectly-plastic crack tip elements. Kanninen MF. Dynamic fracture initiation and propagation in 4340steel under impact loading. editors. Fast fracture resistance and crack arrest in structural steels. Int J Fract 1974. [21] Sumi Y. Element-free Galerkin methods for static and dynamic fracture. Nucl Eng Des 1995. [28] Shahani AR.13(5):655–65. and this implies that the arrest occurs when the strain energy reaches a pronounced value related to the value of crack arrest toughness.36:2473–99. Comput Mech 1988. Nilsson F.R. Lucas RA. Wawrzynek PA. Computational crack path prediction. Hoagland RG. In: Luxmoore AR. The accuracy of this investigation is more noticeable when it concerns with the crack arrest length or arrest time predicted in the specimens. . 109–22. Compos Struct 1985. AMD. dynamic crack propagation. Application of quadratic singular distorted isoparametric elements to linear fracture mechanics. [9] Caldis ES. ASTM STP 711. Forqani M. Columbus. 35. Philadelphia: American Society of Testing and Materials. p. vol. In: Hahn GT.11:58–98. 161–76. [12] Nishioka T. [32] Rosakis AJ. [31] Kobayashi T. Popelar CH. Rosenﬁeld AR. Nonlinear dynamic transient methods in crack propagation studies. p. Eng Fract Mech 1998. Lee SH. [5] Malluck JF. Kanninen MF. Fin Elem Anal Des 1989. [30] Freund LB.41:3793–807. [27] Rethore J.47(3):570–6. the magnitude of crack arrest length and also the arrest time in the ﬁxed-displacement loading condition.15(3): 281–94.. Rosakis AJ. [37] Barsoum RS. Duffy J. Eng Fract Mech 1982. [7] Yagawa G. [15] Koh HM. 1973. Swansea (UK): Pineridge Press. An experimental investigation of the velocity characteristics of a ﬁxed boundary fracture.53:839–44. Dynamic fracture mechanics. vol. J Mech Phys Solids 1984. ASME AMD. Kanninen MF.3(4):443–60. Sousa JL. 1998. Numerical methods in fracture mechanics. [16] Belytschko T.41:1313–29. [14] Koh HM. Burd GS. [19] Lim IL. [26] Tradegard A. p. Int J Fract 1975.55(2):321–34. p. Static and dynamic fracture mechanics analysis of a DCB specimen considering shear deformation effects. Numerical methods in fracture mechanics. 1979. [6] Shahani AR. Simulation of glass cutting with an impinging hot air jet. RDCB-4 and RDCB-8. is nearly identical.4:149–56. 415–26. [33] Zehnder AT. Quasi-automatic simulation for 2D LEFM problems. Lu YY. Ingraffea AR. References [1] Berry JP. M. Sejnoha R. 612–33. A stable numerical scheme for the ﬁnite element simulation of dynamic crack propagation with remeshing. Crack arrest methodology and applications. Owen DR.16(3):303–32. Ingraffea AR. Gravouil A.R. Comput Meth Appl Mech Eng 1998. 1985. [40] Hahn GT. Modeling mixed-mode dynamic crack propagation using ﬁnite elements. [39] Kobayashi AS.39:923–38. [3] Kanninen MF.13(4):443–54. Haber RB. Papelar CH. SSC-progress report on project SR-201. Ingraffea AR. The determination of dynamic fracture toughness of AISI 4340 steel by the shadow spot method. 189–210. Choi SK.8:207–16. Atluri SN. Owen DRJ. Int J Fract 1977. [35] Kanninen MF. ASTM STP. ASTM. Kanninen MF. University College. Kanninen MF. Atluri SN. 627. [25] Bittencourt TN. [23] Wawrzynek PA. [20] Shephard MS. Int J Fract 1990.10(3):415–30. [10] Kanninen MF. Some kinetic considerations of the Grifﬁth criterion for fracture II. Ando Y. In: Nonlinear and dynamic fracture mechanics. ASTM STP 627. 1973. Tabbara M. Johnston IW. [22] Wawrzynek PA. [38] Gehlen PC. Elastodynamic formulation of the Eulerian–Lagrangian kinematic description. In: Hahn GT. In: Luxmore AR. Zienkiewicz OC. 634–47. theory and applications. 1977. I: Validation of one dimensional analysis. Berlin: Cambridge University Press. Beinert J. Seyyedian M. [36] Kalthoff JF. Advanced fracture mechanics. Interactive ﬁnite element analysis of fracture processes: an integrated approach. 1978. editor. p. Int J Solids Struct 1995.5:87–96. Tabbara M. J Mech Phys Solids 1960.e. Swensea. Numerical modeling of dynamic crack propagation in ﬁnite bodies by moving singular elements. Automatic crack propagation tracking. An Interactive approach to local remeshing around a propagating crack. Winkler S. A dynamic analysis of unstable crack propagation and arrest in DCB test specimen. Battelle Columbus Laboratories. J Appl Mech 1986. New-York: Oxford University Press. Johnson GR. In: Sih GC. 1978. editors. [8] Keegstra PNR. Popelar CH. Lee SH. Int J Solids Struct 2004.193:4493–510. FEM-remeshing technique applied to crack growth problems. 1–17. Int J Numer Meth Eng 1977. [34] Paxton TL.A. 19–37. Analysis of fast fracture and crack arrest by ﬁnite differences. ASME. Comput Mech 1988. Gu L. Dally JW. [4] Freund LB.61: 407–25. [11] Nishioka T.160:115–31. Ingraffea AR. A critical appraisal of solution techniques in dynamic fracture mechanics. p. J Mech Phys Solids 1977. [2] Berry JP. Fast fracture and crack arrest. Numerical analysis of dynamic crack propagation: generation and prediction studies.

- 2 Python and Scripting Summary
- Strength prediction of single- and double-lap joints by standard and extended finite element modelling
- Par4all User Guide
- Python Tutorial
- Lecture7 Meshing
- Xfem Crack Propagation 2D
- Elements of Vibration Analysis
- Pg Fuser Guide
- Stress Invariants
- MD-2
- 43781037 NX Nastran Basic Nonlinear Analysis Users Guid
- 43781037 NX Nastran Basic Nonlinear Analysis Users Guid
- Propped Cantilever Beam Tutorial Abaqus Final
- Estimaci´on y acotaci´on del error de discretizaci´on
- Autism
- Bode Table.1
- uprimer
- ANSYS Advantage Vol3 Iss1
- Analis de Pelicula Cast Away

- Master Thesis
- Mechanics of Dynmaic Fracture
- The Mechanics of Dynamic Fracture
- L6B 24Oct07 Fracture
- Ansys excer.
- Sun 2012
- Alfano Problem
- Chap 7
- Meschke Crack Propagation Criteria X FEM
- fracture
- International Journal of Engineering Research and Development (IJERD)
- Fracture problems with ANSYS.pdf
- pm10_tisk
- 12.1. Introduction to Fracture
- Fracture
- Notes on Fracture
- Fracture
- Estimation of Stress Intensity Factor of a Central Cracked Plate
- Sd Article 41
- Fm Lecture 2
- 06song_investigationdynamic
- 134420100303
- Fracture
- 03zhang_msthesis
- fracture bem aliabadi
- Caballero Dyskin 2008
- Fracture Toughness Characterization
- J Integral
- Investigations on the effect of shape of traction separation law on the results of cohesive zone model
- tmpF05D.tmp
- FEA_dynamiccrackprop

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue reading from where you left off, or restart the preview.

scribd