Materials Science & Engineering A 651 (2016) 556–566

Contents lists available at ScienceDirect

Materials Science & Engineering A

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

Low Cycle Fatigue behavior of SMAW welded Alloy28 superaustenitic

stainless steel at room temperature
Y. Kchaou a,b,n, V. Pelosin a, G. Hénaff a, N. Haddar b, K. Elleuch b
a
Institut Pprime, Département Physique et Mécanique des Matériaux, UPR 3346 CNRS ISAE-ENSMA Université de Poitiers, Téléport 2, 1, avenue Clément Ader,
BP 40109, F – 86961 Futuroscope Chasseneuil Cedex, France
b
Laboratoire de Génie des Matériaux et Environnement (LGME), ENIS, BPW 1173, Sfax, Tunisia

art ic l e i nf o a b s t r a c t

Article history: This paper focused on the study of Low Cycle Fatigue of welded joints of superaustenitic (Alloy28)
Received 15 September 2015 stainless steels. Chemical composition and microstructure investigation of Base Metal (BM) and Weld
Received in revised form Metal (WM) were identified. The results showed that both of composition is fully austenitic with a
5 November 2015
dendritic microstructure in the WM. Low cycle fatigue tests at different strain levels were performed on
Accepted 6 November 2015
Base Metal (BM) and Welded Joint (WJ) specimens with a strain ratio Rε ¼  1. The results indicated that
Available online 10 November 2015
the fatigue life of welded joints is lower than the base metal. This is mainly due to the low ductility of the
Keywords: Welded Metal (WM) and the presence of welding defects. Simultaneously, Scanning Electron Microscope
Welding (SEM) observations of fractured specimens show that WJ have brittle behavior compared to BM with the
Low Cycle Fatigue
presence of several welding defects especially in the crack initiation site. An estimation of the crack
Austenitic
growth rate during LCF tests of BM and WJ was performed using distance between striations. The results
Microscopy
Fracture showed that the crack initiation stage is shorter in the case of WJ compared to BM because of the
Dendrite presence of welding defects in WJ specimens.
& 2015 Elsevier B.V. All rights reserved.

1. Introduction the fatigue behavior of electron beam welding joint of austenitic

stainless steel sheet using a tension–tension fatigue tests at room
The weldments are microstructurally and mechanically het- temperature. The authors found that failure occurs always on
erogeneous. Because of the presence of many possible sites of welded joint. It has been explained by the presence of a certain
failure due to the weldment process, the resulting structures are degree of stress concentration at weld root in the back of welding
subjected to fatigue phenomenon, which is manifested by initia- bead. Buchely et al. studied the effect of SMAW welding on fatigue
tion and propagation of cracks. Several studies have been con- behavior of 304 austenitic stainless steel [9]. They found that the
ducted on Low Cycle Fatigue of welded joints at room tempera- final failure occurred into welding deposits. It has been explained
ture, high temperature or under thermomechanical loading [1–5]. by the presence of dendritic microstructure which can influence
The studies generally focused on the cyclic behavior, the fatigue the growth and propagation of the crack. Zaletelj et al. [10] studied
life and damage mechanisms of the Base Metal (BM), Weld Metal the LCF of ferritic stainless steel welded joints. Authors found that
(WM) and Welded Joint (WJ). Madi et al. [6] investigated fatigue welded joint have a reduced fatigue life compared to base metal.
life of welded 316 L stainless steel under different strain levels. This result can be attributed to the difference of microstructure
These authors found that WM have a reduced fatigue life com- and the lower ductility of welded samples.
pared to BM specimens. The fatigue life of the 16MnR welded steel Concerning the superaustenitic stainless steel, until now sev-
used for pressure vessels construction has been studied by Cheng eral works were performed focusing on corrosion properties [11–
et al. [7]. The authors have shown that WM has a lower lifetime 13]. However, no studies concern their fatigue properties which
than the BM due to its lack of ductility and also to the presence of are crucial for further applications. For instance, the Alloy28 su-
residual stress in the structure created after the rapid cooling of peraustenitic stainless steel is largely used for pressure vessel.
the filler metal during the welding process. Zhao et al. [8] studied Therefore, the main purpose of the present work is to compare the
Low Cycle Fatigue behavior of this specific superautenitic steel
n
(Alloy28) then referring to LCF test carried on different materials
Corresponding author at: Institut Pprime, Département Physique et Mécanique
des Matériaux, UPR 3346 CNRS ISAE-ENSMA Université de Poitiers, Téléport 2, 1,
[1,5–7]. Firstly, the fatigue lives of Base Metal and Welded Joint
avenue Clément Ader, BP 40109, F – 86961 Futuroscope Chasseneuil Cedex, France. specimen at room temperature are investigated. Secondly, damage
E-mail address: [email protected] (Y. Kchaou). process is studied using Secondary Electron Microscope (SEM) of

http://dx.doi.org/10.1016/j.msea.2015.11.015
0921-5093/& 2015 Elsevier B.V. All rights reserved.
 Y. Kchaou et al. / Materials Science & Engineering A 651 (2016) 556–566 557

fractured specimens in order to identify the origin of the differ- microstructure of the BM consists of equiaxed grains with sizes
ence between BM and WJ. Finally, spacing between striations is ranging between 100 mm and 200 mm. Concerning the WM, the
determined in order to estimate fatigue propagation life of BM and microstructure is fully dendritic. For both alloys, the micro-
WJ specimen at different imposed strain level. structure is supposed to be fully austenitic as it does not appear to
be any delta ferrite or carbide precipitates.

2. Materials and methods 3.2. Low Cycle Fatigue properties of Alloy28 base metal and weld
joint
Hot rolled superaustenitic stainless steels were welded using
Shielded Metal Arc Welding process. The welding procedure was The cyclic stress–strain curves derived from experimental re-
made on base metal sheets of Sanicro28 with commercially sults on BM and WJ specimens are summarized in Figs. 5 and 6
available filler metal. The steel sheet was first prepared by respectively, taking into account duplicated tests. Fatigue life of
chamfering and grinding the edges in order to make an X form. duplicated tests will be treated later. These results were obtained
Details of the welding method can be found in [14]. For metallo- with total strain-controlled amplitudes of 0.3%, 0.4%, 0.6% and
graphic examinations, several specimens were prepared by 0.8%. A hardening peak is observed in BM curves at the 50th cycle
grinding using 120, 240, 320, 600, 800, 1000, 2000 and 4000 grits for 0.4% and 0.6% of strain levels, and at 30th and 60th cycle for
of SiC paper, followed by the final polishing with 3 mm and 1 mm 0.3% and 0.8% of total imposed strain level, respectively. Further-
alumina powders. Then, the samples were etched using an elec- more, the peak is more pronounced at higher strain amplitudes.
trolytic etching in an aqueous solution of 10 wt% oxalic acid at 5 V This initial hardening stage is followed by a softening and stabi-
for 20 s. The microstructure features were investigated using an lization until final rupture of the specimen. In addition, a decrease
optical microscope. Chemical composition of BM and WM was in the stress amplitude is observed at the end of each test which is
identified using a scanning electron microscope (SEM) equipped characteristic of the propagation of macroscopic cracks.
with energy dispersive spectroscopy (EDS) point analysis. Fatigue The cyclic stress response curves of WJ are divided into
tests until rupture were conducted at room temperature on BM 3 stages. The first stage corresponds with the initial hardening
and WJ specimens at different imposed total strain levels (0.3%, with a peak observed at the 10th cycle for 0.3% and 0.4% and at the
0.4%, 0.6% and 0.8%) with a strain rate equal to 10  3 s  1 and a 30th cycle for 0.6%, and 0.8% of total strain amplitude, followed by
strain ratio Rε ¼  1. Fatigue tests at 0.4% and 0.6% were duplicated a cyclic softening and finally the fracture of the specimen. It should
for BM and WJ specimens. Fatigue tests were performed following be noted that the stabilization phase is not present in the WJ
the standard ASTM: E606/E606M-12 Standard Test Method for specimens.
Strain-Controlled Fatigue Testing. The samples, as shown in Fig. 1, Table 2 shows the hardening and softening rates for BM and WJ
have a cylindrical geometry with a diameter of 4.3 mm and a at different imposed strain level. Both rates were calculated using
length of 56 mm. Two types of specimens were used in this in- Eqs. (1) and (2) [17]:
vestigation: BM specimens were machined in the rolling direction, σpeak − σ1stcycle
Hardening rate = × 100(%)
and WJ specimens which contain a weld nugget in its center as σ1stcycle (1)
shown in Fig. 2. The specimens were grinded by SiC grit paper at
different number in order to eliminate surface defects obtained σpeak − σ NR
2
after machining. Scanning electron microscope (SEM) images of Softening rate = × 100(%)
σ1stcycle (2)
the failed specimens after testing are also used in order to identify
the failure modes of the materials. where σpeak , σ1stcycle and σ NR are stress amplitude at the hard-
2
ening peak, at the first cycle and at half-life cycle, respectively.
3. Results and discussions As can be seen from Table 2 the hardening rate increases when
strain amplitude increases for BM and WM specimens. For the two
3.1. Chemical and microstructure investigation studied materials, the softening rate does not exhibit a marked
dependence on strain amplitude.
The chemical compositions of the Base Metal and the Weld In order to compare the cyclic stress–strain behavior between
BM and WJ, the cyclic stress–strain response and the stabilized
Metal were evaluated using EDS X-ray analysis and are given in
stress–strain hysteresis loops obtained at 0.4% and 0.8% of total
Table 1. They are compared to those determined for UNS N08028
imposed strain are plotted in Fig. 7(a) and (b) respectively. Ac-
and Sanicro28 studied by [15,16]. It can be seen that both com-
cording to cyclic hardening curves already presented [14], the
positions are nearly similar, which confirm the good compatibility
stress level in WJ material is higher than in BM and the difference
of the two materials for welding. Optical micrographs of the BM
is about 100 MPa. Moreover, the consolidation peak appears for a
and the WM are shown in Figs. 3 and 4 respectively. The
lower number of cycles in the WJ compared to the base metal. A
similar result was also found in the work of Rao et al. on the
welded 304 austenitic stainless steel [18]. These authors have
shown that the WM of 304 austenitic stainless steel contains
brittle particles and a very high dislocation density (around
8.109 cm  2). As a result, the presence of these defects has been
directly linked to the early hardening during the cyclic loading. In
our study, many precipitates have also been detected in the WM of
Alloy28 stainless steel [14]. This concordance confirms the crucial
role of the precipitates, generated by the weldment process, in the
hardening of the material. Previous study on 304 L austenitic
stainless steel has shown that pre-hardening shifts the cycle cor-
responding to the hardening peak to the left [19]. Author con-
Fig. 1. Specimen's geometry. firmed that pre-hardening increases dislocation density in the
558 Y. Kchaou et al. / Materials Science & Engineering A 651 (2016) 556–566

Fig. 2. Tested specimens, (a) Welded Joint (WJ) specimen configuration and position of the extensometer, (b) Base Metal (BM) specimen installed on the fatigue test
machine.

Table 1
Chemical composition (wt%) of Base Metal (BM) and Weld Metal (WM).

Fe Ni Cr Mo Si

UNS N08028 [15] 39.07 30.6 26.6 3.29 0.44

BM 36.40 31.95 28 3.41 0.22
Sanicro 28 [16] 40.74 29.88 25.84 3.06 0.48
WM 35.85 33.36 27 3.14 0.64

Fig. 4. Microstructure observation of Alloy28 Weld Metal (WM).

Fig. 3. Microstructure observation of Alloy28 Base Metal (BM).

stainless steel which changes the consolidation peak's cycle

comparing to initial material. Then, the change of the number of
cycle corresponding to the consolidation peak can be associated to
high dislocation density in the welded stainless steel caused by
welding procedure. Concerning the high softening rate of WJ, as
predicted by Rao et al. [18], it can be explained by the initial high
density of dislocations which would subsequently annihilate and
rearrange to a cellular structure during cyclic loading. Fig. 5. Cyclic stress evolution of Alloy28 BM specimens under different strain level.

3.3. Fatigue life analysis duplicated tests was calculated and it is close to 1 for the two
imposed strain level. This result leads us to consider that dupli-
Table 3 shows the fatigue life results of duplicated tests per- cated tests show a good reproducibility of fatigue tests.
formed at 0.4% and 0.6% of total imposed strain level in order to In order to compare the fatigue life of the BM and WJ speci-
study their reproducibility. The ratio between fatigue lives of mens, the curves of strain amplitude vs. number of cycles to failure
 Y. Kchaou et al. / Materials Science & Engineering A 651 (2016) 556–566 559

Table 3
Fatigue test results of duplicated tests.

Δεt/2 (%) NR1 (cycles) NR2 (cycles) NR1/NR2

test 1 test 2

BM 7 0.4 14,571 14,135 1.03

7 0.6 4837 3300 1.4

WJ 7 0.4 6135 6017 1.01

7 0.6 2158 1937 1.11

Fig. 6. Cyclic stress evolution of Alloy28 WJ specimens under different strain level.

Table 2
Hardening and softening rate of BM and WJ at different strain level.

Strain amplitude (%) Hardening rate (%) Softening rate (%)

BM WJ BM WJ

0.3 7 10 5 14.5
0.4 12 7.6 6.3 15.2
0.6 10.7 8.6 8.2 11.6
0.8 36 15.9 5.7 8
Fig. 8. Fatigue life curves of Alloy28 BM and WJ specimens.

are reported in the Fig. 8. It is important to mention that all tests This result could suggest that, at low plastic strain amplitude, WM
performed at 0.4% and 0.6% of total imposed strain amplitude were is more resistant than BM.
duplicated in order to check fatigue life reproducibility. It can be The Manson–Coffin and Basquin equation, Eq. (3), presents a
noted that the fatigue life of BM is superior to that of WJ for all relationship between experimental parameters: plastic strain εp,
investigated imposed strain levels and this result can be probably elastic strain εe, total strain level εt and number of cycles to failure,
attributed to the low ductility of WM. In previous studies [6,10] NR [22–24]
authors have also shown that the WM of 316 L stainless steel and a ∆εt ∆εe ∆εp σ f′
ferritic stainless steel has a lower fatigue resistance than BM = + = (2NR )b + εf′(2NR )c
2 2 2 E (3)
caused by the difference of microstructure and the low ductility of
the welded joint. where σ′f is the fatigue strength coefficient, calculated as the
The observations of the WJ specimen gage length using an value of the stress amplitude for 2NR ¼1; b is the fatigue strength
optical microscope are shown in Fig. 9. It can be seen that the exponent, evaluated as the slope of the line log ( ∆ εe ) against log (2
2
failure occurs in the weld metal for 0.4%, 0.6% and 0.8% of imposed NR); ε′f is the fatigue ductility coefficient, defined as the value of
strain level. These results are in good agreement with those ob- the plastic strain component for 2NR ¼1; c is the fatigue ductility
∆ εp
tained in the work of [18,20] on a welded 304 stainless steel [9,21] exponent, calculated as the slope of the line log ( ) against log (2
2
also found that the failure occurs in the WM of a welded austenitic NR); and E is the material elastic modulus.
304 and 316 L (N) stainless steel specimen respectively. Otherwise, ε–N curves are plotted in Fig. 10(a) and (b) for BM and WJ,
only at 0.3% of imposed strain level, failure occurs in the BM zone. respectively. The coefficients of Eq. (3) identified from these data

Fig. 7. Comparison between Alloy28 BM and WJ, (a) cyclic stress response, (b) stress–strain hysteresis loop at half-life.
560 Y. Kchaou et al. / Materials Science & Engineering A 651 (2016) 556–566

Fig. 9. Fractured WJ specimen's profile.

Fig. 10. Manson–Coffin curves, (a) BM, (b) WJ.

Table 4
Manson–Coffin and Basquin law's Parameters of Alloy28 BM and WJ.

ε′f c s′f (MPa) b

BM 0.134  0.43 1170  0.163

WJ 0.132  0.479 496  0.049

Fig. 12. Half-life stress amplitude at cycle vs. number of cycles to failure for BM and
WJ specimens.

previous studies on superaustenitic stainless steel 654SMO [25]

and on AISI 316N stainless steel [26]. The authors explained the
slope change by the activation of a single deformation slip system
while, at high amplitudes, multiple slip systems are activated.
Fig. 11. Comparison of half-life stress amplitude between BM and WJ specimens A comparison of the stress at mid-life of the BM and WJ for the
under different strain level. same imposed strain level is presented in Fig. 11. It can be seen
that this stress is higher in the case of WJ than BM specimens at
are reported in Table 4. The data treatment does not consider the the same imposed strain amplitude. Fig. 12 shows the half-life
test performed at 0.3% of imposed total strain level, because of the stress amplitude plotted against the number of cycles to failure for
change in the slope at this strain level. A change in the slope in the all imposed strain levels. It can be noted that the curves diverge
ε–N curve has also been noticed at low levels of plastic strain in when the imposed strain level is diminished. The evolution of
 Y. Kchaou et al. / Materials Science & Engineering A 651 (2016) 556–566 561

In order to account for differences in mechanical behavior and

isolate the effect of the microstructure on the fatigue life of the
studied materials, another analysis procedure for LCF has been
used which is based on plastic energy concept. The plastic strain
energy density Wp is defined as the inner area of the cyclic stress–
strain hysteresis loop. Generally, the plastic strain energy is cal-
culated at mid-life using Eq. (6) [27–30]:

W= ∫cycle σ : ϵ̇ pdt
(6)

In our case, the inner area of the cyclic stress–strain hysteresis

loop is determined using trapezoidal rule in order to have an ap-
proximate value of the plastic strain energy density.
A relationship between Wp and the number of cycles to failure
is presented in Eq. (7) [31]:

Fig. 13. Plastic strain energy density vs. number of cycles to failure for BM and WJ Wp = C⋅NR α (7)
specimens.
where C and α are material constants representing the material
energy absorption capacity and the fatigue exponent, respectively.
∆σ
¼ f(NR) laws were identified for BM in Eq. (4) and for WJ in Eq.
2 The plastic strain energy density is plotted against the number of
(5):
cycles to failure in Fig. 13 and then, values of material energy ab-
Δσ sorption capacity and fatigue exponent for Alloy28 BM and WJ are
= 949.5⋅NR−0.129
2 (4) deduced from the curves. They are found equal to 286.5 and
 0.523 for BM and to 289.4 and  0.57 for WJ, respectively. It can
Δσ be seen that, for the same plastic strain energy, fatigue life is
= 730.8⋅NR−0.079
2 (5) higher for BM specimens, which suggests that the difference in

Fig. 14. SEM observation of fractured BM Alloy28 gauge length; (a) Δεt/2 ¼ 0.4%, (b) zoom of (a), (c) Δεt/2 ¼0.8%, (d) zoom of (c).
562 Y. Kchaou et al. / Materials Science & Engineering A 651 (2016) 556–566

Fig. 15. Fractured surface observation of Alloy28 BM tested at 0.4% of total strain amplitude.

Fig. 16. Fractured surface observation of Alloy28 BM tested at 0.8% of total strain amplitude.

fatigue resistance cannot be fully interpreted by a mechanical at 0.8% of total imposed strain level. It can be observed that at
approach and reflects an actual effect of microstructure. 0.4%, slip bands are heterogeneously distributed in the grains.
They are parallel to the crack direction which is in their proximity.
3.4. Fracture analysis of Base Metal specimens It can be seen in Fig. 14(a) two stages of crack propagation: the
first stage corresponds with crack propagation along a slip plan of
Fig. 14 shows SEM observations of the specimen gage length maximum shear stress, while the second stage represents propa-
after rupture, Fig. 14(a) and (b) for test at 0.4% and Fig. 14(c) for test gation in a plane perpendicular to the applied stress direction.
 Y. Kchaou et al. / Materials Science & Engineering A 651 (2016) 556–566 563

Fig. 17. SEM observation of fractured Alloy28 WJ gauge length; (a) Δεt/2¼ 0.4%, (b) Zoom of (a), (c) Δεt/2 ¼ 0.8%, (d) zoom of (c).

Fig. 14(b) shows a magnification of the Fig. 14(a) where crack ra- imposed strain amplitude is characterized by the presence of
mifications can be seen on a grain boundary. At 0.8% of imposed many secondary cracks, suggesting a higher damage.
strain level (Fig. 14(c)), damage is more marked by the presence of
several parallel slip bands. These orientations are different from 3.5. Fracture analysis of Welded Joint specimens
one grain to another. Moreover, the observed cracks have more
ramifications compared to those observed at 0.4% of total imposed SEM observations of fractured WJ specimen gage length tested
strain level showing a higher damage. at 0.3%, 0.4% and 0.8% are shown in Fig. 17a) b) and c) respectively.
In order to determine fracture mechanisms, SEM observations It can be seen in Fig. 17a) that crack path is dependent on WM
of fractured surfaces were performed on specimens tested at 0.4% dendrite orientation. Fig. 17 b) shows that the crack direction is
and 0.8% of total imposed strain level. Fig. 15 shows the SEM parallel to dendrite axis. At 0.8% of total imposed strain level, a
fracture surface of BM specimen failed at 0.4% of total imposed secondary crack measuring more than 1.5 mm was observed in the
strain level. It can be seen that the fractured surface has a rough specimen's gage length. In addition, the crack path is very tortuous
appearance with the presence of fatigue striations. The crack in- in the case of WM specimens. Such a propagation behavior seems
itiation zone could be caused by extrusion–intrusion bands pro- to be highly dependent on microstructure orientation.
duced in the external surface of the specimen. In addition, rupture SEM observations were carried out on WJ specimens tested at
zone shows that many microcracks join the main crack path to 0.4% and 0.8% of total imposed strain level. From Fig. 18, it can be
cause final failure of the specimen. It appears that the observed seen that fracture surface of tested specimen at 0.4% of imposed
propagation mode is transgranular. strain level has many reliefs especially near the fracture zone.
Fig. 16 presents SEM fractography of BM specimen tested at Several crack initiation sites are also observed in these zones from
0.8% of total imposed strain level. It can be observed that fracture which many microcracks initiate and propagate until coalescence.
surface is less rough than for the specimen tested at 0.4% of total Furthermore and as already mentioned [14], the presence of
imposed strain level and it contains several fatigue striations. The welding defects, precipitates, brittle particles, has already been ob-
crack initiation is caused by a microcrack which length reaches served on the welded metal. As evident from the Fig. 19, SEM ob-
roughly 150 mm. In addition, the fracture surface at 0.8% of total servation of the specimen tested at 0.6% of imposed strain level
564 Y. Kchaou et al. / Materials Science & Engineering A 651 (2016) 556–566

Fig. 18. Fractured surface observation of Alloy28 WJ tested at 0.4% of total strain amplitude.

Fig. 20 shows fractured surface of WJ specimen tested at 0.8% of

total strain level. Crack propagates marking semi elliptical tip.
Once again, many crack initiation sites were observed on brittle
particles and inclusions considered as welding defects.

3.6. Crack growth rate in Low Cycle Fatigue tests

The fatigue life can be divided into two phases: the number of
cycles to crack initiation and the remaining life until the failure of the
specimen. In order to investigate the crack propagation rate in BM and
WJ specimens, spacing between striations (showed in Figs. 15 and 16)
vs. crack length was plotted using a log–log scale in Fig. 21(a) and
(b) respectively. In this study it has been supposed, as shown in per-
vious works [33,34], that one striation corresponds to one cycle. The
Crack growth rates can be fitted by a power law as in Eq. (8) [19]:
da
= B × aC
dN (8)

where a is the crack length, B and c are material constants. The

Fig. 19. Fractured surface center of Alloy28 WJ tested at 0.6% of total strain number of cycles of crack propagation Np is determined after in-
amplitude. tegration of Eq. (8) as following Eq. (9) [19]:
1 1 1
Np = × [ C −1 − C −1 ]
highlights the brittle character of these particles by the presence of B×(C − 1) a0 af (9)
many cracks crossing their center. These particles are found in dim-
ples and presents decohesion with the matrix. Moreover, it is well where a0 and af are equal to 30 mm and 3 mm respectively. In
known that brittle particles can lock dislocation movements and Table 5 are reported Np and Ni (number of cycles to crack initia-
increase the fragility of the material [32]. This fragility would en- tion) and Ni/NR ratio for BM and WJ at the most representative test,
hance the crack propagation rate and lead to a reduced fatigue life. 0.4% of imposed strain level. Number of cycle to crack initiation Ni
 Y. Kchaou et al. / Materials Science & Engineering A 651 (2016) 556–566 565

Fig. 20. Fractured surface observation of Alloy28 WJ tested at 0.8% of total strain amplitude.

Fig. 21. Spacing between striations measured for various imposed stain amplitude; (a) BM and (b) WJ.

Table 5 NR is the experimental number of cycles to rupture and Np

Estimation of the number of cycles to crack initiation and propagation in BM and corresponds to number of cycles of crack propagation determined
WJ specimens (Δεt/2 ¼0.4%).
using Eq. (9).
Δεt/2 0.4% At 0.4% of imposed strain level, Ni/NR ratio is higher in BM
specimen than in WJ. This result confirms that the crack initiation
Specimen BM WJ stage is lower in WJ than in BM. This can be attributed to the
NR exp 14,135 6135 presence of welding defects such as brittle particles and inclusions
Np 3607 2470
Ni ¼ NR exp  Np 10,527 3664
observed in fractured surface, which accelerate the fatigue life of
Ni/NR 65% 54% WJ specimens. The same conclusions have been provide from a
strain amplitude of 0.6%.

is determined using Eq. (10) [34]: 4. Conclusions

Ni = NR –NP (10) The Low Cycle Fatigue behavior of welded Alloy28 super-
austenitic stainless steel has been studied. Based on the
566 Y. Kchaou et al. / Materials Science & Engineering A 651 (2016) 556–566

experimental results, it can be concluded that the WJ presents a [12] M. BenSalah, R. Sabot, E. Triki, L. Dhouibi, Ph. Refait, M. Jeannin, Passivity of
hardening peak earlier than the BM at a given imposed strain Sanicro28 (UNS N-08028) stainless steel in polluted phosphoric acid at dif-
ferent temperatures studied by electrochemical impedance spectroscopy and
amplitude. The softening rate is higher in the case of WJ speci- Mott–Schottky analysis, Corros. Sci. 86 (2014) 61–70.
mens. These behaviors are attributed to the high initial disloca- [13] M. Ben Salah, R. Sabot, Ph. Refait, I. Liascukiene, C. Méthivier, J. Landoulsi,
tions density generated by the welding process in the WJ. In ad- L. Dhouibi, M. Jeannin, Passivation behaviour of stainless steel (UNS N-08028)
dition, the fatigue life is lower in the case of WJ than BM speci- in industrial or simplified phosphoric acid solutions at different temperatures,
Corros. Sci. 99 (2015) 320–332.
mens. This result is related to the reduced ductility of WM in re- [14] Y. Kchaou, N. Haddar, G. Hénaff, V. Pelosin, K. Elleuch., Microstructural, com-
lation with microstructural modifications, in particular the for- positional and mechanical investigation of Shielded Metal Arc Welding
mation of brittle particles provided from the shielded electrode (SMAW) welded superaustenitic UNS N08028 (Alloy28) stainless steel, Mater.
Des. 63 (2014) 278–285.
during welding process.
[15] X. Go´mez, J. Echeberria, Microstructure and mechanical properties of carbon
The failure mode is transgranular in the BM and WJ specimens steel A210/superalloy Sanicro 28 bimetallic tubes, Mater. Sci. Eng. A 348
with the presence of fatigue striations. The observations of spe- (2003) 180/191.
cimen gage length at failure suggest that crack initiation is mainly [16] A. Mirzaei, A. Zarei-Hanzaki, N. Haghdadi, A. Marandi, Constitutive description
of high temperature flow behavior of Sanicro-28 super-austenitic stainless
controlled by the formation of persistent slip bands. The crack steel, Mater. Sci. Eng. A 589 (2014) 76–82.
path is tortuous and depends on the orientation of dendritic grains [17] D. Ye, S. Matsuoka, N. Nagashima, N. Suzuki., The low-cycle fatigue, de-
in WJ specimens. The spacing between striation analysis shows formation and final fracture behaviour of an austenitic stainless steel, Mater.
that crack initiation life is shorter in WJ specimens which could be Sci. Eng. 415 (2006) 104–117.
[18] K. Bhanu Sankara Rao, R. Sandhya, S.L. Mannan, Creep-fatigue interaction
caused by the presence of welding defects. behaviour of type 308 stainless steel weld metal and type 304 stainless steel
Finally, it can be deduced that the difference of the LCF beha- base metal, Int. J. Fatigue 1 (3) (1993) 221–229.
vior between BM and WJ is mainly due to the welding process. As [19] Nicolas Huin, Effets d'environnement sur la fatigue de l'acier inoxydable 304L
en milieu primaire REP sous chargement cyclique, Ecole Nationale Supérieure
for many austenitic steels this process affects drastically the mi-
de Mécanique et d’Aérotechnique de Poitiers, 2013.
crostructure of the WJ by introducing inclusions and welding de- [20] C.R. Brinkman, G.E. Korth, J.M. Beeston, Comparison of the strain controlled
fects that consequently weakens the welded structure that must low cycle fatigue behaviour of type 304/308 weld and base material, in:
be taken into account for further applications. Proceedings of the Int. Conf. on Creep and Fatigue in Elevated Temperature
Applications (9th Ed.), Paper C218/73, Philadelphia, PA, September 1973 , In-
stitution of Mechanical Engineers, April 1974, p. 218.
[21] Sayan Kalyan Chandra, Vani Shankar, K. Mariappan, R. Sandhya, P.
References C. Chakraborty, Effect of strain rate on the low cycle fatigue behavior of 316 L
(N) stainless steel weld joints, Procedia Eng. 55 (2013) 176–180.
[22] L.F. Coffin, A study of the effect of cyclic thermal stresses on a ductile metal,
[1] A. Benoit, L. Rémy, A. Köster, H. Maitournam, F. Oger, Experimental in- Trans. ASME 76 (1954) 931–950.
vestigation of the behavior and the low cycle fatigue life of a welded structure, [23] O.H. Basquin, The exponential law of endurance tests, Am. Soc. Test. Mater.
Mater. Sci. Eng.: A 595 (2014) 64–76. Proc. 10 (1910) 625–630.
[2] C.W. Chan, H.C. Man, F.T. Cheng, Fatigue behavior of laser-welded NiTi wires in [24] Mauro Ricotta, Simple expressions to estimate the Manson–Coffin curves of
small-strain cyclic bending, Mater. Sci. Eng.: A 559 (2013) 407–415. ductile cast irons, Int. J. Fatigue 78 (2015) 38–45.
[3] M.-C. Robin, D. Delagnes, R. Logé, P.-O. Bouchard, S. Da Costa, M. Monteagudo- [25] S. Heino, B. Karlsson, Cyclic deformation and fatigue behaviour of 7Mo–0.5 N
Galindo, V. Velay, Thermo-mechanical fatigue behaviour of welded tubular
superaustenitic stainless steel stress-strain relations and fatigue life, Acta
parts made of ferritic stainless steel, Int. J. Fatigue 54 (2013) 84–98.
Mater. 49 (2) (2001) 339–351.
[4] D.J. Lee, J.C. Byun, J.H. Sunga, H.W. Lee, The dependence of crack properties on
[26] J. Polak, K. Obrtlik, M. Hajek, Cyclic plasticity in type 316 L austenitic stainless
the Cr/Ni equivalent ratio in AISI 304L austenitic stainless steel weld metals,
steel, Fatigue Fract. Eng. Mater. Struct. 17 (7) (1994) 773–782.
Mater. Sci. Eng. A 513–514 (2009) 154–159.
[27] E. Charkaluk, A. Constantinescu, Dissipation and fatigue damage, Mater. Test.
[5] Seon-Jin Kim, Pil-Ho Choi, Rando Tungga Dewa, Woo-Gon Kim, Min-
(46) (2004) 524–530.
Hwan Kim, Low cycle fatigue properties of Alloy 617 base metal and weld joint
[28] M.D. Callaghan, S.R. Humphries, M. Law, M. Ho, P. Bendeich, H. Li, W.Y. Yeung,
at room temperature, Procedia Mater. Sci. 3 (2014) 2201–2206.
Energy-based approach for the evaluation of low cycle fatigue behaviour of
[6] Yazid Madi, Philippe Matheron, Naman Recho, Philippe Mongabure, low cycle
fatigue of welded joints: new experimental approach, Nucl. Eng. Des. 228 2.25Cr–1Mo steel at elevated temperature, Mater. Sci. Eng. A 527 (2010)
(Issues 1–3) (2004) 161–177. 5619–5623.
[7] Guangxu Cheng, Z.B. Kuang, Z.W. Lou Hua Li, Experimental investigation of [29] D. Lefebvre, F. Ellyin, Cyclic response and inelastic strain energy in low cycle
fatigue behaviour for welded joint with mechanical heterogeneity, J. Press. fatigue, Int. J. Fatigue 6 (1984).
Vessels Pip. 67 (1996) 229–242. [30] Y.R. Luo, C.X. Huang, Y. Guo, Q.Y. Wang, Energy-based prediction of low cycle
[8] Xiaohui Zhao, Yu Liu, Research on fatigue behavior of electron beam welding fatigue life of high-strength structural steel, J. Iron Steel Res. Int. 19 (2012)
joint of 06Cr19Ni10 austenitic stainless steel sheet, Mater. Des. 57 (2014) 47–53.
494–502. [31] Balazs Fekete, New energy-based low cycle fatigue model for reactor steels,
[9] M.F. Buchely, H.A. Colorado, H.E. Jaramillo, Effect of SMAW manufacturing Mater. Des. 79 (2015) 42–52.
process in high-cycle fatigue of AISI 304 base metal using AISI 308 L filler [32] J. Friedel, Dislocations, Pergamon Press, Oxford, 1964.
metal, J. Manuf. Process. (2015), Technical Paper, 10.1016/j.jmapro.2015.08.005. [33] W.C. Connors, Fatigue striation spacing analysis, Mater. Charact. 33 (3) (1994)
[10] H. Zaletelj, G. Fajdiga, M. Nagode, Low cycle fatigue of welded joints with aging 245–253.
influence, Mater. Sci. Eng.: A 564 (2013) 478–484. [34] Laurent De Baglion de La Dufferie, Comportement et endommagement en
[11] C. Pettersson, J. Pettersson, H. Asteman, J.-E. Svensson, L.-G. Johansson, KCl- fatigue oligocyclique d’un acier inoxydable austénitique 304 L en fonction de
induced high temperature corrosion of the austenitic Fe–Cr–Ni alloys 304 L l’environnement (vide, air, eau primaire REP) à 300 °C, Ecole Nationale Su-
and Sanicro 28 at 600 °C, Corros. Sci. 48 (Issue 6) (2006) 1368–1378. périeure de Mécanique et d’Aérotechnique de Poitiers, 2011.