Anda di halaman 1dari 9

ARTICLE IN PRESS

International Journal of Pressure Vessels and Piping 85 (2008) 219–227


www.elsevier.com/locate/ijpvp

A ductile fracture analysis using a local damage model


N. Benseddiqa,, A. Imadb
a
Laboratoire de Mécanique et de Rhéologie de Tours, Ecole Nationale d’Ingéniéurs du Val de Loire (ENIVL), Rue de la Chocolaterie,
41000 Blois Cedex, France
b
Laboratoire de Mécanique de Lille (UMR CNRS 8107), USTL, Ecole Polytechnique Universitaire de Lille Cité Scientifique,
Avenue P. Langevin, 59655 Villeneuve d’Ascq Cedex, France
Received 26 July 2006; received in revised form 27 September 2007; accepted 27 September 2007

Abstract

In this study, the Gurson–Tvergaard–Needleman (GTN) model is used to investigate ductile tearing. The sensitivity of the model
parameters has been examined from literature data. Three types of parameters have been reported: the ‘‘constitutive parameters’’ q1, q2
and q3, the ‘‘initial material and nucleation parameters’’ and the ‘‘critical and final failure parameters’’. Each parameter in this model has
been analysed in terms of various results in the literature.
Both experimental and numerical results have been obtained for notched round and CT specimens to characterize ductile failure in a
NiCr steel (12NC6) with a small initial void volume fraction f0 (f0 ¼ 0.001%). Ductile crack growth, defined by the J–Da curve, has been
correctly simulated using the numerical calculations by adjusting the different parameters of the GTN model in the calibration
procedure.
r 2007 Elsevier Ltd. All rights reserved.

Keywords: Ductile tearing; Local damage; Numerical simulation

1. Introduction Secondly, coupled models are based upon continuum


damage mechanics such as Rousselier [9] and Gurson [10].
It is well known that the ductile failure of materials is The use of these models is based upon a constitutive
controlled by three stages, i.e. micro-void nucleation, equation introducing the material behaviour and a failure
growth and coalescence mechanisms. In order to describe criterion. In this type of model, the crack growth
this damage and fracture process, several models using simulation is automatically performed using the complete
local approaches have been proposed in the literature. deterioration of the element at the crack tip [6,11–14].
Globally, local approach modelling may be described by Among these models, the micro-mechanical model
two main families. Firstly, uncoupled models are based proposed by Gurson is the most widely used for ductile
upon void growth, such as McClintock [1], Thomason porous materials. The Gurson model employs equations to
[2,4], Rice and Tracey [3]. For these models, the material describe the constitutive response of the metal [10]. These
failure stage is characterized by a critical void growth ratio expressions are based upon a continuum elastic–plastic
(R/R0)c, corresponding to crack initiation. This parameter model that accounts for micro-void nucleation and growth.
is considered as a material intrinsic property and it is Tvergaard and Needleman [15] have subsequently modified
obtained by a combination of numerical results and this model introducing the following yield condition
experimental data. Crack growth simulation is achieved (Gurson–Tvergaard–Needleman (GTN) model):
by a node release technique when the critical void growth
ratio is reached [5–8].  
s2eq  3 sm
Fðsy ; seq ; f Þ ¼ þ 2f q 1 cosh q
s2y 2 2 sy
Corresponding author.
E-mail address: nbensedd@polytech-lille.fr (N. Benseddiq).
 ð1 þ q3 ðf  Þ2 Þ ¼ 0, ð1Þ

0308-0161/$ - see front matter r 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijpvp.2007.09.003
ARTICLE IN PRESS
220 N. Benseddiq, A. Imad / International Journal of Pressure Vessels and Piping 85 (2008) 219–227

Nomenclature f* modified damage parameter


f u ultimate value of damage parameter
J J-integral q1, q 2, q3 constitutive parameters
a crack length SN standard deviation
Da crack extension d accelerating factor
f void volume fraction sy yield stress of material
f0 initial void volume fraction seq equivalent stress
fc critical void volume fraction sm hydrostatic stress
fF void volume fraction at final failure E elastic modulus
fN volume fraction of void-forming particles

where q1 q2 and q3 are the constitutive parameters with For strain-controlled void nucleation B ¼ 0. But in the
q3 ¼ (q1)2; sm is the mean normal stress defined by case of macroscopic ductility, the hydrostatic stress
sm ¼ skk/3, skk is the trace of the stress tensor; seq is the dependence of void nucleation leads to a strong non-
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi von Mises equivalent stress defined by seq ¼
conventional normality in the plastic flow rule, which promotes early
ð3=2Þsij sij (sij is the stress deviator); sy is the yield stress; flow localization. It is worth noting that a stress-controlled
f*(f) is a function of the void volume fraction f and nucleation criterion was found to be important in captur-
represents the modified damage parameter. This function is ing the effect of stress state on damage initiation. In this
defined by case A ¼ 0 and
8 "   #
>
<
f for f pf c ; fN 1 seq þ csm  sN 2
B¼ pffiffiffiffiffiffi exp  ,

f ðf Þ ¼ f c þ dðf  f c Þ for f c pf pf F ; (2) S N 2p 2 SN
>
:f
u f Xf F : c is a constant parameter and can take values between 0.3
and 0.4.
The expression d ¼ ðf u  f c Þ=ðf F  f c Þ is considered as The micro-void volume fraction rate due to growth is
an accelerating factor, which is introduced in order to given by
describe the final stage of ductile failure, where fc is the
critical void volume fraction corresponding to void df growth ¼ ð1  f Þ dpkk , (6)
coalescence which at first occurs. This is considered as a where ekkp is the plastic hydrostatic strain.
material intrinsic parameter, f u is the ultimate value of the
damage parameter which can be written as a function of q1 2. Scope of study
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
and q3 by f u ¼ ðq1 þ ðq1 Þ2  q3 Þ=q3 . If q3 ¼ (q1)2 the
ultimate void volume fraction f u equals (1/q1). fF is the In the current paper, the GTN model is used to analyse
final void volume fraction. According to Eq. (2), f ¼ fF ductile tearing. The sensitivity of the model parameters is
corresponds to the failure stage, f  ¼ f u . analysed from literature data. Three types of parameters
In general, the evolution equation for f consists of two have been reported: ‘‘the constitutive parameters’’ q1, q2
terms describing the nucleation and growth of voids: and q3, ‘‘the initial material and nucleation parameters’’
and ‘‘the critical and final failure parameters’’.
df ¼ df nucleation þ df growth with f ðt0 Þ ¼ f 0 , (3) Both experimental and numerical results have been
obtained using notched round and CT specimens to
where f0 is the initial void volume fraction. characterize rupture behaviour in a NiCr steel (12NC6)
Nucleation is considered to depend exclusively on the with a small initial void volume fraction f0 (f0 ¼ 0.001%).
effective strain of the matrix material: The parameters in this model are determined by a
df nucleation ¼ Adpeq þ Bðdseq þ cdsm Þ. (4) calibration procedure using three types of mechanical tests:
classical tensile (true stress versus true strain), tensile with
The parameter ‘‘A’’ is chosen so that void nucleation an axisymmetric notched (AN) specimen AE (load versus
follows a normal distribution around a critical plastic diameter reduction) and ductile tearing tests (J-parameter
strain ep, which is estimated using the following expression: versus crack growth Da). The results obtained by the GTN
"  # model are discussed in the context of the extensive
fN 1 p  N 2 literature on model parameters.
A¼ pffiffiffiffiffiffi exp  , (5)
SN 2p 2 SN
3. Analysis of GTN model parameters
where fN is the volume fraction of void nucleating particles,
eN is the mean void nucleation strain and SN is the Generally, the GTN model is used to predict the damage
corresponding standard deviation. of ductile materials [4–43]. To apply this model nine
ARTICLE IN PRESS
N. Benseddiq, A. Imad / International Journal of Pressure Vessels and Piping 85 (2008) 219–227 221

Table 1
GTN model parameter values from the literature

Ref. q1 f0 fc fF d fN eN SN Le (mm) Material sy (MPa) Rm (MPa) Specimen

[6] 1.5 0.002 0.004 – 23 – – – 0.2 0.5 C–Mn steel (at 300 1C) – – SENT25 CT22.5
[12] 1.47 0.00016 0.0005 – 2.8 – – – 0.8 A508 C13 A 471 593 AE2
1.47 0.0001 0.0003 – 4.3 – – – – A508 C13 B 470 608 AE2
[13] 1.5 0.0015 0.035 0.15 5.49 0.00085 0.3 0.1 0.1 E690 779 840 AE2 SENB (B ¼ 25)
[14] 1.5 0.001 0.02 1.141 67 0.04 0.3 0.1 0.1 20MnMoNi5 5
1.43 0.114 0.13 0.272 4 – – – – 366 – AE4
1.1 0.114 0.2 0.35 4.73 – – – – – –
[17] 1.1 0.114 0.3 0.44 4.35 – – – – GGG40 – –
1.2 0.114 0.19 0.235 14.3 – – – – – –
1.2 0.114 0.175 0.235 10.97 – – – – – –
[29] 1.25 0.00033 0.026 0.15 5.17 0.006 0.3 0.1 0.033 370 –
[31] 1.15 0.002 0.033 0.15 – 0.004 0.3 0.1 – 22NiMoCr37 – – –
[32] 1.5 0.077 0.12 0.2 6.8 – – – 0.2 GGG40 230 433 CT25 AE4
[33] 1.5 0.0023 0.004 3 – – – 0.5 CMn steel (300 1C) 190 500 AE2, 4 and 10
[34] – 0.001 0.003 – – 0.01 – 0.01 – AlMgSi alloy 260 – –
[35] 1.5 – 0.15 0.25 5.17 0.04 0.3 0.1 – – – – –
[36] 1.5 0.0 0.04 0.195 – 0.008 0.3 0.1 – – 366 – –
[37] 1.5 0.08 0.15 0.28 4 0 0 0 Composite
1.5 0 0.02 0.34 2 0.04 0.5 0.2 – Al-Al3Ti 54.7 80 .
1.5 0 0.2 0.32 4 0.4 0.05 0.02
1.5 0 0.15 0.28 4 0.08 0.025 0.025
[38] 1.5 0.00057 0.03 1.3 0.004 0.3 0.1 612 707 AE0.25 and 4 CT
[39] 1.5 0.0025 0.021 3.4 0.02 0.3 0.1 0.2 E460 steel
[40] 1.25 0.005 – 0.2 – 0.001 0.04 0.01 0.2 A533B – – –
[41] 1.5 0 0.06 0.212 4 0.002 0.3 0.1 0.1 Steel 440 600
1.5 0 0.04 0.197 4 0.002 0.3 0.1 0.1 620 750 SENB
1.5 0 0.03 0.189 4 0.002 0.3 0.1 0.1 320 580
[42] 1.25 0.00033 0.026 0.15 6.24 0.006 0.3 0.1 0.033 CMn steel 360 – –
[43] 1.47 0.002 0.028 – – – – – – Cu 312 325 AE1 2 4 8

parameters are necessary. They can be classified into two and q2 ¼ 1. These values have been used in several
principal families: investigations (see Table 1).
In a recent study, Perrin et al. [16] have determined a
 the ‘‘constitutive parameters’’: q1, q2 and q3 correlation between the parameter q and porosity f: q ¼
 the ‘‘material parameters’’, which are classified in two q(f). The authors show that when the porosity tends to
parts. Firstly, ‘‘the initial material and nucleation zero, the q value tends to 1.47. This value is similar to that
parameters’’, which are determined as the initial void suggested by Tvergaard (q ¼ 1.5) [15]. In the modelling of
volume fraction f0 and the void nucleation parameters nodular cast iron, Steglich et al. [17] have proposed a yield
fn, en and Sn. Secondly, ‘‘the critical and final failure condition determined from a cell model that results in using
parameters’’, the critical void volume fraction fc a quadratic equation to estimate the parameter q1. The
and fF. calculations of this parameter for various stress triaxiality
ratios T give an average value of q1 equal to 1.5. The
A combination of numerical results and experimental authors have noted that the parameter q1 does not depend
data is necessary in order to determine some of the on the stress triaxiality ratio T but little physical
parameters. It is important to note that most of the significance is given to this value.
parameters are not easy to define. Globally, there is no Gao et al. [18] have performed several calculations
unique method to determine these parameters. For this considering a wide range of material flow properties
reason, a global analysis of the large data available in the (hardening N, sy/E) and stress triaxiality ratios (2–3.3)
literature has been performed, in order to examine the with sy/E ¼ 0.001–0.004.
validity of the choices of these parameters, and suggest a For a specific set of flow properties, the authors have
global method to determine them (see Table 1). obtained various q1 and q2 parameters. It is interesting to
note that the parameter q ¼ q1  q2 is approximately
3.1. Constitutive parameters: q1, q2, q3 always constant and equal to about 1.5 for all cases
studied (Table 2). Faleskog [19] indicates that the two
In order to describe material ductile fracture, Tvergaard parameters depend on the material hardening exponent.
[15] has suggested fixing the q1 and q2 values at q1 ¼ 1.5 The q parameters depend strongly on strain hardening and
ARTICLE IN PRESS
222 N. Benseddiq, A. Imad / International Journal of Pressure Vessels and Piping 85 (2008) 219–227

Table 2
Optimal values for micro-mechanics parameters (q1, q2) [19]

Hardening (N) sy/E ¼ 0.001 sy/E ¼ 0.002 sy/E ¼ 0.004

q1 q2 q1  q2 q1 q2 q1  q2 q1 q2 q1  q2

5 1.96 0.78 1.531 1.87 0.8 1.496 1.71 0.84 1.43


6.7 1.78 0.83 1.483 1.68 0.86 1.438 1.49 0.9 1.342
10 1.58 0.9 1.425 1.46 0.93 1.359 1.29 0.98 1.265

material strength; for a material of moderate hardening


and strength, e.g., N ¼ 0.1 and sy/E ¼ 0.002, the calibra-
tion gives q1 ¼ 1.46 and q2 ¼ 0.93. High strain hardening
reduces the effect of stress triaxiality on void growth rate, Fig. 1. Dependence of fc on f0.
resulting in q2 values much below unity [19] (Table 2).
It is important to note that the value of q1 may be
considered as essential to describe the failure stage. Because
of Eq. (2), fracture of material occurs (i.e. f ¼ fF) when f* than the value of 0.15. Fig. 1 shows a dependence of fc on
reaches its ultimate value f u ¼ 1=q1 (when q1 ¼ 1.5, f0. Only for small f0, as a first approximation, can the value
f u ¼ 0:666). The physical validity of this value is not of fc be taken as a constant. Tvergaard and Needleman [15]
proven. suggested that the value of fc can be taken as 0.15.
Table 1 shows that the value of the constitutive These observations show that fc is not a material
parameter q1 varies from 1.1 to 1.5 for many materials. parameter but is a parameter determined by fitting.
However, this parameter is often fixed to 1.5 and q2 ¼ 1. Table 1, which gives numerous values from the literature,
shows the variability of this parameter for a considered
3.2. Material parameters: f0, fN, fc, fF, eN and SN material.
On the other hand, the final failure void volume fraction
3.2.1. Initial material and nucleation parameters fF is considered a parameter that may be experimentally
Generally, the initial void volume fraction f0 and volume determined [6]. The mentioned authors show that fF affects
fraction of void nucleating particles fN are evaluated by the post-initiation load diameter–reduction curve and that
microscopical examination of the undamaged material. As the smaller the fF the faster the load decreases. In FE
a first approximation, f0 can be obtained from the MnS modelling, Chambert [24] has analysed curves illustrating
inclusions by Franklin’s formula using the chemical the evolution of equivalent stress with equivalent plastic
composition [20]. strain using three fF values: 0.075, 0.15 and 0.225. The
Table 1 also shows that the values of the nucleation author noted that this parameter controls the final slope of
parameters fN, eN and SN can be arbitrarily fixed. The these curves. Originally, Tvergaard and Needleman [15]
values eN ¼ 0.3 and SN ¼ 0.1 have been used in several proposed a value equal to 0.25.
studies, and they are determined by fitting. From Eq. (2), the final failure void volume fraction fF
may be determined if the values of q1, d and fc are
3.2.2. Critical and final failure parameters previously known using fF ¼ fc+(1/q1fc)/d. This shows
In most investigations, only the critical void volume that the final failure parameter may be given to a first
fraction fc is considered as a parameter, obtained by fitting approximation by f F ¼ 1=dq1 ¼ ð1=dÞ  f u when the value
the numerical calculations with experimental results, and of fc is low.
the other parameters are fixed arbitrary. According to Recently, Zhang et al. [28] have proposed an empirical
Zhang et al. [21], fc is not a constant but decreases when the expression for fF and f0, which is written as a linear
stress triaxiality ratio T increases. However, other authors equation: fF ¼ 0.15+2  f0. This signifies that fF can be
note that fc can be taken as a constant only for small f0 fixed to a first approximation at 0.15 for low f0 values. As
values. In a similar study, Steglich et al. [17] confirm that shown in Table 1, this parameter can take values between
the fc value depends on stress triaxiality T: fc decreases with 0.15 and 0.44.
the increase of T. In a recent experimental study [22], these
observations were confirmed and the author proposed a 4. Experimental technique and results
linear correlation between fc and T for a high f0 value
(f0 ¼ 13%): fc ¼ 0.2550.146  T. 4.1. Material
According to Koplik and Needleman [23], the fc value,
which signifies the onset of coalescence, seems to vary The nickel and chromium steel used in this investigation
slowly with stress triaxiality ratio but depends strongly on is the 12NC6 steel. Its chemical composition was (wt%):
the initial void volume fraction f0 and is generally smaller 0.12C, 0.007S, 0.32Si, 0.6Mn, 1.6M, 0.85Cr and 0.076A1.
ARTICLE IN PRESS
N. Benseddiq, A. Imad / International Journal of Pressure Vessels and Piping 85 (2008) 219–227 223

The steel was austenitized at 880 1C for 1 h, and then and Young’s modulus of 194 GPa. The uniaxial plastic flow
quenched in an oven. The final steel micro-structure is behaviour was assumed to follow a Ramberg–Osgood
illustrated in Fig. 2. This shows the ferritic–pearlitic bands. stress–strain law:
The initial void volume fraction f0 is small (f0 ¼ 0.001%). 8
>  ¼ e þ p
This micro-structural parameter is necessary to use the <
GTN model. s ¼ Ee if spsy ; (7)
>
: s ¼ sy þ kn if s4sy ;
p
4.1.1. Mechanical properties
Three tensile tests were carried out at ambient tempera- with the hardening exponent N ¼ 0.45 and ductility
ture on cylindrical specimens with a diameter d ¼ 11 mm. coefficient k ¼ 544 MPa.
The mechanical properties of this material are as follows: To experimentally determine the crack initiation point,
0.2% proof stress of 340 MPa, ultimate stress of 489 MPa other tensile tests have been carried out using three sets of
standard axisymmetrically notched bar specimens: AE2,
AE4 and AE10. This specimen is often used in order to
vary the stress triaxiality ratio T with the notch radius
according to Brigdman’s formula [26]:
 
sm 1 r
T¼ ¼ þ ln 1 þ , (8)
seq 3 2r
where r is notch radius and r is the minimum radius of the
notched round specimen. As shown in Fig. 3, the specimens
used in this study have a notch radius r ¼ 2, 4 and 10 mm,
and minimum radius r ¼ 3 mm. This leads to three values
of stress traixiality ratio T ¼ 0.89, 0.65 and 0.47.
During this kind of test, it is difficult to correctly
measure the diameter reduction. In our study, the variation
of diameter has been measured using an image technique,
which has the advantage of not requiring contact. An
example of an image is shown in Fig. 4. This technique can
follow the evolution of the diameter reduction exactly at
Fig. 2. Micro-structure of the steel used in this study. the notch bottom and at the minimum diameter. In

Fig. 3. Notched round specimens: (a) NR2, (b) NR4 and (c) NR10.
ARTICLE IN PRESS
224 N. Benseddiq, A. Imad / International Journal of Pressure Vessels and Piping 85 (2008) 219–227

Fig. 5. Load versus diameter reduction for three notched axisymmetric


specimens.

Fig. 6. Failure strain versus stress triaxiality ratio.

when T increases, i.e., when the minimum diameter


decreases. This observation is in agreement with several
literature results.

4.2. Fracture toughness tests


Fig. 4. An example of typical images to measure the diameter reduction.
Global profile corresponding to (a) first image, (b) at initial diameter and Six fracture mechanics tests were performed at room
(c) at crack initiation.
temperature using side-grooved CT specimens (width
W ¼ 50 mm, thickness BN ¼ 20 mm) pre-cracked to
addition, the image analysis technique easily detects the a/W ¼ 0.5. The parameter J was determined according to
point where crack initiation is localized. the ASTM standard procedure [25] using the area under
Fig. 5 shows the evolution of applied load as a function the load–displacement curve and crack extension was
of diameter reduction DF for the three configurations deduced from the partial unloading compliance method.
examined. These curves give the critical diameter reduction Fig. 7 shows the evolution of J with crack extension. The
DFc at crack initiation. The ductility er is defined as the J0.2 value, corresponding to a crack extension
average longitudinal strain at fracture: Da ¼ 0.2 mm, is about 210 kJ/m2. The high value of this
  parameter can be explained by the small value of initial
F0
r ¼ 2 ln . void volume fraction of the steel used in this study. The
F0  DFc
stable growth stage, within the valid domain of crack
Fig. 6 shows the variation of failure strain with stress extension defined by two parallel offset lines at 0.15 and
triaxiality ratio T. The failure strain is sensitive to the stress 1.5 mm crack growth, is described by a power function
triaxiality ratio T. Indeed, the coalescence strain decreases J ¼ J1.0Dan, with J1.0 ¼ 529 kJ/m2 and n ¼ 0.89.
ARTICLE IN PRESS
N. Benseddiq, A. Imad / International Journal of Pressure Vessels and Piping 85 (2008) 219–227 225

Fig. 7. J-integral versus crack extension Da (experimental results).

5. Numerical investigation

The GTN model has been applied in the FE simulation


using the SYSTUS software. The material is assumed to be
homogeneous, elastoplastic with isotropic hardening. Five
parameters of this model are necessary: q1 ¼ q (i.e.,
q2 ¼ 1), A, f0, d and fc.
Two configurations have been analysed: the tensile AN
specimen and the CT specimen. Fig. 8 shows the meshing
corresponding to each specimen using quadratic elements Fig. 8. (a) Tensile axisymmetric notched, (b) typical meshing for CT
with eight nodes. In the CT specimen, the mesh size around specimen and (c) detail around crack tip (Le ¼ 0.2 mm).
the crack tip is refined and chosen equal to Le ¼ 0.2 mm,
which is a size usually used for ductile tearing simulation. It
is well known that the Le value affects the computation
results using the local approach [9,26]. In the literature,
proposals have been made to estimate the mesh size as a
function of the number of inclusions per unit volume Nv
[9,26].
For the ductile tearing computation, the whole specimen
must be considered even where the specimen has geome-
trical symmetry [24,27].
At first, the following parameter set has been chosen in
order to simulate the P–DF curve of the AN specimen:

 f0: initial void volume fraction is evaluated by


observations f0=0.001%,
 q: interaction between void parameters, q=q1=1.5
and q2=l ((with q3(q1)2),
 d: amplifying factor of f in the coalescence phase is
arbitrarily fixed to 3 with d ¼ ðf u  f c Þ=ðf F  f c Þ
where fu*=1/q1=0.666,
 A: continuous void nucleation parameter is arbitrarily Fig. 9. Critical void volume fraction versus stress triaxiality ratio.
fixed to 0.001 and
 fc: critical void volume fraction or critical porosity
(adjustable parameter) with fF determined as a func- ality ratio values. As expected, fc decreases with increasing
tion of 1/q1, d and fc from Eq. (2). stress triaxiality ratio T (Fig. 9). The value of fc depends
on the set of parameters initially fixed and the average
From this set of parameters, the critical void volume value in this case is fc ¼ 0.004. Using this set of para-
fraction fc has been computed by fitting the experimental meters (f0 ¼ 0.001%, q ¼ q1 ¼ 1.5, A ¼ 0.001, d ¼ 3 and
load–diameter reduction curves for different stress triaxi- fc ¼ 0.004), crack growth simulation has been carried out
ARTICLE IN PRESS
226 N. Benseddiq, A. Imad / International Journal of Pressure Vessels and Piping 85 (2008) 219–227

parameters characterizing this model have been studied


using literature data and have been classified as follows:

 The constitutive parameters (q1, q2, q3) which are always


fixed and not dependent on material. In the majority of
cases q1 ¼ 1.5.
 The material parameters (f0, fN, fc, fF, eN and SN), where
f0 can be determined experimentally and it is possible to
determine the relationship between f0, fc and fF.

The critical void volume fraction fc decreases with


increasing stress triaxiality. In numerical calculations,
related to experimental data on a nickel–chrome steel, an
average value of this parameter has been used.
The crack growth resistance curve has been simulated by
using the following parameters: the constitutive parameters
are constant, q1 ¼ 1.5 and q2 ¼ 1, the amplifying factor in
the coalescence phase d is 3, the void nucleation factor is
arbitrarily fixed at 0.001. The first calculations lead to a
difference between the numerical results and experimental
data. This may be due to using a value for the coalescence
Fig. 10. J–Da curves. Comparison between numerical results and
experimental data.
phase, d, which is too high. Indeed, a value of this chosen
parameter equal to 2 leads to the correct results. In this
model case, it is difficult to determine a unique solution.

References

[1] McClintock FA. A criterion for ductile fracture by growth of holes.


J Appl Mech 1968;35:363–434.
[2] Thomason PF. Ductile fracture and the stability of incompressible
plasticity in the presence of microvoids. Acta Metall 1968;29:360–5.
[3] Rice JR, Tracey DM. On ductile enlargement of voids in triaxial
stress fields. J Mech Phys Solids 1969;17:201–17.
[4] Thomason PF. Ductile fracture of metals. Oxford: Pergamon Press;
1990.
[5] Wilsius J, Inad A, Nait Abdelaziz M, Mesmacque G, Eripret C. Void
growth and damage models for predicting ductile fracture in welds.
Fatigue Fract Eng Mater Struct 2000;23:105–12.
Fig. 11. Load versus diameter reduction for d ¼ 2 and 3. [6] Bauvineau L, Burlet H, Eripret C, Pineau A. Modelling ductile stable
crack growth in a C–Mn steel with local approaches. In: First
European mechanics of materials conference on local approach to
fracture, Fontainebleau, September 1996.
using the GTN model, but underestimates the resistance [7] Imad A, Wilsius J, Na-t Abdelaziz M, Mesmacque G. Experiments
curve. This may be explained by the high value of the and numerical approaches to ductile tearing in an 2024-T351
aluminium alloy. Int J Mech Sci 2003;45:1849–61.
amplifying factor d (Fig. 10). Fig. 10 also shows that the [8] Dutta BK, Saini S, Arora N. Application of a modified damage
crack growth curve is correctly simulated when d is reduced potential to predict ductile crack initiation in welded pipes. Int J Press
to 2. This indicates that the amplifying factor plays an Vessels Piping 2005;82(11):833–9.
important role in crack growth simulation. On the other [9] Rousselier G. Ductile fracture models and their potential in local
hand, d plays a weak part on the end of the curve (load approach to fracture. Nucl Eng Design 1987;105:97–111.
[10] Gurson AL. Continuum theory of ductile rupture by void nucleation
versus diameter reduction) for AN numerical result and growth: Part I—yield criteria and flow rules for porous ductile
(see Fig. 11). In our case we have chosen arbitrarily this media. ASME J Eng Mater Technol 1977;99:2–15.
parameter (d ¼ 3), but we suggest taking smaller values [11] Wilsius J. Etude experimentale et numerique de la dechirure ductile
(1–2) to obtain good results. basee sur des approches locales en mecanique de la rupture. PhD
thesis, Universite des Sciences et Technologies de Lille, France, 1999.
[12] Devaux J, Joly P, Leblond JB. Simulation by the local approach of
6. Conclusions ductile crack growth in a pressure vessel steel using an improved
Gurson–Tvergaard model. 21st MPA seminar, October 1995.
[13] Betegon C, Rodriguez C, Belzunce FJ. Analysis and modelisation of
This study has examined ductile fracture analysis using short crack growth by ductile fracture micromechanisms. Fatigue
the GTN model based on local damage. The nine Fract Eng Mater Struct 1997;20(5):633–44.
ARTICLE IN PRESS
N. Benseddiq, A. Imad / International Journal of Pressure Vessels and Piping 85 (2008) 219–227 227

[14] Hao S, Brocks W, Heerens J, Hellmann D. Simulation of 3D ductile [30] Brocks W, Hao S, Steglich D. Micromechanical modelling of the
crack growth by the Gurson–Tvergaard–Needleman model. ECF11 damage and toughness behaviour of nodular cast iron materials. In:
Mechanisms and mechanics of damage and failure, September 1996. First European mechanics of materials conference on local approach
p. 805–10. to fracture, Fontainebleau, September 1996.
[15] Tvergaard V, Needleman A. Analysis of the cup–cone fracture in a [31] Brocks W, Sun DZ, Honig A. Verification of micromechanical
round tensile bar. Acta Metall 1984;32:157–69. models for ductile fracture by cell model calculations. Comput Mater
[16] Perrin G, Leblond JB. Analytical study of a hollow sphere made of Sci 1996;7:235–41.
plastic porous material and subjected to hydrostatic tension— [32] Dong MJ, Berdin C, Beranger AS, Prioul C. Damage effect in the
application to some problems in ductile fracture metals. Int J fracture toughness of nodular cast iron. In: First European mechanics
Plasticity 1990;6:677–98. of materials conference on local approach to fracture, Fontainebleau,
[17] Steglich D, Brocks W. Micromechanical modelling of damage and September 1996.
fracture of ductile materials. Fatigue Fract Eng Mater Struct 1998; [33] Decamp K, Bauvineau L, Besson J, Pineau A. Size and geometry
21(10):1175–88. effects on ductile rupture of notched bars in a C–Mn steel.
[18] Gao X, Faleskog J, Shih CF, Dodds RH. Ductile tearing in part- Experiments and modelling. Int J Fract 1997;88:1–18.
through cracks: experiments and cell-model predictions. Eng Fract [34] Sovik OP. Experimental and numerical investigation of void
Mech 1998;59:761–77. nucleation in an AlMgSi alloy. In: First European mechanics of
[19] Faleskog J, Gao X, Shih CF. Cell model for nonlinear fracture materials conference on local approach to fracture, Fontainebleau,
mechanics—I. Micromechanics calibration. Int J Fract 1998;89:365–73. September 1996.
[20] Franklin AG. Comparison between a quantitative microscope and [35] Ghosal AK, Narasimhan R. Numerical simulations of hole growth
chemical methods for assessment of non-metallic inclusions. J Iron and ductile fracture initiation under mixed-mode loading. Int J Fract
Steel Inst 1969:181–6. 1996;77:281–304.
[21] Zhang ZL, Niemi E. A new failure criterion for the Gurson–Tver- [36] Kuna M, Sun DZ. Three-dimensional cell model analyses of void
gaard dilatation constitutive model. Int J Fract 1995;70:321–34. growth in ductile materials. Int J Fract 1996;81:235–58.
[22] Guillemer-neel C. Comportement mecanique et endommagement de [37] He R, Steglich D, Heerens J, Wang G-X, Brocks W, Dahms M.
la fonte a graphite spheroidal. PhD thesis, UTC France, 1999. Influence of particle size and volume fraction on damage and fracture
[23] Koplik J, Needleman A. Void growth and coalescence in porous in Al-A13Ti composites and micromechanical modelling using
plastic solids. Int J Solids Struct 1988;24(8):835–53. the GTN model. Fatigue Fract Eng Mater Struct 1998;21(10):
[24] Chambert J. Etude parametrique d’un modele d’endommagement de 1189–201.
type Gurson. Prise en compte d’un critere de coalescence. In: [38] Sun DZ, Siegle D, Voss B, Schmitt W. Application of local damage
Proceeding du 14eme Congres de Mecanique, Toulouse, 1999. models to the numerical analysis of ductile rupture. Fatigue Fract
[25] Chung WN, Williams JG. Elastic–plastic fracture test methods: the Eng Mater Struct 1989;12(3):201–12.
user’s experience (second volume). In: Joyce JA, editor. ASTM STP, [39] Siegmund T, Bernauer G, Brocks W. Two models of ductile fracture
1114, vol. 2. ASTM; 1991. p. 320–39. in contest: porous metal plasticity and cohesive elements. ECF12
[26] Devaux JC, Mudry F, Pineau A, Rousselier G. Experimental and fracture from defects, September 1998. p. 933–8.
numerical validation of a ductile fracture local criterion based on a [40] Xia L, Cheng L. Transition from ductile tearing to cleavage fracture:
simulation of cavity growth. In: Saxena A, Merkle JG, editors. Non a cell-model approach. Int J Fract 1997;87(3):289–306.
linear fracture mechanics, vol. 2II. Philadelphia: American Society [41] Schmitt W, Sun DZ, Blauel JG. Damage mechanics analysis (Gurson
for Testing and Materials; 1989. p. 332–54. model) and experimental verification of the behaviour of a crack in a
[27] Zhang ZL, Niemi E. Analyzing ductile fracture using dual dilational weld-cladded component. Nucl Eng Design 1997;74:237–46.
constitutive equations. Fatigue Fract Eng Mater Struct 1994;17(6): [42] Skallerud B, Zhang ZL. A 3D numerical study of ductile tearing and
695–707. fatigue crack growth under nominal cyclic plasticity. Int J Solids
[28] Zhang ZL, Thaulow C, Phidegard J. A complete Gurson model Struct 1997;34(24):3141–61.
approach for ductile fracture. Eng Fract Mech 2000;67:155–68. [43] Pardoen T, Doghri I, Delannay F. Experimental and numerical
[29] Zhang ZL. A sensitivity analysis of material parameters for the comparison of void growth models and void coalescence criteria for
Gurson constitutive model. Fatigue Fract Eng Mater Struct 1996; the prediction of ductile fracture in copper bars. Acta Mater
19(5):561–70. 1998;46(2):541–52.

Anda mungkin juga menyukai