Journal of Nuclear Materials
Journal of Nuclear Materials
Journal of Nuclear Materials
a r t i c l e i n f o a b s t r a c t
Article history: A physical-and-mechanical model of ductile fracture has been developed to predict fracture toughness
Available online 3 April 2014 and fracture strain of irradiated austenitic steels taking into account stress-state triaxiality and radiation
swelling. The model is based on criterion of plastic collapse of a material unit cell controlled by strain
hardening of a material and criterion of voids coalescence due to channel shearing of voids. The model
takes into account deformation voids nucleation and growth of deformation and vacancy voids.
For justification of the model experimental data on fracture strain and fracture toughness of austenitic
steel 18Cr–10Ni–Ti grade irradiated up to maximal dose 150 dpa with various swelling were used. Exper-
imental data on fracture strain and fracture toughness were compared with the results predicted by the
model. It has been shown that for prediction of the swelling effect on fracture toughness the dependence
of process zone size on swelling should be taken into account.
Ó 2014 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jnucmat.2014.03.047
0022-3115/Ó 2014 Elsevier B.V. All rights reserved.
596 B. Margolin et al. / Journal of Nuclear Materials 452 (2014) 595–606
Nomenclature
R
ef is the fracture strain of material for tensile cylindri- epeq ¼ depeq is the Odquist parameter where depeq is the
cal specimen equivalent of a plastic strain increment
ef_crack is the fracture strain of material near the crack tip req is the equivalent stress
Jc is the critical value of J-integral rY is the yield strength of material
D is the neutron irradiation dose in displacement per rul is the ultimate tensile strength of material
atom (dpa) rf is the true fracture stress
Tirr is the irradiation temperature rth
nuc is the threshold stress for deformation voids
Ttest is the test temperature nucleation
S is the swelling A is the parameter of stress–strain curve
qdef
v is the concentration of deformation voids n is the parameter of stress–strain curve
qvac
v is the concentration of vacancy voids ddef is the deformation void diameter
qmax
v is the maximum concentration of deformation void dvac is the vacancy void diameter
nucleation sites ldef is the distance between centres of deformation voids
rd is the effective local strength of matrix–inclusion lvac is the distance between centres of vacancy voids
interphase rf is the process zone size under ductile fracture
rnuc is the stress controlling deformation voids
nucleation
The aim of first part of this paper is to improve MKSK-model for where qdefv is the concentration of deformation voids per unit vol-
prediction of fracture strain and fracture toughness of irradiated ume of the material, qmax v is the maximum concentration of void
austenitic steels for different stress states, neutron irradiation nucleation sites, rnuc is the stress resulting in void nucleation, rd
doses and irradiation swelling values. It is understood that it is is the effective local strength of matrix–inclusion interphase.
not possible to take into account all features of deformation and Integrating Eq. (1) and assuming that the nucleation of voids
fracture micromechanisms of the irradiated austenitic steels. Nev- begins only at stresses rnuc P rth nuc we obtain
ertheless authors have tried to consider only basic mechanisms 8
controlling ductility of smooth and notched specimens and also < 0; for rnuc < rth
nuc
def
qv ¼ h i ð2Þ
fracture toughness of austenitic steels. Deformation features of : qmax r rth
v 1 exp nucr nuc ; for rnuc P rth
nuc
d
irradiated austenitic steels are taken into account on the basis of
stress–strain curves obtained directly from test results. where rth
nuc is the threshold stress for void nucleation.
For verification of the model applicability to the real irradiated In Eqs. (1) and (2) rnuc is the stress controlling nucleation of dis-
austenitic material the predicted results are compared with the continuity near some barriers and, in particular, nucleation of voids
obtained experimental data. or microcracks near some inclusions [10,11]. Equation describing
In the second part of this paper prediction of fracture at high level the stress of cleavage microcrack initiation [10,11] has been
of swelling when stress less than yield strength will be presented. derived from the analysis of stresses near the head of dislocation
pile-up
2. Main considerations of the physical–mechanical ductile
fracture model rnuc ¼ r1 þ mT e reff ð3Þ
result in a weakening of the inclusion–matrix interphase. In addi- length p2d. Thats why in this case vacancy voids coalescence may
tion, rd decreases due to more easily dislocation pile-ups forma- occur when the distance between voids lv ac ¼ pd2vac , i.e. when
tion under irradiation near the sites of void nucleation [10,11]. 3
Let us consider the neutron irradiation effect on deformation f ¼ 1dvac
vac
25:8% that corresponds to the irradiation swelling
voids nucleation. From Eq. (3) it follows that as the neutron dose S 34.8%. In most cases with S < 34.8% channel fracture will occur
D increases, rnuc increases too due to an increase in rY and, as a after a considerable voids growth when f reaches 25.8%. Such voids
consequence, in r1. As the dose D increases, rd decreases [10,11]. volume fraction corresponds to the stage of unstable voids growth
Then from Eq. (2) it follows that irradiation results in an increase in some cross-section of a unit cell when the condition
in voids concentration. This conclusion based on the above equa-
epeq P epeq collapse is already met, at least for tensile specimens.
tions is confirmed by the experimental data. Dimples concentra-
Therefore, with small irradiation swelling values (S < 34.8%) the
tion on the fracture surface of a specimen in irradiated condition p collapse p shear
has been shown to be higher than that in the initial one [2]. inequality eeq < eeq is valid. The case when
p collapse p shear
eeq > eeq is appear to be valid for specimens tested
(d) When analysing the growth of vacancy and deformation by torsion.
voids the Huang equation [13] modified by us is used. It The criterion of a unit cell plastic collapse or, in other words, the
should be noted that the Huang equation is valid for a single criterion of plastic instability is described in the following form
void in an infinite body. In the case when the distance [9,16]
between voids is comparable with their sizes the rate of void
growth increases due to an additional strain concentration dFeq
¼0 ð9Þ
in the vicinity of voids. To describe the voids growth allow- depeq
1
ing for their interaction an additional factor in the form of 1f
where Feq = req(1 AR ), req is the equivalent stresses acting in a
was introduced into the Huang equation
void-free matrix material, AR is the relative void area, i.e. the ratio
of voids cross-section area to the cross-section area of a unit cell
dVvoid 3a
¼ depeq ð4Þ with voids. It should be noted that when analysing the conditions
V void 1f of Eq. (9) the stress triaxiality is taken to be constant [9].
where according to equation given below [13] The proposed criterion describes the start of unstable voids
( growth. An unstable voids growth may occur when no necking
k 0:25; for rm
61
rm 3 rm req between each pair of voids takes place, as the conglomerate of a
a¼b exp ; k¼ rm ð5Þ
req 2 req 0; for req >1 matrix material with voids differing in site is considered. In other
words, the total necking between each pair of voids does not occur
It should be noted that the Huang equation seems to be applied to necessarily when an unstable voids growth occurs according to the
vacancy voids as their size of 10–30 nm is much larger than that of above criterion. So the criterion (9) allows one to calculate the frac-
the Burgers vector being a plastic deformation ‘‘quantum’’. ture strain ef irrespective of necking between voids.
In the general case deformation and vacancy voids may differ in Besides, it should be noted that there is no need to introduce
the coefficient b as vacancy voids are much smaller than deforma- some empirical critical parameters like a critical void volume frac-
tion ones. As the Huang equation was derived for deformation tion, a critical distance between voids or a critical void size when
voids, the value of b for such voids may be assumed to be equal using the criterion (9).
to 0.427. So we can write The parameter AR is calculated from the following consider-
( ations. As mentioned above, the value a in Eq. (5) may be different
bdef for deformation voids for deformation and vacancy voids. Then relying on Eq. (4) an
b¼ ð6Þ
bvac for vacancy voids increase in the volume of vacancy and deformation voids may be
calculated by equation:
where bdef = 0.427.
3a h i
In common case the value of bvac depends on material micro- dV def V def p def def p
eeq þ depeq qdef epeq ;
R ¼ R deeq þ V nuc qv v
structure and features of plastic deformation mechanisms. 1f
The value of bvac may be determined from the test results ð10Þ
obtained for specimens with different swelling.
In Eq. (4) Vvoid is the volume of a void; depeq is the equivalent of a 3a
plastic strain increment, f is the void volume fraction in a material dV vac
R ¼ V vac p
R deeq ; ð11Þ
1f
VR
f ¼ ð7Þ dV R ¼ dV def vac
V þ VR R þ dV R ð12Þ
In Eq. (7) VR is the total volume of vacancy and deformation voids in where V R ¼ VR
V
; V def
nuc is the volume of a nucleus deformation void.
the volume V of the matrix material.
It is assumed that the radius of nucleus deformation void and
(e) Unit cell fracture occurs when epeq ¼ ef calculated from the
V def
nuc do not depend on the type of void-nucleating inclusions. So,
condition
shear
collapse the radius of nucleus deformation void is taken to be the same
ef ¼ min epeq ; epeq ð8Þ for all inclusions.
When integrating Eq. (12) the initial conditions are formulated
collapse
where epeq is the strain meeting the conditions of unit cell in the form:
shear
plastic collapse, epeq is the strain when fracture occurs at the h i
expense of channel shearing of voids. Such a mode of fracture is V def
R 0
¼ V def def p
nuc qv eeq ¼ 0 ; V vac
R 0
¼S ð13Þ
called the channel fracture [2,3,14,15]. vac
Under shear deformation the growth of voids is hampered. A where V def
R 0 and V R 0 are the relative void volumes when
p
maximum increase in void size is restricted by p2 times when a eeq ¼ 0 for deformation and vacancy voids, respectively, S is the
spherical void with diameter d transforms into a microcrack with material swelling.The average voids area will be determined as
598 B. Margolin et al. / Journal of Nuclear Materials 452 (2014) 595–606
!2=3
VR 3. Simulation of material fracture under different irradiation
Avoid ¼ ð14Þ and testing conditions
qdef
v þ qv
vac
where qvac
v is the vacancy voids concentration.
The weld metal of 18Cr–10Ni–Ti steel in the initial and irradi-
Considering that the total area of voids per unit of area of a ated conditions was chosen for modelling. Welding was made
matrix material is calculated by the formula using 19Cr–11Ni–3Mo welding wire without subsequent heat
treatment. The weld metal specimens were irradiated in the
vac 2=3
AR ¼ Avoid qdef
v þ qv ; ð15Þ BOR-60 fast neutron reactor by neutron doses over the range from
6–7 to 46 dpa at a temperature of Tirr = 320–340 °C [1]. Tensile
we obtain
tests were carried out at a strain rate of about 103 sec1.
AR ¼ V 2=3
R : ð16Þ All the examined specimens were fractured by the ductile
mechanism due to void evolution (see Fig. 1).
The volume of a unit cell increases (1 þ V R ) times due to vacancy 18Cr–9Ni and 18Cr–10Ni–Ti grade steels are known to be
and deformation voids. widely used in atomic power engineering for manufacturing com-
Then the voids cross-section area in a unit cell, whose volume ponents to be highly irradiated. The choice of the weld metal to be
increased (1 þ V R ) times, can be calculated by the formula investigated instead of these steels was determined by the follow-
2=3 ing circumstances.
1
AR ¼ AR ; ð17Þ In the general case the dependence of ef on the test temperature
1 þ VR Ttest under ductile fracture is monotonically decreasing (at least at
Taking into account Eq. (16) and Eq. (17) we obtain Ttest P 20 °C) [1,9,18]. For the weld metal under consideration the
!2=3 behaviour of ef (Ttest) dependence is similar to ef (Ttest) dependence
VR for ductile fracture (it is a monotonically decreasing function). But,
AR ¼ : ð18Þ
1 þ VR as seen in Fig. 2, this dependence shows a local nonmonotonic
behaviour. The use of such experimental results for the model cal-
Lets compare number of input parameters for proposed model and ibration allows one to estimate the possibility of describing the
for widely used for prediction of material behaviour in ductile frac- local nonmonotony of material properties using the model as well
ture regime GTN- model [7,8]. According to latter model the plastic as to reveal material properties controlling the local nonmonotony
potential introduced by Gurson [7] is presented in the form of ef (Ttest).
r2eq 3 rm 2
U¼ 2f q 1 cosh q ½1 ðq1 f Þ ¼ 0; ð19Þ 3.1. Stress–strain curves for the weld metal
r2S 2 2 rS
This subsection gives the prediction of stress–strain curves
f for f 6 fc (SSCs) for the weld metal irradiated up to neutron dose D and
f ¼ : ð20Þ
fc þ ka ðf fc Þ for f > fc tested at temperature Ttest. These dependences were obtained upon
processing tensile test data on the weld metal irradiated in the fast
where rS is the function describing the yield surface for material
neutron reactor BOR-60 [1]. The dependencies are input informa-
without voids, q1, q2, fc and ka are material constants;
tion for calculation of SST for different types of tested specimens.
The increase of void volume fraction is written
The stress–strain curve is approximated by the following equa-
df ¼ df nucl þ df growth ; ð21Þ tion [1]:
n
where req ¼ rY þ A epeq ð25Þ
df growth ¼ ð1 f Þdepii ; ð22Þ
where A and n are the parameters describing strain hardening of a
In (Eq. (22)) depii is the sum of normal plastic strain. Value of df nucl
p
deeq material.
may be calculated by formula [17]
df nucl fN 1 epeq eN
p ¼ pffiffiffiffiffiffiffi exp ; ð23Þ
deeq SN 2p 2 SN
0.9
0.8 0.9
0.7 0.8
0.6 0.7
0.5 0.6
εf
0.5
0.4
εf
0.4
0.3
0.3
0.2 0.2
0.1 0.1
0 0
0 200 400 600 0 200 400 Ttest, °C
Ttest, °C
(a) (b)
Fig. 2. Temperature dependence of the fracture strain of a material irradiated by the dose D = 46 dpa at Tirr = 320–340 °C: s are the experimental data; solid line is a curve
predicted on the basis of SSCs calculated from Eqs. (25)–(34) (case ‘a’); N are the fracture strains predicted on the basis of individual SSCs for each specimen (case ‘b’).
In Eq. (25) the dependence of rY on Ttest and D for the investi- r0ul ðT test Þ ¼ 439 þ 222 expð9:74 103 T test Þ; MPa ð34Þ
gated weld metal can be calculated by the formula [1]:
where the increment of ultimate tensile strength under irradiation
rY ðT test ; DÞ ¼ r0Y ðT test Þ þ DrYG ðDÞ ð26Þ is described by equation:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
where Drul ðDÞ ¼ 440 1 expð0:25 DÞ MPa; ð35Þ
r0Y ðT test Þ ¼ 255 þ 420 exp 2:22 103 ðT test þ 273Þ ; MPa: ð27Þ As is shown in [1], the stress–strain curve parameters (rY, A and n)
do not depend on irradiation temperature (at least in the range
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Tirr = 320–450 °C) if radiation swelling is small. If it is not possible
DrYG ðDÞ ¼ 498 1 expð0:3 DÞ; MPa: ð28Þ to neglect swelling, the above equations may be applied only for
In Eq. (26) r is the yield strength of the weld metal in the initial
0 the matrix material.
Y
condition, DrYG is the increment of the athermal (temperature inde-
pendent) part of yield strength. 3.2. Simulation of fracture strain – test temperature dependences
In Eq. (26) et seq. Ttest is the test temperature in °C, D is the neu-
tron dose in dpa. The aim of this subsection is to predict the dependence of
The parameters A and n in Ref. [1] were determined on the basis fracture strain of an irradiated material on test temperature and
of tensile test results by the following way. It is used two pairs of compare the predicted results with experimental data. The weld
calc metal irradiated up to 46 dpa at Tirr = 320–340 °C and tested in
data for each specimen: (ef; rf) and ecalc ul ; rul exp eul , where rf is
the temperature range from 80 °C to 425 °C is considered.
the true fracture stress, rul is the ultimate tensile strength; ecalc
ul is
For calibration of the basic model parameters rd and qmax v the
the calculated value of plastic strain at which the loss of tensile
experimental data on fracture strain, eexp , at Ttest = 80 °C are used.
cylindrical specimen carrying capacity takes place. In other words, f
ecalc is the plastic strain corresponding to the specimen necking. The pair of parameters rd and qmaxv is formed in such a way that
ul
The value of ecalc was calculated from the necking condition [1]: the calculated fracture strain ecalc
f at T = 80 °C is assumed to be
ul
equal toeexp
f . Relying on the calculations performed, the following
dreq
¼ req : ð29Þ parameter values were chosen: rd = 1115 MPa, qmax ¼ 8 106
depeq ep calc
v
eq ¼eul mm3 . The parameters rd and qmax are taken as independent of
v
Processing the data obtained on the parameter A gives the following Ttest.
dependence of A on Ttest and D [1] Others parameters are chosen from the following
considerations
AðT test ; DÞ ¼ 734:4 0:77 T test 0:337 DrY ðDÞ; MPa: ð30Þ
The parameter n, according to [1], is calculated from the known (1) Radiation swelling from measurements of specimens is close
values of A, rY and ultimate tensile strength rul by solving the to zero. Therefore, when simulating fracture under the spec-
combined equations: ified irradiation condition the effect of vacancy voids on ef is
neglected.
ecalc
ul (2) The parameter mTe is assumed to be constant. The value of
n¼ rY ; ð31Þ
1r mTe is taken as 1.0 [12].
ul exp ðecalc
ul Þ
(3) The value of rth
nuc in Eq. (2) is taken to be equal to rY.
rY
A¼ : ð32Þ As a fracture criterion of a tensile cylindrical specimen the frac-
calc n1
e ul n ecalc
ul
ture of the specimen neck centre is taken. To describe the SST in the
The temperature-dose dependence of ultimate tensile strength for centre of specimen neck qm rreqm
and q1 rreq1 as a function of plastic
the investigated weld metal is calculated in this case by equation strain the Bridgman formulas [19] is used in the following form
[1]: 8
< 13 ; for epeq 6 eul
rul ðT test ; DÞ ¼ r 0
ul ðT test Þ þ Drul ðDÞ ð33Þ qm epeq ¼ 1 ep e
: þ ln 1 þ 3 1 exp eq ul
3 2 3
; for epeq > eul ;
where the temperature dependence of ultimate tensile strength in
the initial condition is described by equation ð36Þ
600 B. Margolin et al. / Journal of Nuclear Materials 452 (2014) 595–606
2
the dose D = 46 dpa at Tirr = 320–340 °C are used for predicting
q1 epeq ¼ qm epeq þ : ð37Þ
3 the ductility of the weld metal irradiated by the dose D = 49 dpa
at Tirr = 400–450 °C.
In Eq. (36) eul is the value of plastic strain corresponding to the spec-
Numerical simulation of the swelling effect on fracture strain is
imen necking.
based on the following considerations:
The prediction is carried out for two cases:
(1) The SSCs for a matrix material, i.e. for material without
(a) For the case ‘a’ the SSCs calculated by dependences pre-
vacancy and deformation voids, are curves invariant to
sented in the previous subsection (see Eqs. (25)–(35)) are
swelling and irradiation temperature and depend only on a
used.
neutron dose and test temperature.
(b) For the case ‘b’ the individual SSCs obtained on the test
(2) When calculating fracture strain for each temperature the
results for each specimen are used; these curves are pre-
individual swelling values for each specimen [18] are used.
sented in Fig. 3.
(3) The parameters rd and qmax
v values are determined from the
equality condition eexp ¼ e calc
at Tirr = 320–340 °C (i.e. with-
In the range of Ttest = 80–495 °C the fracture strain ecalc
f values f f
were calculated from the plastic collapse condition Eq. (9). out swelling [1]) and Ttest = 80 °C; these values are the same
Fig. 2 compares the experimental data and the calculated as in the previous subsection.
dependences ecalc (Ttest) for both cases (‘a’ and ‘b’). (4) It is assumed that b = 0.427 in Eq. (5) for both deformation
f
As is seen from Fig. 2, the experimental data are in good agree- and vacancy types of voids.
ment with the model-calculated results. The data obtained show
that with invariant values of rd and qmax the model makes correct The results of fracture strain calculations in the absence of
v
ef predictions at different Ttest, especially with taking into account swelling and with it are shown in Fig. 4.
the peculiarities of individual stress–strain curves for each Ttest. It is seen from the figure that there is a good agreement
The anomalously low value ef at Ttest = 350 °C is determined by between the experimental and calculated data for specimens with
the anomalously low strain hardening of the material of specimen and without swelling (in case of specimens without swelling, the
tested at this temperature (see Fig. 3). Evidently, the given specimen values of eexp
f and ecalc
f are presented in Fig. 2). The results obtained
was cut out from the weld zone with anomalous properties. Never- suggest that a decrease in ef at Tirr = 400–450 °C compared with ef
theless, it should be noted that a relation between ef and strain at Tirr = 320–340 °C is connected solely with a vacancy void volume
hardening of a material is observed even for anomalous properties. fraction that determines the material radiation swelling. It should
Thus the effect of Ttest on ef is basically determined by the influ- be noted that a nonmonotonic behaviour of the dependence eexp f
ence of strain hardening on the plastic collapse condition for a unit (Ttest) at Tirr = 400–450 °C is evidently determined by the inhomo-
cell. geneity of specimens swelling. At the same time, the average value
of the fracture strain ef in the temperature range of 80–425 °C
3.3. Simulation of the swelling effect on fracture strain obtained experimentally approaches the calculated value:
eexp
f ¼ 0:26; ecalc
f ¼ 0:25.
For analysing the swelling effect on the material ductility the Let us estimate the effect of radiation swelling on the relative
weld metal in two conditions (with and without swelling) is con- eSw
fracture strain f
where eSw and ef are the fracture strains for mate-
sidered. These conditions were reached by neutron irradiation of ef f
the weld metal in BOR-60 reactor practically with the same doses rial irradiated by the same neutron dose with and without swell-
but at the different irradiation temperature Tirr. One set of ing. The following input information is used for simulation of the
specimens was irradiated by the neutron dose D = 46 dpa at eSw
f
dependence ef on swelling.
Tirr = 320–340 °C; swelling of these specimens is close to zero.
The second set of specimens was irradiated by the dose
D = 49 dpa at Tirr = 400–450 °C; swelling of these specimens varies
from 3% to 7% [18].
For the swelling effect on the material ductility to be predicted, 0.7
the model parameters obtained for the weld metal irradiated by
0.6
600 0.5
80°C
200°C 0.4
290°C
εf
0.3
400 340°C
σeq - σY , MPa
3.4%
0.2 6.4%
6.8%
4.0%
0.1
200
0
425°C 0 100 200 300 400 500
350°C Ttest, °C
(1) The parameters rd and qmax v values are taken the same as in According to [1], the dependence
Def
ðDÞ for the weld metal
e0f
the previous calculations.
(2) The strain hardening parameters for a matrix material are tested at Ttest = 80 °C is nearly the same as the corresponding
calculated for the irradiated condition corresponding to the dependence at Ttest = 290–350 °C. Therefore, for the specified Ttest
saturation of mechanical properties for the matrix material the coefficients Ae and Be can be taken the same: Ae = 0.52 and
according to the formulas presented in [1]. It should be Be = 0.37 dpa1.
noted that the mechanical properties saturation of 18Cr– At the Ttest = 80 °C the value of e0f are taken as equal to 1.23 [1].
10Ni–Ti steel and its weld metal, especially of the fracture Fig. 6 shows the calculative dependence rd(D) and dependence
strain, occurs in the dose range of 10–20 dpa [1]. At the same of DrY = DrYG on neutron dose calculated by Eq. (28).
time, in the general case radiation swelling for these doses is As is seen from the figure, the dependence rd(D) correlates well
sufficiently small [20,21]. So we can accept that for the doses with the dependence DrY(D).
at which swelling occurs, the mechanical properties of a With changing a neutron dose from 0 to 6 dpa a rapid change in
matrix material do not change anymore. That is why the rd and DrY occurs. On further increase in neutron dose the values
use of strain hardening parameters corresponding to the sat- of rd and DrY remain nearly unchanged. The results obtained are
uration of mechanical properties allows one to describe the agreed with theoretical dependencies on the influence of radiation
swelling effect on fracture strain properly. defects on rd. [10,11,22].
(3) The strain hardening parameters are calculated at test tem- A decrease in rd is connected with several mechanisms, in
perature equal to 450 °C which is close to the temperature of particular, with the segregation of different impurities at the inter-
maximum swelling for 18Cr–10Ni–Ti or similar grade steels phase boundaries and with the internal stress at these boundaries
[20,21]. Such an approach gives a conservative swelling due to dislocation loops formation [10,11,22]. These processes
effect estimation. result in the weakening inclusion–matrix interphase. In addition,
rd decreases due to an easier formation of dislocation pile-ups near
Fig. 5 shows the dependence of calculated relative decrease in void nucleation sites under irradiation. The mentioned mecha-
fracture strain on swelling and its approximation by the following nisms of decreasing rd due to the formation of dislocation loops
exponential function and pile-ups are unambiguously connected with the yield strength
increment DrY.
eSw
f Thus the results obtained suggest that the model may describe
¼ expða Sb Þ ð38Þ
ef the influence of a neutron dose on the fracture strain ef if the influ-
ence of D on rd is taken into account.
where a and b are the material constants; a = 7.47, b = 0.69; these
values are obtained by least-squares method.
3.5. Determination of the model parameters
3.4. Analysis of the effect of a neutron dose on the parameter rd
Based on the performed analysis, the following procedure of
For the analysis of the effect of a neutron dose on the parameter determining the model parameters may be proposed.
rd we considered the weld metal irradiated over the neutron dose Stress–strain curves are determined by processing smooth ten-
range from 0 to 46 dpa at Tirr = 320–340 °C and tested at Ttest = 80 °C. sile specimen test results.
Generally, with simulating a fracture strain temperature depen- The values of qmax
v and rd can be determined unambiguously by
dence for the irradiated material we took meff = 1 and qmax v ¼ 8 a tensile test of specimens of two types: a smooth cylindrical spec-
106 mm3 . imen and a cylindrical specimen with a circumferential notch. The
The parameter rd is determined from the equality condition values of ef for these specimens will differ because of a difference in
eexp ¼ ecalc for a given dose. The dependence of eexp on the dose D stress triaxiality. Thus, the values of qmax
v and rd can be determined
f f f
may be obtained by the following dependence [1] from the equality condition eexpf ¼ ecalc
f for both types of specimens.
A similar approach was used in the paper [16].
Def
¼ Ae ½1 expðBe DÞ0;5 ; ð39Þ Since qmax
v does not depend on an irradiation dose, it is reason-
e0f able to determine this parameter from the test results for a mate-
rial in the initial condition.
where De ¼ e0f ef ; e0f is a fracture strain of material in initial
condition.
10000 600
1
8000
0.8
400
ΔσY, MPa
6000
σd, MPa
0.6
4000
200
0.4 2000
0 0
0.2 0 10 20 30 40 50
0 0.02 0.04 0.06 0.08
D, dpa
Swelling S
Fig. 6. Dependences of rd and DrY on neutron dose: s and —— are the calculated
eSw
Fig. 5. Dependence of relative decrease in fracture strain ef on radiation swelling: values rd and their approximation respectively; –– –– –– is the dependence of yield
f
s is the model-calculated prediction, the curve is the approximation by Eq. (38). strength increment on the dose D, see Eq. (28).
602 B. Margolin et al. / Journal of Nuclear Materials 452 (2014) 595–606
The dependence rd(D) may be presented in the form similar to ductile fracture may be presented in the form similar to that pro-
that given in Refs. [10,11,22] relying on the analysis of the physical posed in the paper [16]:
mechanisms of radiation damage effects on rd:
J c ¼ g rflow r f ef crack ð41Þ
rd ¼ r0d rmin
d exp ½CD Dm þ rmin
d ð40Þ
where Jc is the critical value of J-integral, rf is the process zone size,
where r0d is the value of rd for a material in the initial condition, efcrack is the fracture strain near the crack tip determined with
rmin is the minimum value of rd for a material; CD and m are the regard to SST typical of a crack, g is a some numerical coefficient
d
material constants. equal 36 [26], rflow is the flow stress
From Eq. (40) it follows that for the coefficients in Eq. (40) to be rY þ rul
determined it is necessary to have at least test results of a material rflow ¼ : ð42Þ
2
in the initial condition and irradiated up to 3 different doses.
As is known, the process zone size under ductile fracture by voids is
related to the distance between voids l and may be estimated by a
4. Analysis of the irradiation effect on fracture toughness simple formula
First of all let us estimate the influence of SST on the depen- rf ¼ k 1 ð43Þ
ef
dence ef(D) and compare the dependence e0f
ðDÞ deduced from Eq. where k is the numerical coefficient; k = 2–8 [16,27,28].
(39) and the dependence
ef crack
ðDÞ calculated by the model. For Assuming that no new deformation void sites arise under irra-
e0f crack diation, the process zone size rf may be taken to be invariant with
ef crack let us take ecalc
f calculated for the stress state typical for a respect to a neutron irradiation dose. The invariance of rf is valid if
material near a crack tip on the line of its extension. The values there is no swelling. In the general case, when neutron irradiation
of e0f and e0f crack is the fracture strain of a material in the initial con- is accompanied by swelling, rf should be swelling-dependent.
dition for a smooth tensile specimen and a cracked specimen, When swelling is close to zero, rf is related to the distance between
respectively. deformation voids ldef : r f ¼ rdef
f . When the value of swelling is
The calculation of qm ðepeq Þ and q1 ðepeq Þ for tensile specimens was large, rf is connected with the distance between vacancy voids
made according to the Bridgman formulas (36)and (37) [19], while lvac : rf ¼ r vac
f . The dependence of rf on swelling may be presented
calculations for a material near the crack tip were made by the in the following form
dependences described in [23] (see Fig. 7). (
The calculation by the model
ef crack
ðDÞ was made for Ttest = r def
f x S; when S < S
e0f rf ¼ ; ð44Þ
when S P S
crack
r vac
f ;
290 °C. The choice of such a temperature is related to the available
representative data on Jc(D) for Ttest = 290–350 °C [24]. Table 1 where x is the coefficient of decreasing process zone size due to
ef ef crack
gives the calculation results for e0f
by Eq. (39) and e0f
calculated swelling, S is the some value of swelling for which the condition
crack
4 So for the case where swelling is close to zero, rf in the initial con-
dition is the same as in the irradiated conditions. Then
Jc ef crack rflow
3 ¼ ð47Þ
J 0c e0f crack r0flow
where Jc and J 0c are the critical J-integral values for an irradiated
qm, q1
Table 1
ef ef crack Jc
Values of ;
e0f e0f
and J0c
at different doses and Ttest = 290 °C.
crack
0 1 0.0551 1 413 1
27 0.480 0.0044 0.08 883 0.16
46 0.480 0.0044 0.08 883 0.16
rf
the dependence rf(S). Fracture toughness values were determined
as per the standard ASTM 1820 [30].
rfdef The critical value of J-integral was calculated by the formula
Jc = min(JQ, Jbr), where JQ is determined according to the standard
ASTM 1820 [30]; Jbr is the value of J-integral at quasi-brittle frac-
ture of specimen.
Before tests the swelling of every specimen was determined by
a hydrostatic weighing method.
The tensile test results of tensile cylindrical specimens and
tensile cylindrical specimens with a circumferential notch are
rfvac presented in Tables 2 and 3. Besides, Table 2 reports the strain
hardening parameters A and n obtained for every specimen by
S
the procedure proposed in [1]. The tensile tests were carried out
Fig. 8. Scheme of a dependence of the process zone size rf on swelling S. at a strain rate of about 1.0 103 sec1.
The values of ef presented in Tables 2 and 3 were calculated by
the formula ef = ln(1 Z/100%), where Z is the reduction of area,
Pmax and Pf are the maximum load and fracture load of tensile cylin-
5. Effect of swelling on fracture toughness drical specimens with a circumferential notch, correspondingly
It should be noted that strain distribution over the central
For experimental investigation of the effect of swelling on cross-section of the neck of cylindrical tensile specimens is uni-
fracture toughness the material of BOR-60 reactor shield assembly form. Since the SST is higher in the specimen centre, the this place
E-65 was used. This assembly operated in the BOR-60 reactor for is targeted for fracture initiation. The stress and strain distribution
41 years. The neutron dose for different assembly components var- in the neck of the tested notched specimens is similar to that in the
ied over the range from 90 to 150 dpa. Swelling of assembly com- neck of cylindrical tensile specimens. The notch radius was
ponents varied over the range from 0% to 6% as a result of height selected so that at Z > 2% the strain distribution in the neck is close
distribution of irradiation temperatures and neutron doses. The to uniform distribution [16]. Calculations made for notched speci-
shield was made of 18Cr–10Ni–Ti grade steel with high carbon mens show that the SST hardly depends on the strain and may be
content (0.11 wt.%). taken as qm = 0.9 and q1 = 1.57.
Three types of specimens were cut out from the shield assembly The SST for cylindrical tensile specimens is strain dependent
core: cylindrical tensile specimens, cylindrical tensile specimens and is calculated from the Bridgman equation (see Section 4).
with a circumferential notch and surface edge bending (SEB) spec- The fracture toughness results for SEB-specimens are presented
imens. The drawings of specimens are presented in Fig. 10. in Table 4.
The test results of smooth and notched specimens at Ttest = It should be noted that in spite of the quasi-brittle type of load–
290 °C were used for calibrating the parameters of the model to displacement diagram for SEB specimens with 3–6% swelling, in all
predict the value of ef_crack. The results of fracture toughness tests the cases fracture proceeded by the void nucleation and growth
of SEB specimens at Ttest = 290 °C were used for determination of mechanism. This conclusion is based on SEM analysis of fracture
350 1.4
300 1.2
250 1
Jc, N/mm
200 Jc 0.8
150 J 0c 0.6
100 0.4
50 0.2
0 0
0 10 20 30 40 50 0 10 20 30 40 50
Dose D, dpa Dose D, dpa
(a) (b)
Fig. 9. Experimental data on fracture toughness of the weld metal tested at Ttest = 290 °C [24] (s) and data predicted by modelling (d).
604 B. Margolin et al. / Journal of Nuclear Materials 452 (2014) 595–606
Fig. 10. Drawings of cylindrical tensile specimens (a), cylindrical tensile specimens with a circumferential notch (b) and surface edge bending (SEB) specimens (c) cut out
from the shield assembly of BOR-60 reactor.
Table 2
Tensile test results and strain hardening parameters for cylindrical specimens cut out from the shield assembly E-65 of BOR-60 reactor irradiated over the range of doses from 90
to 150 dpa.
(a) (b)
fatigue
precrack
(c)
Fig. 11. Typical fracture surfaces of tested SEB specimens with different swelling cut out from the component of the shield assembly E-65 after operation in the BOR-60
reactor: (a) S = 0.4%, (b) and (c) S = 5.5% with different scale.
0.9
140
0.8
120
0.7
0.6 100
0.5
80
εf
rf , μm
0.4
60
0.3
0.2 40
0.1
20
0
0 2 4 6
0
Swelling S, % 0 2 4 6
Swelling S, %
Fig. 12. Dependence of fracture strain on radiation swelling for smooth (s, ——)
and notched (d, – – –) cylindrical specimens: experimental data (s, d), predicted Fig. 13. Effect of the swelling S on the process zone size rf with linear
curves (——, – – –). approximation.
The value of rf was determined for every fracture toughness dvac 20 nm [29] estimation of rvac f ¼ klvac with regard for Eq.
tested specimen based on Eq. (41). The value of ef_crack in Eq. (41) (45) at S = 5.2% gives (0.11
0.44) lm.
was calculated by the model for the dependences qm ðepeq Þ and Fig. 14 shows the comparison of experimental data and the pre-
q1 ðepeq Þ typical of the material near the crack tip (see Section 4). dicted curve Jc(S) with regard for the linear dependence rf(S) and
The experimental values of Jc were taken from Table 4. with regard for assumption that r f ¼ const ¼ rdeff . It is seen that
Fig. 13 shows the values of rf determined for the specimens with the experimental data are suitably described by the predicted
differing swelling. From the figure it is seen that rf decreases line- dependence. This indicates that rf is adequately approximated by
arly with increasing S. Linear approximation of presented data by a linear function. Assumption that r f ¼ const ¼ rdef f results in
Eq. (44) for S < S⁄ gives the following values of parameters: underestimation of swelling effect on fracture toughness.
r def
f ¼ 0:1 mm and x = 1.884 mm. The comparison of Figs. 12 and 14 shows that swelling has a
Value of swelling S⁄ corresponding to the condition r f ¼ rvac
f is stronger effect on Jc than on ef. Such a strong effect is related not
estimated to be equal 5.2% (see Fig. 13). Assuming that value only to high stress state triaxiality but to a considerable decrease
606 B. Margolin et al. / Journal of Nuclear Materials 452 (2014) 595–606