Engineering Fracture Mechanics 99 (2013) 239250

Contents lists available at SciVerse ScienceDirect

Engineering Fracture Mechanics

journal homepage: www.elsevier.com/locate/engfracmech

Loss of constraint during fracture toughness testing of duplex

stainless steels
Johan Pilhagen , Rolf Sandstrm
Materials Science and Engineering, KTH, Brinellvgen 23, S-10044 Stockholm, Sweden

a r t i c l e

i n f o

Article history:
Received 2 December 2011
Received in revised form 10 September
2012
Accepted 8 January 2013

Keywords:
Duplex stainless steel
Fracture toughness
Delamination
Splits
Pop-in
Normalization
Master curve

a b s t r a c t
Delamination of the fracture surfaces, so called splits, is an important phenomenon that
occurs at sub-zero temperature for hot-rolled duplex stainless steels during impact and
fracture toughness testing. To evaluate how the splits inuence the fracture toughness,
sub-zero temperature fracture toughness testing of 50, 30 and 10 mm thick plates of hot
rolled 2205 duplex stainless steel was performed. The results show that the splits cause
loss of constraint along the crack front. This can be observed as local difference in crack
growth in the specimen. The initiation fracture toughness is not inuenced by the specimen thickness. Furthermore, due to the delamination the material exhibits a stable fracture
process despite the presence of cleavage fracture. This is interfering with the master curve
method so for evaluating the fracture toughness at sub-zero temperatures an assessment
of the fracture resistance curve is instead suggested. For assessing the brittle crack behaviour at sub-zero temperatures it is proposed to use the split initiation as a failure criteria.
The splits are also the cause of the pop-in behaviour observed for the duplex stainless
steels. The susceptibility for pop-in is inuenced by the microstructure.
2013 Elsevier Ltd. All rights reserved.

1. Introduction
Reported fracture toughness values for duplex stainless steels (DSSs) demonstrate high toughness at low temperatures
[15]. A common feature is secondary cracks, so called splits, which are delamination of the material growing normal to
the plane of the fatigue crack. They are often seen after fracture toughness and impact testing when tested through-thickness
along the rolling direction (TL) [16]. Their size, numbers and proximity to the fatigue crack increase with decreasing temperature [7]. It has been proposed that splits affect the scatter in fracture toughness measurements [3].
The splits are assumed to occur in the ferritic phase or at austenite/ferrite phase boundaries and the orientation of the
splits is governed by the orientation of the fracture plane in relation to the microstructure. A TL specimen will have splits
parallel to the crack growth direction while a TS specimen will have splits perpendicular to the crack growth direction.
Delamination of the fracture surface during fracture toughness testing has been observed for a range of materials and the
mechanism has been explained by the existence of weak planes in the material which delaminate under testing due to the
through-thickness stress [8]. The explanations for these weak planes have either been the interaction of different crystallographic orientations or banding of weaker phases. Two examples of the former case is the splits in a ferritic stainless steel
which was found to be due to texture banding from the rolling procedure [9] and the splits found in a pipeline steel where
rolling at low nishing temperature resulted in a texture that promoted cleavage fracture parallel to the plane of rolling [10].

Corresponding author. Tel.: +46 8 7906252.

E-mail address: pilhagen@kth.se (J. Pilhagen).
0013-7944/$ - see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.engfracmech.2013.01.002

240

J. Pilhagen, R. Sandstrm / Engineering Fracture Mechanics 99 (2013) 239250

Nomenclature
Ap
b0
B0
B1T
BN
C1
C2
E
Je
JIc
Jpl
JQ
K
KJc(0)
KJc(1T)
Kmin
m
M
T0
d
Da

g
t
rys
rY
CMOD
DSS
LOM
SEM

plastic work
initial ligament length
plate thickness in Fig. 3 or specimen thickness in Eq. (4)
reference thickness
net thickness of the specimen
coefcient from the power law regression of the JDa data
coefcient from the power law regression of the JDa data
Youngs modulus
elastic component of the J-integral
initiation fracture toughness
plastic component of the J-integral
tentative initiation fracture toughness
stress intensity factor
measured fracture toughness at the point of instability
size adjusted fracture toughness
threshold value for cleavage fracture
constraint parameter, function of crack depth and strain hardening exponent
non-dimensional deformation limit
reference temperature
crack-tip opening displacement (CTOD)
crack extension
dimensionless parameter
Poissons ratio
yield strength at the test temperature
effective yield strength at the test temperature
Crack Mouth Opening Displacement
duplex stainless steels
light optical microscope
Scanning Electron Microscope

In the case of banding of weaker phases there is an example of another pipeline steel where the thermomechanically process
caused thin layers of martensitic or bainitic grains to occur through the thickness parallel to the rolling direction [11].
In the work by Nilsson [6] the impact toughness of 50 mm and 12 mm plates of 2205 duplex stainless steel were tested
from room temperature down to liquid nitrogen temperature for all the major orientations in Fig. 1. The conclusions were
that the specimens are toughest when the notch is oriented in the normal plane (LS, TS and 45-S) followed by the case
when the notch is oriented perpendicular to the normal plane (LT, TL and 4545). The lowest toughness is reached when
the notch is oriented parallel to the normal plane (SL, ST and S-45) which is the same plane as for the splits. It is interesting
to note that the 12 mm plate has higher toughness than the 50 mm plate in all orientations except for the last-mentioned
where the 12 mm plate transition region has moved to higher temperature and lower upper-shelf energy compared to
the 50 mm plate. A texture characterisation showed that the ferrite in the 12 mm plate had strong texture while the ferrite
in the 50 mm plate was more prone to exhibit random texture. However, no unambiguous effect of the crystallographic orientation could be related to anisotropy of the impact toughness [6].

Fig. 1. Schematic drawing of specimen orientations. L is the rolling direction, T is the transversal direction and S is the through-thickness direction.

241

J. Pilhagen, R. Sandstrm / Engineering Fracture Mechanics 99 (2013) 239250

The objective of this paper is to analyse how the splits inuence the fracture toughness and how the microstructure affects the splitting phenomenon in hot-rolled duplex stainless steels. The approach was to test single-edge notched bend bars,
SE(B), with different dimensions from a 50 mm plate and compare these with 30 mm plate and 10 mm plate.

2. Material and testing procedure

2.1. Material
The material used in this work was a commercially produced duplex stainless steel designated 2205 (EN 1.4462, UNS
S32205), produced by Outokumpu Stainless. The material was hot-rolled to the desired plate thickness (50 mm, 30 mm
and 10 mm) followed by solution treating at 1100 C and water quenching. The chemical composition can be found in
Table 1.
The 50 and 30 mm plate have almost the same tensile properties at room temperature with yield strength of 496 MPa and
492 MPa and tensile strength of 739 MPa and 737 MPa respectively. For the 10 mm plate the yield strength is 604 MPa and
the tensile strength is 804 MPa at room temperature. The values for yield and tensile strength at the testing temperatures for
the 30 and 50 mm plate were obtained from a cubic polynomial t of the yield and tensile strength data found in the European research project EcoPress [12]. For the 10 mm plate the difference in yield and tensile strength at room temperature
between the 10 mm plate and the polynomial t at room temperature was assumed to be constant.

Table 1
Chemical composition (wt.%) of the plates.
Plate thickness (mm)

Si

Mn

Cr

Ni

Mo

50
30
10

0.023
0.022
0.016

0.40
0.44
0.50

1.40
1.47
1.40

0.020
0.026
0.023

0.001
0.001
0.001

22.4
22.4
22.4

5.70
5.61
5.20

3.20
3.12
2.80

0.17
0.17
0.18

Fig. 2. Microstructure of the 10 and 50 mm plate: (a) 10 mm plate and (b) 50 mm plate. Ferrite is the dark phase and the rolling direction is horizontal with
the gure.

242

J. Pilhagen, R. Sandstrm / Engineering Fracture Mechanics 99 (2013) 239250

Polished and etched light optical microscope (LOM) photos of the microstructure (transversal plane) for the 10 and
50 mm plate are shown in Fig. 2. It is evident that the 10 mm plate has a more elongated and aligned austenite lamella structure than the 50 mm plate due to the more extensive hot-rolling. The austenite spacing and thickness were measured perpendicular to the rolling direction at 20 magnication in LOM, see Table 2. The measurements were in the transversal plane
according to Fig. 1. In this work, the rolling direction (L) is normal to the L-plane, the transverse direction (T) is normal to the
T-plane and the through-thickness direction (S) is normal to the S-plane. The 10 mm plate has the smallest distance between
the austenite lamellae (austenite spacing) followed by the 30 mm plate. The 50 mm plate has a homogenous microstructure
with respect to thickness of the austenite lamellae and their spacing.
2.2. Specimen selection
The specimen orientation for all tested specimens was TL according to Fig. 1 and the specimen dimensions were
50  50  400 mm3, 30  60  400 mm3 and 10  20  200 mm3. The rst number corresponds to the plate thickness. From
a 50 mm plate, 10 mm specimens of dimension 10  20  200 mm3 and 30 mm specimens of dimension 30  60  400 mm3
were cut out. Three types of 10 mm specimens were cut out according to their location in the thickness direction in the
50 mm plate, surface specimens, intermediate specimens and mid-thickness specimens. The methods used were water cutting or sawing. From the 30 mm plate, specimens of dimension 30  60  400 mm3 were produced. All surfaces were machined to comply with the fracture toughness standard ASTM E 1290-02 [13]. Electrical discharge machining was used
for the notch and clip-gauge slots.
2.3. Testing procedure
Standard SE(B) specimens were used in the fracture toughness measurements. Side-grooves were used on all fracture
toughness specimens and the crack length over the specimen width ratio was around 0.55 including fatigue precrack. The
fracture mechanic testing was done with a 100 kN hydraulic testing machine and a clip-gauge to measure the crack mouth
opening displacement (CMOD) of the specimen. The specimens were submerged during the testing in ethanol. Liquid nitrogen was used to cool the ethanol down to the testing temperature.
The fracture toughness testing was based on ASTM E 1290-02 for evaluating the crack-tip opening displacement (CTOD)
[13]. The CTOD can be evaluated from the sum of the elastic (Jel) and plastic (Jpl) component of the J-integral at the point of
instability. The point of instability can either be the unset of unstable crack extension prior to or following stable crack
growth or the attainment of a maximum force plateau [13]. The elastic component can be found from the relation between
the stress intensity factor K and the Jel:
Table 2
Mean austenite thickness and spacing.
Plane

Austenite thickness (lm)

Austenite spacing (lm)

10
30
50 Ca
50 Sb

T
T
T
T

5.8
9.6
11.5
11.6

5.5
6.9
9.2
9.8

C = mid-thickness specimen.
S = surface specimen.

400

50 mm
30 mm, B = 50 mm

350

10 mm, B = 50 mm
0

300

m [m]

a
b

Plate thickness (mm)

250
200
150
100
50
0
100

90

80

70

60

50

Temperature [ C]
Fig. 3. Result from the fracture toughness testing of the 50 mm plate. B0 is the plate thickness.

J. Pilhagen, R. Sandstrm / Engineering Fracture Mechanics 99 (2013) 239250

J el

K 2 1  t2
E

243

where t = 0.3 is the Poissons ratio for duplex stainless steel, 211 GPa is used for the elastic modulus, E. K is the elastic stress
intensity factor at the point of instability. The plastic component, Jpl, can be found from the point of instability as follows:

J pl

gpl Ap

BN b0

where Ap is plastic area under the force versus crack mouth opening displacement records, gpl = 3.7853.101(a/W) + 2.018(a/
W)2, BN is the net thickness and b0 is the remaining ligament. The crack-tip opening displacement can be found from Eq. (3):

J el J pl
mrY

where m is a constraint parameter which is a function of the crack depth and the strain-hardening exponent. rY is the effective yield strength at test temperature, dened as the mean of yield and tensile strength.
It has been reported in the literature that pop-ins are quite common during fracture toughness measurements in rolled
duplex stainless steels [14]. Pop-in is a sudden crack extension with accompanied crack arrest. For a pop-in to be regarded
as signicant and taken as a point of instability it must be severe enough. Two common criteria are the increased crack
length [13,14] or the force drop in combination with increased clip-gauge displacement [13]. In this work a pop-in is evaluated according to Section 9.1.2 in ASTM E 1290-02 [13].
3. Results
No unstable failure occurred during the testing and thus the failure criteria for the tested specimens were chosen to be
maximum force plateau or pop-in. For the rst case when the maximum force plateau was reached the crack extension was
measured by subsequent unloading. Post-fatigue cracking at room temperature was also conducted for measuring the crack
extension with LOM after breaking the specimen. Pop-ins were only observed for the 10 and 30 mm plate. The testing temperatures were between 94 C and 18 C.
The result for the 50 mm plate can be seen in Fig. 3. The 10 mm specimens have in principle the same measured CTOD as
the 50 mm specimens and shows that the 50 mm plate has homogenous fracture toughness as the specimens taken from
different depths of the plate exhibit similar toughness and scatter. The 30 mm specimens seem to have a bit higher CTOD.
The result shows that the CTOD dened from the maximum force plateau gives temperature independent fracture toughness
in the tested temperature interval.
The result from the 30 mm plate is shown in Fig. 4. All the specimens from the 30 mm plate exhibited multiple pop-ins
but only two specimens had signicant pop-ins as shown in Fig. 4. The 30 mm plate has higher CTOD than the 50 mm plate.
For the 10 mm plate multiple pop-ins occurred during the testing and if the maximum force plateau were used instead of
pop-in the CTOD would increase substantially, see Fig. 5. The fracture toughness seems to be temperature independent for
10 mm plate in the tested temperature interval.
The difference in pop-in behaviour between the 30 and 10 mm plate is that for the 10 mm plate the pop-in occurs shortly
after the start of the nonlinear behaviour while for the 30 mm plate the pop-in occurs more closely to the maximum force

400
350
300

[m]

250
200
150
100
50
0
100

Maximum force plateau

Popin

90

80

70

Temperature [ C]

60

50

Fig. 4. Result from the fracture toughness testing of the 30 mm plate. For the lled data points the evaluated maximum force plateau and pop-in are from
the same specimen.

244

J. Pilhagen, R. Sandstrm / Engineering Fracture Mechanics 99 (2013) 239250

400
Maximum force plateau
Popin

350
300

[m]

250
200
150
100
50
0
100 90

80

70

60

50

40

30

20

10

Temperature [C]
Fig. 5. Result from the fracture toughness testing of the 10 mm plate. For the lled data points the evaluated maximum force plateau and pop-in are from
the same specimen.

plateau, compare Fig. 4 with Fig. 5. According to the method used for pop-in evaluation, the pop-in for the 10 mm plate is
also more severe compared to the 30 mm plate.
The fracture surfaces of all specimens are dominated by the splits and extensive shear-lip deformation between the splits.
This is independent of plate thickness and specimen dimension. Their sizes vary from small ones only observable in SEM to
larger ones that are visible to the naked eye. These splits grow both in and perpendicular to the plane of the fatigue crack as
seen in Fig. 6, and can be several mm deep. The true length of these splits during testing is unknown because they grow durp
ing subsequent post-fatigue (30 < KI < 70 MPa m).
Examination of the fracture surfaces shows a distinct difference between the 10 and 30 mm plate and the corresponding
specimens from the 50 mm plate. The 10 and 30 mm plate have numerous and clearly dened splits, see Fig. 6a, compared to
the 10 and 30 mm specimens from the 50 mm plate which have fewer, shallower and less dened splits, see Fig. 6b. The
50 mm specimens have similar splits as the 30 mm plate, but not as deep and well dened.
The fracture surface between these splits is a combination of brittle fracture and microvoid coalescence (MVC). Close to
the splits the MVC fracture process is dominating. The brittle fractures consist of cleavage fractures in the plane of the fatigue
pre-crack, see Fig. 7, which are arrested locally. There are also areas mixed with the MVC that has curved facets and no direction of crack advancement which indicates a quasi-cleavage type of fracture [15]. The amount of growth between the splits
increases with increasing distance between the splits.

Fig. 6. SEM photos of the fracture surface for 30 mm specimens (25 magnication): (a) 30 mm plate and (b) 30 mm specimen from the 50 mm plate. Black
dots indicate the boundary between crack growth and post fatigue cracking.

J. Pilhagen, R. Sandstrm / Engineering Fracture Mechanics 99 (2013) 239250

245

Fig. 7. SEM photo showing cleavage fracture. 30 mm plate specimen tested at 55 C.

4. Discussion
4.1. Cause of splits
For the reported cases of splits in various materials in the literature, the splits were single or few and found in the midthickness region where the triaxial stress is the highest. This is not the case for the tested DSS where the splits were numerous and seemed to be evenly distributed over the thickness (at least for the splits clearly seen by the naked eye).
The likely explanation for the splits in 2205 is that when cleavage crack initiation occurs in the ferrite the cleavage propagation is locked along the austenit lamellae which cause delamination. For crack propagation in the S-plane (SL, ST and S45 specimens) the cleavage crack is free to propagate in the plane of the notch with little resistance from the austenite. This
causes the appearance of a weaker plane. The lower impact toughness of the S-plane for the 12 mm plate compared to the
50 mm plate in the work of Nilsson [6] might be explained by the difference in texture. The increased splitting in the 10 and
30 mm plate compared to the cut out specimens from the 50 mm plate is then explained by the more elongated and ner
microstructure found in the 10 and 30 mm plate, recall Fig. 2 and Table 2.

4.2. Loss of constraint

In an article by Embury et al. [16] the splits inuence on impact toughness was evaluated by the introduction of weaker
individual laminate subunits in the specimen (mild steel plates soldered together). It was concluded that the transition region was markedly shifted to a lower temperature and this effect increased with increasing laminate subunits. The increased
toughness was explained with the impact specimens behaving as the sum of the individual subunits which caused a relaxation of triaxial stress.
In the work by Chan [17] the same behaviour was observed for AlFeX alloys during fracture mechanics testing. It was
found that the specimens exhibited similar splits as the tested duplex stainless steels and this caused measured KIc values to
deviate from theoretical plane-strain stress state toward a plane-stress stress state. This mechanism was called thin sheet
toughening by Chan and it was later conrmed that for delaminated AlFeX alloys the KIc fracture toughness is controlled
by the thin sheet ligaments [18], thus rendering the measured toughness independent of specimen thickness.
With the same argument as Embury et al. and Chan, the local constraint in the tested specimens should decrease when
splits are initiated and propagating. The local constraint should then depend on the distance between two splits and on the
severity of respective split. The crack growth should be higher in a region with higher constraint (triaxial stress state) [19
21]. Fig. 8 shows the mid-thickness fracture surface of a 50 mm and 10 mm specimen and the difference in crack growth due
to the presence of splits is demonstrated. This relation between crack growth and splits has been observed for all tested
specimens.
Furthermore, if the suggested explanation for the splits (Section 4.1) is valid then it means that delamination is favoured
over cleavage crack propagation in the plane of the fatigue pre-crack. This causes the specimen to exhibit stable fracture
behaviour instead of a catastrophic failure, which increased the fracture toughness.
To gain further information on how the splits affect the fracture toughness, JR curves were calculated according to the
normalization data reduction technique described in chapter A15 in ASTM E 1820-06 [22]. This method is based on the key
curve approach and the principle of load separation and has been shown to give equivalent JR results with the elastic compliance method [23]. The normalization data reduction technique is not explained here and the reader is referred to reference ASTM E 1820-06 [22] for more information. This method was only successful on the 50 mm plate test data because the
multiple pop-ins that occurred for the 30 mm and the 10 mm plate caused a discontinuous JDa graph with large crack

246

J. Pilhagen, R. Sandstrm / Engineering Fracture Mechanics 99 (2013) 239250

Fig. 8. SEM photos over crack growth in 50 mm plate: (a) 50 mm specimen (14 magnication) and (b) 10 mm specimen (25 magnication). Black dots
indicate the boundary between crack growth and post fatigue cracking. Areas with lower crack growth are marked with an asterisk ().

extension jumps and an oscillation in J values. For JR measuring of steels which are prone to pop-ins, the elastic compliance
method may be less sensitive to pop-ins than the normalization method [24].
For all JR curves of the 50 mm plate (including 50, 30 and 10 mm specimens) the shape of the curve was the same. After
initial crack extension the J value increased linearly with increasing crack extension instead of having a power law relation,
see Fig. 9. The power law regression line is the tted line between the two exclusion lines according to J = C1DaC2. The tentative initiation toughness JQ is thus the intercept of the 0.2 mm offset line and the tted line. JQ becomes JIc if the thickness
and initial ligament is less than 25  JQ/rY and if the slope of the line in Fig. 9 at DaQ is less than rY [22]. rY is the mean value
of the yield and tensile strength at testing temperature. All specimens evaluated with the normalization method except for
two passed these conditions. These two 10 mm specimens exceeded the constraint on the initial ligament which is explained
by the present of larger than normal splits for the 10 mm specimens.

250

J, [kN/m]

200
150
100

JNorm.
Power law regression line
0.15 mm exclusion line
1.5 mm exclusion line
0.2 mm offset line

50
0

0.5

1.5

Crack extension, [mm]

Fig. 9. Resulting JR curve from the normalization data reduction technique for a 50 mm specimen at T = 55 C. The number of data points in the gure has
been reduced for clarity.

J. Pilhagen, R. Sandstrm / Engineering Fracture Mechanics 99 (2013) 239250

247

350

KJIc [MPam]

300
250
200
150
B = 50 mm
B = 30 mm
B = 10 mm

100
100

90

80

70

60

50

40

Temperature [ C]
Fig. 10. Initiation fracture toughness for 50 mm plate of 2205. The conversion from J-integral units to stress intensity factor units is done according to
p
KJIc = (JIc  E/(1  t2)).

The KJIc values for the 50 mm plate with different specimen dimensions are shown in Fig. 10. With respect to the number
of specimens and the scatter, the results indicate that the specimen thickness does not inuence the initiation fracture
toughness. The splits give almost constant fracture toughness independent of the specimen thickness. A result comparable
to Guo et al. [25].
The splits also seem to inuence the post-fatigue cracking of the specimens where some regions had large fatigue crack
extension where others were not affected at all by the fatigue cracking, see Fig. 6. This gives uncertainties on the extent of the
crack growth during the testing when optical measuring the crack growth with LOM. The highly irregular crack growth found
in the specimens, see Fig. 8, makes it difcult to dene the crack extension, especially if the split should be accounted for.
Using other methods like the potential drop or elastic compliance technique instead of optical measuring of the crack
extension may not solve the problem with the splits. It is plausible to assume that the free surfaces of the splits change both
the electrical resistance in the specimen and the elastic compliance of the specimen. This will give an erroneous measurement of the crack extension that occurred in the plane of the fatigue crack.
4.3. Pop-in
From the examination of fracture surfaces (broken up in liquid nitrogen) of duplex stainless steel specimens that exhibited pop-in, Wiesner [26] found no brittle crack propagation in the plane of the fatigue crack. The pop-in was instead attributed to the splits that had occurred during the testing. In Fig. 11 this lack of brittle crack propagation in the plane of the
fatigue crack is demonstrated. The depth of the split can be seen in Fig. 11b. For this particular specimen the fracture toughness test was stopped with subsequent fatigue cracking immediately after a pop-in for revealing the amount of crack growth
that had occurred during the pop-in. As seen in Fig. 11a no change in fracture surface has occurred between the two fatigue
surfaces except for crack-tip blunting and delamination (split). Pop-in that occur due to the formation of splits has been dened as type II pop-in [8].
The pop-in occurs due to the sudden change in compliance of the specimen when the splits are formed. A microstructure
with more elongated austenite lamellae is likely to cause a more severe pop-in due to the longer uninterrupted crack path
between the cleavage crack initiation and the crack front. This mechanism also explain why it is common to nd splits that
have grown backwards in to the fatigue pre-cracking region when a pop-in has occurred [8], see Fig. 11a. This phenomenon is
more common and more visible for the 10 mm plate compare to the 30 mm plate. For the 50 mm plate the pop-in is likely to
be too small to be observed. The difference in texture for the ferritic phase between the plates could also contribute to the
increased pop-in behaviour for the thinner plates [6].
The validity of type II pop-in in fracture toughness testing has been evaluated in the work by Wiesner and Pisarski [8]. It
was concluded that the fracture toughness value at this pop-in is not a true material property for the crack orientation subjected to testing nor for the through-thickness orientation and can thus be ignored. However, it is recommended that if the
component exhibit through-thickness stresses then the through-thickness fracture toughness (S-L/T/45 orientation in Fig. 1)
of the material should be evaluated [8,10].
4.4. The validity of ASTM E 1921 for the tested material
The ASTM E 1921-05 [14] standard is commonly used for evaluating the fracture toughness in the ductile to brittle transition temperature region for ferritic materials. The standard decides a reference temperature which characterizes the fracture toughness. The reference temperature, T0, is dened as the temperature where the median KJc for a 25 mm thickness

248

J. Pilhagen, R. Sandstrm / Engineering Fracture Mechanics 99 (2013) 239250

Fig. 11. SEM photo over pop-in fracture during testing of the 10 mm plate: (a) overview over the fracture surface showing (from right): pre-cracking
(fatigue), crack-tip blunting and post fatigue cracking and (b) depth of split in (a). Note that several splits are initiated during a pop-in event and in this
example seven similar splits along the crack front were found. L, T and S according to Fig. 1.

p
specimen (B1T) is 100 MPa m. This test method was originally developed for ferritic steels in the transition region where the
fracture toughness is primarily controlled by the statistical event of crack initiation from the cleavage of brittle second phase
particles like carbides and inclusions [14,27].
Specimens with different thicknesses can be size adjusted to the reference thickness (B1T) by Eq. (4).


K Jc1T K min K Jc0  K min 

B0
B1T

1=4
4

p
KJc(0) is the measured toughness, Kmin is the threshold toughness assumed to be 20 MPa m [14] and B0 is the specimen
thickness.
Evaluating the data for the 50 and 30 mm plate according to this method result in a T0 temperature of 135 C for the
50 mm plate and 143 C for the 30 mm plate. This is in agreement with other reported results for the 2205 base metal
[2,28]. However, the standard for determining the reference temperature T0 is based on the criteria of high constraint condition along the crack front which theoretically requires small scale yielding condition [14,29]. To assure that this criterion is
met at fracture for the master curve method, a maximum KJc capacity is dened:

K Jclimit

s
Eb0 rys

M1  t2

where b0 is the remaining ligament length of the specimen, rys is the yield strength at testing temperature and M is a nondimensional deformation limit which is set to 30 in the standard [14]. According to the KJc(limit) criteria the 30 and 50 mm
plate specimens have a sufcient constraint condition along the crack front.
However, if one considers a specimen where splits occur before or at stable crack growth and the specimen later fails by
unstable cleavage fracture, it is reasonable to assume that the effective thickness of the specimen is closer to the distance
between splits than the specimen thickness. The measured fracture toughness is likely to increase due to the reduced constraint in the material which results in higher plastic work done before a critical cleavage initiation occurs. One can therefore
question if the KJc(limit) is still valid for a specimen that delaminate during testing.

J. Pilhagen, R. Sandstrm / Engineering Fracture Mechanics 99 (2013) 239250

249

For the tested material the situation is however different and more complex. The splits seem to be cleavage cracks and
therefore a change in effective thickness occurs after cleavage initiation. All the specimens exhibited a stable fracture mechanism where increasing displacement was needed for increasing crack growth despite the presence of cleavage fracture in
the plane of the fatigue pre-crack. Reported fracture toughness results of a 30 mm plate of 2205 DSS (TL orientation) state
that no brittle failure (i.e. complete cleavage fracture) is observed down to 180 C [30]. This is interfering with the weakestlink theory and the assumption in the master curve method that the specimen failure is primarily cleavage initiation controlled [27,31]. Therefore the master curve analysis is not valid in this case because there is no direct link between cleavage
fracture initiation and specimen failure.
A similar situation exists for the use of the maximum force plateau as a failure criterion. The maximum force plateau occurs when the rate of strain hardening of the material is balanced by the rate of decrease of the remaining cross section [21].
However, this point is not directly associated with the crack extension in the specimen and likely to depend on specimen
geometry (size and crack depth) and displacement rate. The maximum force plateau has therefore been removed from
the CTOD test standard in the later revision of the test standard [32]. For specimens that only show stable crack growth
the CTOD at the end-of-test (deot) can be used for quality control and specications of acceptance [32].
The specimens were stable after reaching maximum force plateau under displacement control which indicates that the
specimens have a rising R curve, recall Fig. 9. Therefore, it is suggested that for assessing the fracture toughness at sub-zero
temperatures for the 2205 base metal the fracture resistance curve should be determined (JR or dR). For assessing the fracture toughness regarding brittle crack behaviour for the tested material at sub-zero temperatures it is suggested to evaluate
the split initiation toughness. The potential drop method might have the potential to detect the split initiation when it occurs
in the specimen.
5. Conclusions
Fracture toughness testing of a duplex stainless steel were performed on SE(B) specimens between 94 C to 18 C.
Specimens were taken from 10, 30 and 50 mm plate. From the 50 mm plate, 10 and 30 mm specimens were also cut out.
The purpose was to examine how the splits inuence the fracture toughness and how the microstructure affects the splitting
phenomenon in hot-rolled duplex stainless steels.
 Thinner plates have more extensive splitting than thicker plates likely due to the more aligned and elongated austenite
lamellae.
 The consequence of initiated and propagating splits is the creation of free surfaces which causes the specimen to behave
as two or more individual subunits. This causes loss of constraints along the crack front which increase the fracture
toughness. The loss of constraints is also visible by the local difference in crack growth in the specimen where the crack
growth decreases with decreasing distance between splits. From the normalization method, fracture initiation values
were obtained and the result shows that specimen thickness does not inuence the initiation fracture toughness.
 The pop-in phenomenon that can occur during fracture toughness testing of rolled duplex stainless steels is coupled to
the initiation of splits. This phenomenon is frequently observed for the 10 and 30 mm plates but not for the 50 mm plate.
The frequency and severity of the pop-ins seems to increase with decreasing plate thickness due to the renement of the
microstructure.
 The tested specimens exhibit a stable fracture process despite the presence of cleavage fracture likely due to the presence
of splits. For assessing the fracture toughness regarding brittle failure, the master curve method is suggested to not to be
valid for the 2205 base metal at the tested conditions. For evaluating the fracture toughness at sub-zero temperatures the
fracture resistance curve can be used.
 A proposed failure criterion for assessing brittle crack behaviour for these materials is the split initiation which may be
observed with the potential drop method.

Acknowledgements
The authors would like to express their gratitude to the VINN Excellence Center Hero-M and Outokumpu Stainless for
nancing this study. Outokumpu stainless is also gratefully acknowledged for delivering the material and for all help. Valuable support from Mikael Johansson and Jan Y. Jonsson at Outokumpu Avesta Research Centre is acknowledged.
References
[1] Dlouhy I, Chlup Z, Holzmann M. Crack length effect on fracture behaviour of duplex steels. In: Brown MW, de los Rios ER, Miller KJ, editors, Fracture
from defects, ECF 12 1998;2:72732.
[2] Sieurin H, Sandstrm R. Fracture toughness of a welded duplex stainless steel. Eng Fract Mech 2006;73:37790.
[3] Sieurin H, Sandstrm R, Westin ME. Fracture toughness of the lean duplex stainless steel LDX 2101. Metall Mater Trans A 2006;37A:297581.
[4] Dhooge A, Deleu E. Fracture toughness of duplex and super duplex stainless steels at low temperatures. Stainless steel world, September 1995.
[5] Sieurin H, Westin M E, Liljas M, Sandstrm R. Fracture toughness of welded commercial duplex stainless steel. Duplex 2007, 1820 June 2007, Grado,
Italy.

250

J. Pilhagen, R. Sandstrm / Engineering Fracture Mechanics 99 (2013) 239250

[6] Nilsson SA. Anisotropy in duplex stainless steel SS 2377. Swedish Institute for Metals Research, IM-2551, February 1992. Can be ordered at: <http://
www.swerea.se>.
[7] Erauzkin E, Irisarri MA. Effect of the testing temperature on the fracture toughness of a duplex stainless steel. Scripta Metall Mater 1991;25:17316.
[8] Wiesner SC, Pisarski GH. The signicance of pop-ins during initiation fracture toughness tests. 3R International 1996;35:63843.
[9] Chao CH. Mechanism of anisotropic lamellar fractures. Metall Trans A 1978;9A:50914.
[10] Pisarski GH, Hammond R, Watt K. Signicance of splits and pop-ins observed during fracture toughness testing on line pipe steel. In: Proceedings of
IPC2008, seventh international pipeline conference, Alberta, Canada, 2008. p. 47382.
[11] Yang Z, Huo C, Guo W. The charpy notch impact test of X70 pipeline steel with delamination cracks. Key Eng Mater 2005;297300:23916.
[12] Ericsson C, Sandstrm R, Sieurin H, Lagerqvist O, Eisele U, Schiedermaier J, et al., Material and welding data. Properties and test results, background
document 3.5: duplex stainless steel, EcoPress, European research 5th, framework; 2003.
[13] ASTM E 1290-02: Standard test method for crack-tip opening displacement (CTOD) fracture toughness measurement. vol. 03.01, 2007 ed.
[14] ASTM E 1921-05: Standard test method for determination of reference temperature, T0, for ferritic steels in the transition range. vol. 03.01.
[15] Engel L, Klingele H. An atlas of metal damage. London: Wolfe Publishing Ltd.; 1981.
[16] Embury DJ, Petch JN, Wraith EA, Wrigth SE. The fracture of mild steel laminates. Trans Metall Soc AIME 1967;239:1148.
[17] Chan SK. Evidence of a thin sheet toughening mechanism in AlFeX alloys. Metall Trans A 1989;20A:15564.
[18] Chan SK. Conrmation of a thin sheet toughening mechanism and anisotropic fracture in AlFeX alloys. Metall Trans A 1989;20A:233744.
[19] Marrow JT, Humphreys OA, Strangwood M. The crack initiation toughness for brittle fracture of super duplex stainless steel. Fatigue Fract Eng Mater
Struct 1997;20:100514.
[20] Kikuchi M, Ishihara T. Study on thickness effect of three-point-bend specimen. JSME Int J Ser A 2006;49:4117.
[21] Anderson LT. Fracture mechanics: fundamentals and applications. 2nd ed. CRC Press; 1995.
[22] ASTM E 1820-06: Standard test method for measurement of fracture toughness. vol. 03.01.
[23] Landes DJ, Zhou Z, Lee K, Herrera R. Normalization method for developing JR curves with the LMN function. J Test Eval 1991;19:30511.
[24] Zhu KX, Levis NB. Experimental determination of constraint dependent JR curves for X80 pipeline steel using normalization method. In: ASME
international mechanics engineering congress and exposition. 2006: p. 695703.
[25] Guo W, Dong H, Lu M, Zhao X. The coupled effects of thickness and delamination on cracking resistance of X70 pipeline steel. Int J Pres Ves Pip
2002;79:40312.
[26] Wiesner SC. Toughness requirements for duplex and super duplex stainless steels. In: Duplex stainless steels 975th world conference. 1997: p. 979
90.
[27] Wallin K, Laukkanen A. New developments of the Wallin, Saario, Trrnen cleavage fracture model. Eng Fract Mech 2008;75:336777.
[28] Wallin K. Quantifying Tstress controlled constraint by the master curve transition temperature T0. Eng Fract Mech 2001;68:30128.
[29] Ruggieri C, Doods Jr HR, Wallin K. Constraint effects on reference temperature, T0, for ferritic steels in the transition region. Eng Fract Mech
1998;60:1936.
[30] U. Eisele, Schiedermaier J. Limit state models derived, background document 7.2: ductile fracture, EcoPress, European research 5th, framework; 2003.
[31] Wallin K. The master curve method: a new concept for a brittle fracture. Int J Mater Product Technol 1999;14.
[32] ASTM E 1290-08: standard test method for crack-tip opening displacement (CTOD) fracture toughness measurement. vol. 03.01, 2010 ed.