Académique Documents
Professionnel Documents
Culture Documents
Co n te n ts l i s ts a va i l a b l e a t S ci Ve rs e S ci e n ce D i re ct
j o u rn a l h o m e p a g e : w w w . e l s e v i e r . c o m /l o c a t e /c o m m a t s c i
Prediction of creep crack growth behavior in ASME P92 steel welded joint
Lei Zhao, Hongyang Jing, Yongdian Han, Junjie Xiu, Lianyong Xu ⇑
Sch o o l o f Ma teria ls Scie n ce an d En gin e erin g, Tia n jin Un ive rsity, Tia n jin 3 00072 , Ch in a
a r t i c le i n f o a b s t r a c t
Article h isto ry:
In the present study, the prediction of creep crack growth behavior in ASME P92 steel welded joint at
Received 27 February 201 2 650 ° C was carried out using elastic–plastic–creep finite element method (FEM) and theoretical predic-
Received in revised form 6 April 201 2 tion models, respectively. Then the predicted creep crack growth rate was correlated with parameter
Accepted 6 April 201 2 Ã
C and compared with experimental data. It was observed that the creep crack growth behaviors calcu-
Available online 5 May 201 2
lated by the FEM and predicted by the modified NSW model under plain stress condition agreed well with
experimental results. Hence, the appropriate creep residual life for the in-service components containing
defects could be provided without conducting creep crack growth tests. Furthermore, creep crack growth
Keywo rds:
1 . Introduction conditions and also studied the creep crack initiation behavior in
1 /2CrMoV steel welded joint [ 7,8]; Tu et al. reported the reduction
The ASME P92 steel has been widely used in the Ultra Supercritical in the structure life due to the presence of a weld and the effect of
Power Plants Components due to high creep resistance and high weld repair [ 9]. Becker et al. predicted the creep rupture life in a
corrosion resistance at temperature above 600 ° C and service pres- multi-material cross-weld bar using continuum damage mechan-
sure above 30 MPa [ 1 –3]. Components always contain some macro- ics approach incorporated within FEM [ 1 0]. However, the studies
scopical or microscopical defects (e.g. inclusions or micro cracks) mentioned above only focused on the creep fracture life and the
during manufacturing. These defects may grow during service. In creep initiation time, and the creep crack process was not studied
addition, they can also stimulate the nucleation and coalescence in detail. In addition, a number of FEM analyses had been carried
of creep voids during creep [ 4] . For the components servicing at ele- out to obtain the creep crack process by releasing the nodes con-
vated temperature, crack propagation may occur by the nucleation, strained in the crack plane. Using this method, Saxena et al. [ 1 1 ]
growth and coalescence of the creep voids, which in turn may accel- analyzed creep crack growth data by assessing the deflection rate
erate the failure in material. Therefore, it’s crucial to develop the and Zhao et al. [ 1 2] numerically derived the creep crack behavior
safe and accurate methods to predict the creep crack initiation in a nickel base superalloy on the basis of experimental creep crack
and creep crack growth behavior Thus catastrophic fracture occur- propagation dada. Although these calculated results showed agree-
ring in fossil power plants components made of ASEM P92 steels ment with the experimental data, the criterion of crack length ver-
during long time service could be avoided. sus time was determined by the experimental result rather than
Up to now, existing methods to predict creep crack growth can the creep damage accumulation of the specimen. Hence, the above
be clarified into two main fields. Firstly, finite element method investigations cannot be employed to provide the appropriately
(FEM) coupled with damage mechanics had been widely employed predicted creep crack growth. Another prevail method for obtain-
to predict the creep crack behavior. Based on Karchanov–Rabotnov ing creep crack growth data was to estimate by theoretical predic-
continuum damage mechanics, Hayhurst et al. investigated the tion model. The commonly adopted method was that a single
creep damage evolution within the material to analyze the rupture damage variable based on continuum damage mechanics was em-
life of structure [ 5,6]; Hyde et al. investigated the failure life of the ployed to account for the creep crack growth. In particular, Nikbin,
butt-welded Inco71 8 sheets at 620 ° C under creep and plasticity Smith and Webster proposed a predicted model based on the duc-
tility exhaustion theory [ 1 3], the stress field in the crack tip (HRR
⇑ Corresponding author at: School of Materials Science and Engineering, Tianjin field) [ 1 4,1 5] and the Cocks and Ashby void growth model [ 1 6] .
University, Tianjin 300072, China. Tel./fax: +86 022 27402439. In this model, it was assumed that the creep damage would occur
E- m a il a ddress: xulianyong@tju.edu.cn (L. Xu). in the zone ahead of crack tip which was in the creep processing
0927-0256/$ - see front matter Ó 201 2 Elsevier B.V. All rights reserved.
http: //dx.doi.org/1 0.1 01 6/j.commatsci.201 2.04.028
1 86 L. Zhao et al. / Computational Materials Science 61 (201 2) 1 85–1 93
zone and the local fracture would occur at the crack tip when the
creep ductility of the material was exhausted. Furthermore, this
model had been widely applied in evaluating the creep crack
growth behavior [1 3].
In the present study, the prediction of creep crack growth
behavior of ASME P92 steel welded joint at elevated temperature
was carried out by the FEM and the theoretical models. Due to
the disadvantages of the prevailed FEM approaches, a new node re-
lease technique on the basis of the creep ductility exhaustion dam-
age model is adopted. The virtual creep crack growth process can
be achieved when the damage in the material reaches to the criti-
cal value [1 7]. Furthermore, for ASME P92 steel components used
in power plants, they were mainly fabricated by welding. The early
failure of welded joint had been reported, which always occurred
in the heat affected zone (HAZ) adjacent the base metal and was Fig. 1 . Dimension and size of a CT specimen used in the present study.
caused by creep crack, known as Type IV cracking [4,1 8,1 9]. There-
fore, the prediction of creep crack growth behavior in welded joint
is important to the safety assessment of the in-service compo- Table 2
Table 1
Chemical composition of ASME P92 steel (all data are given in weight-%.).
Material C Si Mn S P Cr Ni Mo W V Nb B Al N
P92 0.1 0 0.47 0.40 0.001 0.008 8.77 0.1 2 0.38 1 .48 0.1 6 0.054 0.001 0.02 0.043
L. Zhao et al. / Computational Materials Science 61 (201 2) 1 85–1 93 1 87
x¼ x
_ dt ¼ dt ð5Þ
strength and post-yield the material which follows von Mises flow 0 0 eÃf
theory with isotropic strain hardening. Material properties for
ASME P92 steel welded joint at 650 ° C is shown in Table 3, which In this theory, it is assumed that there is no damage occurring in
were obtained from the tensile tests and uniaxial creep tests of the specimen at beginning of the creep crack tests, which means
base metal (BM), welded metal (WM), simulated fine grain heat af- that x = 0; and then as the creep time increases, the creep damage
fected zone (FGHAZ) and coarse grain heat affected zone (CGHAZ). gradually accumulates. When the damage of an integration point
The simulated FGHAZ and CGHAZ were obtained by subjecting big- reaches to 1 or 0.999, this point is assumed to be failed. In addition,
ger steel samples to the expected thermal cycles which corre- it can be noted that the creep deformation is independent of the
sponded to the actual thermal cycles undergone by the FGHAZ creep damage accumulation. As a result, the creep strain rate can
and the CGHAZ during the welding process [21 ]. Tensile tests were not be accelerated by the damage evolution.
carried out on the cylindrical specimens with a cross-sectional
diameter of 5 mm and a gauge length of 25 mm at a constant strain
3. 3. Finite element model
rate of 0.02 minÀ 1 to obtain the elastic–plastic characteristics. Uni-
axial creep tests were performed on the rod-shaped specimens
Creep crack growth simulation was carried out using the finite
with a cross-sectional diameter of 1 0 mm and a gauge length of
element code ABAQUS. Due to the symmetry, only upper half of the
1 00 mm in the stress range of 1 40–1 80 MPa to obtain the creep
specimen with the initial crack located at the BM and WM is em-
properties. The geometry and size of tensile specimens and uniax-
ployed. The finite element mesh used in the analysis is shown in
ial creep specimens were all designed according to ASTM standards
Fig. 3, which is composed of two-dimensional four-node plane
E1 39-06 [22] and E8-04 [23]. Experimental procedures of the ten-
strain or plain stress elements. Eight-node quadratic elements with
sile tests and uniaxial creep tests used to generate the material
reduced integration normally are the first choice to carry out the
properties were published elsewhere [8,9,22].
analysis due to the crack tip singularity in the stationary creep
fracture mechanics. However, as a general for the crack extension
3. 2. Creep damage constitutive equation
analysis, four-node first-order elements can work best [24]. Fur-
thermore, the refined elements are employed in the crack tip and
An uncoupled single damage theory which is on the basis of the
the size of the smallest mesh is 1 5 lm, which is the experimentally
creep ductility exhaustion approach is adopted to investigate the
observed average grain size of ASME P92 steel. A finer mesh in the
creep damage accumulation [1 3]. The damage parameter x , de-
crack tip can be used to eliminate mesh dependency effects in the
fined ranging from 0 (no damage) to 1 (full damage), is employed
analysis.
to represent the creep damage accumulation. The damage rate is
The load was applied on the CT specimen using an analytical ri-
defined to depend on the equivalent creep strain rate as shown:
gid pin that represented the bolt in the experiment and was tied to
e_ c e_ c the hole of the specimen. The analytical rigid and the specimen
x
_ ¼ ¼ ð3 Þ
efÃ
MSFe f were tied through the contact surfaces in ABAQUS. A pre-crack in-
serted in the specimen was realized by the nodes that were not
where x is the creep damage rate; e is the equivalent creep strain
_ _
c
constrained in the y-direction. The creep crack propagation for
rate, which is determined by the average creep strain rate in this pa- one-half model is carried out using the node release technology.
per; eà is the critical multiaxial failure strain, which depends on the
f The schematic of this technology is illustrated in Fig. 4. In this anal-
uniaxial creep fracture strain e and the appropriate multiaxial
f ysis, it is assumed that the creep crack extension only occurs along
stress factor MSF. e depends on the material and the temperature
f the plane of crack tip, which is the symmetry plane of CT specimen.
and is 1 6% for the present ASME P92 steel at 650 ° C, which is the In addition, the crack tip is a sharp fronted flat crack, idealizing the
measured value of uniaxial creep tests. The form of MSF should be multiple micro cracks ahead of the crack tip during the actual tests.
calculated using an appropriate model. As a rule, the creep crack At beginning, the nodes of crack plane are all constrained in the y-
growth in Cr–Mo steel is caused by the creep voids. Therefore, in direction. When the value of damage at two integration point
the present study, the model proposed by Cocks and Ashby [1 6] belonging to the elements representing the current crack tip
which is based on the void growth and coalescence is used. The reaches to 0.999, the nodes of these elements are released, which
adopted MSF depends on the power creep law stress exponent ( n) means that the constraint of these nodes are deleted. Then the
and the stress triaxiality predicted which is defined as the ratio of creep crack propagates in further. The user-defined multi-point
the hydrostatic stress ( r ) and equivalent stress ( r ). The relation-
m e constraint subroutine in ABAQUS which can alter the constraints
ship can be written as: of the nodes during analysis is used to carry out the node releasing.
eÃf 2 n 05À n 05À rm
sinh sinh 2 ð 4Þ
: :
MSF ¼ ¼
ef 3 nþ0 5 : nþ0 5 : re
Table 3
Mechanical properties and creep properties of ASME P92 steel welded joint at 650 ° C.
Material Elastic modulus Yield strength A (hÀ 1 /MPa )n
n ef (%)
E (GPa) r (MPa)
y
BM 1 25 1 80 2.6353EÀ 1 6 5.23 16
WM 1 20 1 70 1 .4594eEÀ 20 7.63 16
FGHAZ 90 1 60 2.7567EÀ 1 8 6.80 16
CGHAZ 110 1 90 6.791 3EÀ 21 7.1 5 16 Fig. Two-dimensional FEM model of CT specimen for creep crack growth
3.
analysis.
1 88 L. Zhao et al. / Computational Materials Science 61 (201 2) 1 85–1 93
the regular elements in the crack tip plane are used ( see Fig. 3 ).
the test is also given by FEM analysis , due to the reduction in liga-
noted that the experimental data generally lie between the plane
numerical s imulations are in agreement with the experimental Fig. 6. Relations hip between FEM calculated creep crack growth length and
normalized time.
data. The relations hips between creep crack length and normalized
the current time of load application and tf is the life of creep crack
thickness of 1 0 mm is us ed and the stres s condition under testing
growth for each s pecimen, res pectively. The portions of the total
is close to the plain s tress condition.
creep crack growth taken by the linear relationship between creep
Representative creep damage accumulation contour of creep
crack length and normalized time of the FEM calculated res ults un-
crack propagation under the plain stres s condition for BM s peci-
der plain s tres s condition is nearly the s ame as that of experimen-
men is shown in Fig. 7 . It can be obs erved that the failed elements
tal results . While the calculated crack length under plain s train
are behind the current crack tip. In addition, the relatively uniform
condition is higher than that of the experimental res ult at the same
size and s hape of failed elements s uggest that steady creep crack
normalized time. This phenomenon may be caus ed by the dimen-
growth is occurring. As s tated in Eqs. ( 4) and ( 5 ), the creep damage
sion of CT specimen. In the present s tudy, CT specimen with a
accumulation depends on the s tress triaxiality. Therefore, the var-
reveals that under the plain s tress condition, the stress triaxiality
is relatively ins ens itive to the creep crack extension length ( see
Fig. 8 a) while under the plain s train condition, the s tres s triaxiality
varies as the creep crack length propagates ( see Fig. 8 b). The high-
creep crack length. Its value increas es as the creep crack grows. The
Ã
4. 2. Characterization of creep crack growth data using parameter C
temperature, the dis tribution of stress and strain rate in the crack
Ã
tip is us ually characterized by the parameter C . The creep zone in
the crack tip is s urrounded by the annular region where the elas-
Fig. 5 . Relationship between FEM calculated creep crack growth length and time. tic– plastic filed remains valid and the creep condition prevails .
L. Zhao et al. / Computational Materials Science 61 (201 2) 1 85–1 93 1 89
à 1 = ðnþ1 Þ
C
rij ¼ rij ð h
~ ; nÞ ð 6Þ
AIn r
3
eij Are
nÀ 1
_ ¼ Sij ð7Þ
2
1
Sij ¼ rij À d ij rkk ð8Þ
3
polar coordinates with the origin at the crack tip, which are referred
to the radial distance from the crack tip and the crack tip angle,
[ 2 0] :
à PV
_
c n 2
C ¼ þ 0: 5 2 2 ð9Þ
Bn ð W À aÞ n þ 1 1 À a= W
V
_
c ¼V
_
t À Ve À V p
_ _
ð 1 0Þ
where V
_
t ; Ve
_
and V
_
p are the total, elastic and plastic load line dis-
2
2K Bn a
_
V
_
e ¼ ð1 1 Þ
PE
Fig. 8. Variation of s tress triaxiality during creep crack growth ( a) under plain
ð m þ 1 Þ Jp Bn a_
V
_
p ¼ ð1 2Þ
P
Stres s fields in the crack tip under creep condition are addres sed by
The corresponding creep crack growth rates evaluated in terms creep crack extens ion. In ASTM E1 45 7 -07 [ 2 0] , it specifies that the
Ã
of C for experimental res ults and FEM res ults of BM and WM s pec- load dis placement rate ratio s hould be: V
_
c = Vt
_
! 0 : 5 . When this ra-
imens are shown in Fig. 9a and b, respectively. It can be obs erved tio is les s than 0.2 , the s tress intensity factor K is the dominant
that there is a tail occurring first in experimental and predicted re- parameter to correlate with creep crack growth rate. Fig. 1 0 s hows
s ults. After the tail, a linear relations hip between the creep crack the variation of load line displacement rate ratio during the calcu-
Ã
growth rate and C occurs , which represents the s teady creep crack lated creep crack proces s. For plain strain predictions , the load line
propagation. The creep crack propagation in Fig. 9 can be divided displacement ratio ranges from 0.65 to 0. 98 and the average value
into three bands , which represent plain strain prediction, plain is about 0.7 5 . While for the cas e of plain stres s predictions , this ra-
s tres s prediction and experimental data, res pectively. The plain tio keeps cons tant at a value of 0.95 . Therefore, the ratio exceeds
s train prediction exhibits a high creep crack growth rate and a 0.5 in all these s imulations, which stratifies the s pecification on
Ã
low value of C . The creep crack growth rate of plain s tress predic- the creep ductility failure in ASTM E1 45 7 -07 . Hence, it can be de-
Ã
tion is a little higher than that of experimental res ults but it con- duced that the relations hip between C and creep crack growth
Ã
verges to the experimental data at a high value of C . rate obtained from the FEM res ults is effective.
Ã
4. 3. Effectiveness of parameter C 4. 4. Prediction of creep crack growth behavior using theory models
Ã
As a rule, the parameter C is only applicable for the condition In addition to the prediction calculated by FEM, on the bas is of
that that the material shows creep-ductile behavior, where the creep ductility exhaustion theory and s tress filed ahead of crack tip
creep proces s zone near the crack tip is extensively wide. In ASTM as shown in Eq. ( 6), Nikbin, Smith and Webster als o propos ed that
E1 45 7 -07 , some specific terms of conditions are defined to ens ure the creep crack growth rate can be predicted as follows [ 1 3 ] :
that for plane stress condition e Ãf was taken to be the uniaxial failure
strain ef while for plane strain condition was e f/3 0, which was ex-
However, in this model, the stress state ahead of crack tip during
of crack tip. The value of e Ãf for different stress states could be calcu-
lated from the uniaxial failure strain and the MSF as mentioned in
Eq. ( 4).
s hown in Eq. ( 6), thus the MSF is als o a function of n and h. There-
Fig. 9. Comparison of FEM calculated creep crack growth rate with experimental
data ( a) BM specimen and ( b) WM s pecimen. Fig. 1 0. Variation of dis placement rate ratio V
_
c = Vt
_
during creep crack growth.
L. Zhao et al. / Computational Materials Science 61 (201 2) 1 85–1 93 1 91
Ã
r ð h nÞ
n = nþ 1 n
C the plain stres s NSW model underestimates it. However, the plain
1 = ð nþ 1 Þ 1 ~
e
;
¼ ð n þ 1 Þ ð Arc Þ ð 1 4Þ
MSFð h n Þ
a
_
In ef ;
s tres s and plain s train NSW-MOD models give more accurate lower
max
and upper prediction lines for the experimental res ults compared
In the present s tudy, A and n are s hown in Table 3 for different with NSW model. In particular, the NSW-MOD predictions under
s ub-regions of ASME P92 welded j oint; rc is taken as the size of the plain stres s condition are cons is tent with experimental data of
smallest element in the crack plane used in the FEM analys is , WM and BM s pecimens. It can be concluded that the NSW-MOD
l
n
r
~ ð h; n Þ
m; In and depend on the different values of n
e
which is 1 5 model can give appropriate creep crack growth rate. Hence, it
MSFð h ; n Þ
for different s ub-regions of welded j oint. Table 4 s hows the above can be us ed to evaluate the s afety of s tructures containing defects.
cons tants of modified NSW prediction model, which are calculated As analyzed in Sections 4.1 and 4.2 , the FEM res ults under plain
according to the reference [ 2 7 ] . s tres s condition exhibit agreement with experimental res ults. Fur-
Fig. 1 1 a and b s hows the predicted creep crack growth rates thermore, the NSW-MOD predictions under plain stres s condition
using NSW and NSW-MOD models for BM and WM specimens, are also consistent with experimental data. Therefore, it can be de-
res pectively. In addition, the prediction res ults are also compared duced that us ing the relations hip between the creep crack growth
Ã
with the experimental results. It can be obs erved that the plain rate and C obtained from the numerical simulations and coupling
s train NSW model overestimates the creep crack growth rate while with the modified NSW prediction models can be employed to give
the appropriate predicted res idual life for the actual in-s ervice
MSFð h ; nÞ max
mechanical testing technique, which s hows great potentiality as
Plain s train 15 4.98 1 1 .63 2 components in both nuclear and foss il fuel power plants [ 2 8 ,2 9] .
Therefore, this approach not only can reduce the cos t and time of
WM Plain s tres s 15 3 .1 4 1 . 944
Plain s train 15 4.7 3 1 1 .61 1 the experiment but als o is nondes tructive to the in-s ervice
components.
failure always occurs in the FGHAZ adj acent to the bas e metal
from each other. When the initial crack is located at the interface
the CGHAZ, which is 0.5 mm; the initial crack length and the initial
stress intensity factor K for all the cas es are kept constant, which is
1 /2
1 0 mm and 1 4 MPam , res pectively. The creep crack behaviors of
specimens with cracks pos ition located at the FGHAZ, the FGHAZ/
ary creep crack analys es under plane stres s condition are con-
HAZ. Because the node release technique can not be carried out
in the whole model, the creep crack extension is carried out us ing
and the remaining forces exerted by the element on adj acent nodes
are relaxed to a value near zero [ 6,9 ] . The creep crack length is cal-
imens with different crack locations. It can be noted that the creep
Fig. 1 1 . Comparison of NSW and NSW-MOD predicted res ults with experimental
the micro zones in the welded j oint. This means that the FGHAZ
data ( a) for BM s pecimen and ( b) for WM s pecimen. creeps more eas ily. As a res ult, the creep strain rate in this zone is
1 92 L. Zhao et al. / Computational Materials Science 61 (201 2) 1 85–1 93
5 . Conclusions
Fig. 1 2. A whole FEM model of CT specimen with crack in HAZ. In the present study, the predictions of creep crack growth
behavior in ASME P92 steel welded joint at 650 ° C are carried out
on CT specimens using FEM and a modified NSW model under
plain stress and plain strain conditions. The calculated results
and the predicted results are compared with experimental data un-
der the same condition. Furthermore, the effect of crack position on
creep crack growth is also investigated using FEM analysis. The
conclusions are listed as follows:
(1 ) The FEM predictions provide lower and upper creep crack
growth lines. The lower creep crack line under plain stress
condition is in agreement with experimental result.
(2) The FEM results all show higher creep crack growth rate
than the experimental data at the same value of C⁄ . How-
ever, the calculated results under plain stress condition con-
verge to the experimental results at high value of C⁄.
(3) The modified NSW model under plain stress condition gives
accurate prediction results for CT specimens with base metal
and welded metal compared with experimental results.
However, the modified NSW model under plain strain condi-
tion overestimates the creep crack growth rate at the same
value of C⁄.
Effect of crack location on creep crack growth behavior.
Fig. 1 3 .
(4) The FEM analysis coupled with the modified NSW prediction
model may be employed to give the appropriate predicted
residual life for the in-service components containing flaws
without conducting the creep crack growth tests.
(5) The specimen with crack in the FGHAZ/BM interface exhibits
a higher creep crack growth rate than those with cracks in
other zones of welded joints, i.e. FGHAZ, FGHAZ/CGHAZ
interface and CGHAZ/WM interface. This is due to low creep
strength and high constraint condition in this zone.
Acknowledgements
References
[3] A. Hernas, M . Imosa, B. Formanek, J. Cizner, J. Mater. Process . Technol. 1 5 7 – 1 5 8 [ 1 9] K. Shinozaki, H. K. Dej un Li, H. Harada, K. Ohis hi, T. Sato, Sci. Technol. Weld.
[ 4] L. Zhao, H. -Y. Jing, L. -Y. Xu, J.-C. An, G. -C. Xiao, D.-L. Xu, Y. -C. Chen, Y. Han, J. [ 2 0] ASTM , ASTM E1 45 7 -07 , Standard Test M ethod for M easurement of Creep Crack
M ater. Res . 2 6 ( 2 01 1 ) 93 4– 943 . Growth Times in M etals, ASTM I nternational, 2 007 , pp. 1 01 2 – 1 03 5 .
[ 5] D.R. Hayhurs t, P. R. Dimmer, C. J. M orrison, Philos . Trans . R. Soc. Lond. Ser. A 3 1 1 [ 21 ] Y. Shi, Z. Han, J. Mater. Proces s . Technol. 2 07 ( 2 008 ) 3 0– 3 9.
( 1 98 4) 1 03 – 1 2 9. [ 22] ASTM , Standard Test M ethods for Conducting Creep, Creep-Rupture, and
[ 6] R. J. Hayhurs t, R. Mus tata, D.R. Hayhurs t, Int. J. Pres . Ves . Pip. 8 2 ( 2 005 ) 1 3 7 – 1 44. Stres s -Rupture Tes ts of Metallic M aterials , ASTM E1 3 9-06, United States :
[ 7] T.H. Hyde, W. Sun, A.A. Becker, Int. J. Pres. Ves. Pip. 7 8 ( 2 001 ) 7 65 – 7 7 1 . American Society for Testing M aterials , 2 006.
[ 8] T.H. Hyde, A. A. Becker, Y. Song, W. Sun, Comput. Mater. Sci. 3 5 ( 2 006) 3 5 –41 . [ 23] ASTM , Standard Test Methods for Tension Testing of Metallic Materials, ASTM
[ 9] S.-T. Tu, P. Segle, J. -M. Gong, I nt. J. Pres . Ves . Pip. 8 1 ( 2 004) 1 99– 2 09. E8 -04, United States : American Society for Tes ting Materials, 2 004.
[ 1 0] A.A. Becker, T.H. Hyde, W. Sun, P. Anderss on, Comput. M ater. Sci. 2 5 ( 2 002 ) [ 2 4] M. Göting, J. Röler, Comput. Struct. 8 5 ( 2 007 ) 2 2 5 – 2 3 2 .
3 4– 41 . [ 25] M. Tan, N. J. C. Celard, K. M. Nikbin, G. A. Webs ter, Int. J. Pres . Ves . Pip. 7 8 ( 2 001 )
[ 1 2] L.G. Zhao, J. Tong, J. Byrne, Eng. Fract. Mech. 68 ( 2 001 ) 1 1 5 7 –1 1 7 0. [ 2 6] M. Yatomi, N.P. O’Dowd, K. M. Nikbin, G. A. Webster, Eng. Fract. M ech. 7 3 ( 2 006)
[1 3] K. M. Nikbin, D.J. Smith, G. A. Webs ter, Ser. A: M ath. Phys . Sci. 3 96 ( 1 98 4) 1 8 3 – 1 1 58–1 1 75.
[ 1 4] J. W. Hutchinson, J. Mech. Phys . Solids 1 6 ( 1 968 ) 1 3 – 3 1 . [ 28] D. T. Blagoeva, R.C. Hurs t, M ater. Sci. Eng. A 5 1 0– 5 1 1 ( 2 009) 2 1 9– 2 2 3 .
[ 1 5] J. R. Rice, G.F. Ros engren, J. Mech. Phys . Solids 1 6 ( 1 968 ) 1 – 1 2 . [ 2 9] P. Egan, M. P. Whelan, F. Lakes tani, M .J. Connelly, Comput. Mater. Sci. 40 ( 2 007 )
[ 1 7] M. Yatomi, C. M . Davies, K. M. Nikbin, Eng. Fract. Mech. 7 5 ( 2 008 ) 5 1 40– 5 1 5 0. [ 3 0] D. Li, K. Shinozaki, H. Kuroki, Mater. Sci. Technol. 1 9 ( 2 003 ) 1 2 5 3 – 1 2 60.
Pip. 8 3 ( 2 006) 63 – 7 1 .