Sunteți pe pagina 1din 9

International Journal of Fatigue 33 (2011) 683691

Contents lists available at ScienceDirect

International Journal of Fatigue


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

Thermomechanical and isothermal fatigue behavior of 347 and 316L austenitic


stainless tube and pipe steels
Mageshwaran Ramesh a, Hans J. Leber a,, Koenraad G.F. Janssens a, Markus Diener b, Ralph Spolenak b
a
b

Paul Scherrer Institute, Laboratory for Nuclear Materials, CH-5232 Villigen-PSI, Switzerland
Laboratory for Nanometallurgy, Department of Materials, ETH Zurich, Wolfgang-Pauli-Strasse 10, CH-8093 Zurich, Switzerland

a r t i c l e

i n f o

Article history:
Received 9 June 2010
Received in revised form 25 October 2010
Accepted 9 November 2010
Available online 13 December 2010
Keywords:
Austenitic stainless steel
Thermo-mechanical fatigue
Lifetime prediction
Cyclic plasticity
Cyclic deformation

a b s t r a c t
Many components of nuclear power plant piping systems are made of austenitic stainless steels. These
structures undergo degradation by thermomechanical loading caused by thermal transients and stratications. In scientic literature, most of the studies deal with this problem under isothermal fatigue conditions, which is different from the typical service conditions. In addition, less attention has been paid to
thermo-mechanical fatigue (TMF). This work aims to understand and compare the cyclic deformation
behavior during TMF and isothermal fatigue (IF) testing in air for two of the most commonly used grades
of austenitic stainless tube and pipe (TP) steel, the non-stabilized TP 316L and the niobium-stabilized TP
347, under light water reactor relevant temperature conditions. Three types of tests, i.e. in-phase, outof-phase TMF in the 100340 C temperature range, and IF tests at the maximum temperature Tmax of
TMF were performed. All the tests were carried out under total strain control, for two different mechanical strain amplitudes (Demech/2 = 0.3% and 0.5%). Results revealed that irrespective of the tested strain
amplitude and material, in-phase TMF showed a higher lifetime over out-of-phase TMF and IF at Tmax.
A crossover in fatigue lifetime was observed in TP 316L between out-of-phase TMF and IF test at Tmax with
decreasing strain amplitude. In TP 347, out-of-phase TMF and IF test at Tmax showed similar fatigue lifetimes for both strain amplitudes. Finally, three lifetime prediction models were selected and compared to
evaluate the possibility of estimating TMF lifetime by using IF data. Satisfactory predictions were
obtained from all the prediction approaches.
2010 Elsevier Ltd. All rights reserved.

1. Introduction
Austenitic stainless steels (SS) have a unique advantage of good
mechanical properties and excellent corrosion resistance in a wide
temperature range, which makes this type of steel a favorable candidate for structural components in the nuclear industry. Most of
the piping systems of the light water reactor are made of stabilized
(usually Nb or Ti) and non-stabilized grades of austenitic SS. Light
water reactor operation conditions can induce local plastic deformation in the coolant pipes, owing to thermomechanical loading
(viz thermal transients, instabilities, turbulence, thermal stratication, etc.) [13]. Therefore, pipes in service undergo a complex evolution of damage which can hardly be described by isothermal
conditions. In the past, engineering fatigue curves of these materials have been established from simplied IF conditions. It is also
well known that the mechanical stress strain response of a metallic
material is temperature dependent. With the development and
standardization of TMF testing procedure in the last two decades,

Corresponding author.
E-mail address: hans.leber@psi.ch (H.J. Leber).
0142-1123/$ - see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijfatigue.2010.11.005

the non-uniform temperature distribution caused by physical


boundary conditions in reactor coolant pipe systems can be probed
more adequately with TMF tests than with IF tests [46].
Two basic types of TMF tests are conducted most often in the
laboratory to assess lifetime: namely in-phase and out-of-phase
[4,5,7]. An in-phase TMF cycle is where the temperature and applied mechanical strain of the cycle reach their maximum and minimum values at the same time. The out-of-phase TMF cycle is
where the maximum applied mechanical strain and lowest temperature occur at the same time. During a TMF loading, the primary
damage mechanisms that can play a role in metals are fatigue,
environmental and creep damage. These damage mechanisms
may act alone or in combination with each other, based on operating conditions such as strain range levels, strain rate or temperature intervals [8]. In austenitic SS, the activated damage
mechanism can alter the cracking behavior from intergranular for
creep dominant regimes to that of transgranular in fatigue conditions. In the last two decades, numerous research studies have
been performed focusing mainly on non-stabilized grades (304
and 316L) subjected to TMF loading under varying strain amplitudes, environment and temperature conditions [917]. According
to Kuwabara and Nitta [12], TMF behavior can vary in different

684

M. Ramesh et al. / International Journal of Fatigue 33 (2011) 683691

Nomenclature

a
Demech
Demech/2
Dep
Detherm
Detot

rmax
ANL
DSA
E
EBSD
IF

coefcient of thermal expansion


mechanical strain range
(total) mechanical strain amplitude
plastic strain range
thermal strain range
total strain range
average maximum stress
Argonne National Laboratory
dynamic strain ageing
elastic modulus
electron backscatter diffraction
isothermal fatigue

alloys and can be classied into one of the following four groups
depending on the loading mode: (a) in-phase TMF life is shorter
than out-of-phase TMF life at lower strain ranges; (b) out-of-phase
TMF life is shorter than in-phase TMF life at lower strain ranges; (c)
in-phase TMF and out-of-phase TMF lives are essentially the same;
(d) in-phase TMF life is shorter at higher strain ranges, but is the
same as out-of-phase TMF life at lower strain ranges. Even among
the non-stabilized grades 304L and 316L of austenitic SS, TMF
behavior varies signicantly (Table 1) [1012,1419].
On investigating the non-stabilized variant of 304 subjected to
TMF cycling between 427 C and 593 C, Majumdar [10] found
that in-phase TMF was more damaging than IF at Tmax and more
damaging than out-of-phase TMF for intermediate strain ranges
(Detot = 0.7%). Kuwabara and Nitta [12] performed TMF tests on
the same steel grade for a slightly wider temperature range 200
600 C, and reported considerable intergranular cracking. He found
that the isothermal tests at 600 C have the lowest fatigue lifetime
and also, the in-phase TMF lives were shorter than the of outof-phase TMF lives at higher strain ranges (Detot = 1.5%). Fujino
and co-workers [19] later conrmed the trends found by Kuwabara
by testing the same grade in air for a much broader temperature
interval of 200750 C. In another non-stabilized type 316L steel,
Zamrik and his co-workers [15] investigated the TMF temperature
range 399621 C and observed that in-phase TMF has the lowest
life time followed by IF at Tmax and out-of-phase TMF, for the
mechanical strain ranges greater than 0.7%. He also observed a
crossover between out-of-phase TMF and IF at Tmax at Demech = 0.6%.
A crossover is a phenomenon which describes the effect of varying
strain amplitudes on the lifetime trends for different loading conditions. In other words, it is the point on mechanical strain
amplitude where two or more lifetime trends interlace each other.

IPF
L, C, e0f
LWR
Nf5
SEM
SS
TEM
Tmax
TMF
Tmin
TP

inverse pole gure


material constants
light water reactor
fatigue lifetime
scanning electron microscopy
stainless steels
transmission electron microscopy
maximum temp. of TMF
thermo-mechanical fatigue
minimum temp. of TMF
tube and pipe material

However, at lower mechanical strain range levels (Demech < 0.4%) it


was observed that out-of-phase TMF is more damaging than inphase TMF and IF at Tmax [15]. Zamrik attributed these crossovers
to surface oxidation damage occurring over the long exposure period at extreme temperatures and varying yield stress during TMF
and IF loading. This crossover mechanism was also observed in
other classes of materials like carbon steels and super alloys
[2022]. Shi et al. [14] examined the same grade for a high strain
range Demech = 1.6% for two different temperature ranges, namely
250500 C associated with fatigue damage and 250650 C also
to include creep interactions. During 250500 C TMF tests, the
in-phase TMF showed similar life time as out-of-phase TMF, but
from the tests at another temperature range 250650 C, in-phase
TMF showed a lifetime four times longer than that of out-of-phase
TMF. This lower lifetime in out-of-phase TMF was attributed to the
development of tensile mean stress and greater hysteresis energy
for out-of-phase TMF tests. More recent investigations by Nagesha
et al. [11] in the temperature range of 300650 C and for intermediate amplitudes 0.25% < Demech/2 < 0.4% showed that, for these
conditions, creep and oxidation play dominant roles as damage
mechanisms. With decreasing strain amplitude, a crossover was
observed with higher out-of-phase TMF lifetime and in-phase
TMF becomes more damaging cycle. It was also observed that the
difference in lifetime between in-phase and out-of-phase TMF
diminishes with increasing peak temperature and that the outof-phase TMF lifetime increases by a factor of six when tested
under vacuum. The effect of environment on the TMF lifetime
was rst studied by Zauter et al. [17]. They found that for the same
material under vacuum, the maximum cycling temperature of the
TMF governs the damage mechanics. Most recently, Christ [9]
made a comprehensive study of the inuence of environment

Table 1
Summary of earlier works on comparison of TMF and IF behavior of austenitic stainless steel.
Author (year)

Grade

Strain range, total, Detot; mechanical,


Demech; plastic, Dep

Temperature interval (C)

OP and IP-TMF vs. IF at Tmax

Majumdar et al. (1987)


Kuwabara and Nitta (1979)
Fujino et al. (1979)
Zauter et al. (1994)a
Zamrik et al. (1996)

304L
304L
304L
304L
316L

Shi et al. (1996)

316L

Detot = 0.7%
0.5% < Detot < 1.5%
0.8% < Detot < 1.5%
Dep = 1%
0.35% < Demech < 0.6%
Demech  0.6%
Demech > 0.7%
Detot = 1.6%

Ellison et al. (1994)


Nagesha et al. (2009)

316L
316L

427593
200600
200750
250500
399650
399621
399621
250650
250500
400625
300650
300650

OP > IF > IP
OP > IP > IF
OP > IP > IF
OP > IP
IF > IP > OP
IF > OP > IP
OP > IF > IP
IP > OP
IP > OP
OP > IF > IP
OP > IP
OP > IP > IF

1% < Detot < 1.5%


Demech = 0.5%
Demech = 0.8%

OP: out-of-phase, IP: in-phase, IF: isothermal fatigue at Tmax.


a
Under vacuum.

685

M. Ramesh et al. / International Journal of Fatigue 33 (2011) 683691

fatigue lifetime is compared with standard IF tests at Tmax. The test


results were further analyzed with focus on understanding the differences if any, of their fatigue hardening and softening and cyclic
stress strain behavior. Later, the results from above TMF and IF
investigations were further utilized to evaluate existing conventional energy based life prediction models.

during TMF testing across a wide range of materials including 316L,


and conrmed the earlier observations by Zauter et al. They also
found that under vacuum conditions the crack initiation sites
shifted from the surface to the interior leading to similar lives in
in-phase and out-of-phase TMF tests. There have been numerous
attempts to develop energy based TMF life prediction methods
based on IF life approaches such as strain range partitioning,
CofnManson, SmithWatsonTopper damage parameter, the
Ostergren model and the ductility exhaustion model [8,2325]. It
was understood with such prediction methods that TMF life was
less than IF life in most of the cases in literature on a logarithmic
scale.
The main reasons that necessitate the current work are rstly,
most of the available TMF investigations on austenitic SS in literature were performed for non-stabilized grades (304 and 316L) because of their wider use in (nuclear) power plant designs. Secondly,
the available TMF investigations of those non-stabilized variants
are from high temperature tests (i.e. Tmax > 400 C), primarily
aimed to understand their creep-fatigue-environment interactions.
Thirdly, there is not much work available till date to compare the IF
and TMF behavior of austenitic SS at intermediate strain amplitudes at lower temperatures with only the fatigue mechanism
being activated (since below 400 C austenitic SS exhibits no creep
damage and better oxidation resistance). Fourth, until now no TMF
studies are available from the stabilized grades of austenitic SS
(347 or 321), which are also employed in service at several light
water reactor systems around the world. Finally, in these type of
reactor systems the typical service temperature range of 100
340 C lies partially within the dynamic strain ageing (DSA) temperature range of 250550 C for austenitic SS [26]. Hence, it will
provide us with a unique opportunity to observe the effect of elements in solid solution on the fatigue lifetime during TMF tests.
Lack of earlier investigations for these conditions were mainly
attributed to the fact that, regardless of advancement in temperature control systems in recent times, performing TMF on the LWR
relevant temperature intervals (100340 C) is quite challenging.
The recommended tolerances for TMF testing with the existing
standards are quite tough, as they were developed primarily from
high temperature tests [57]. Similarly, stabilized grades (347 and
321) were less widely used as compared to the older non-stabilized
variants (316L and 304) resulting in fewer number of IF investigations [27].
This article reports the results from TMF tests performed in the
temperature interval specied above, for two most widely used
austenitic tube and pipe materials (TP 316L and TP 347). Their

2. Experimental
2.1. Material and specimens
The investigated materials were portions of commercial niobium-stabilized TP 347 and non-stabilized TP 316L austenitic SS
steel pipes with outer diameter of 219 mm and 18 mm and
23 mm wall thickness, respectively. The seamless pipes were manufactured and processed according to the requirements of the corresponding codes for pressure vessel applications. Their respective
major chemical compositions were determined by using Inductive
Coupled Plasma Optical Emission Spectrometry (ICP-OES) and by
infrared adsorption method after combustion (C, S, N) (Table 2).
The processing sequences of the seamless pipe material consisted
of hot working, solution annealing, water quenching to room temperature, pickling and grinding. The TMF and IF experiments were
both performed on tubular specimens as recommended by the Validated Code-of-Practice for Strain-Controlled Thermo-Mechanical
Fatigue Testing (EUR 22281 EN) [7]. The specimens have an outer
diameter of 8 mm and an inner diameter of 5.5 mm (Fig. 1) [7].
They were machined out of the pipe wall according to the ASTM
E 2368-04 [4], the specimen axis being parallel to the pipe axis.
As recommended by the same ASTM E 2368-04, the bore of the
tubular specimens were honed. The resulting wall thickness of
1.25 mm guaranteed a faster regular temperature adaptation
across the specimen wall within the prescribed temperature variation. All the TMF and IF test specimens were solution annealed in
vacuum at 1050 C for about 1 h after fabrication, furnace cooled
to 950 C and quenched in argon to room temperature (85 C/
min). The machined and solution annealed samples were then
electropolished in a solution containing 10% perchloric acid and
90% acetic acid at 40 V and 10 C before testing in order to attain
a scratch free smooth surface. The starting microstructure of the alloy TP 316L is a typical precipitate free austenitic matrix with the
average grain size of 35 lm. In contrast the TP 347 grade consists
of precipitate strengthened austenitic matrix with an average grain
size of 20 lm measured with EBSD shown in normal direction

Table 2
Major chemical composition of the investigated pipe materials (values in mass%).
Grade

Si

Mn

Cr

Mo

Ni

Nb

TP 347
TP 316L

0.058
0.021

0.20
0.26

1.72
1.69

0.025
0.033

0.006
0.003

17.4
17.5

0.44
2.15

10.4
11.14

0.571
0.0601

0.045
0.012

Fig. 1. Geometry of the TMF and IF tubular specimen (dimensions in mm).

686

M. Ramesh et al. / International Journal of Fatigue 33 (2011) 683691

Fig. 2. The starting microstructure of TP 347. (a) EBSD mapping shown in normal direction inverse pole gure (IPF). (b) The starting texture of the material. (c) TEM image
[28].

inverse pole gure (IPF) as illustrated in Fig. 2a). The black spots in
the map indicate the non-indexed regions from coarser primary
carbides aligned parallel to pipe axis. Fig. 2b) shows the starting
texture of the material measured from the electropolished sample
(area: 1500  1000 lm2) illustrating ber texture about (0 0 1)
axis. Fig. 2c) represents a TEM image illustrating dense homogeneous distribution of NbX precipitates along {1 1 1} planes contributing to harder grain volume [28]. More details of this unique TP
347 microstructure can be found in our earlier work [28].
2.2. Fatigue testing
Fatigue tests were conducted on a low cycle fatigue testing system Instron 8862 100 kN with an axial servo-electric load train and
compact digital control electronics. The tubular fatigue specimen
was heated in an induction coil supplied by a high frequency generator with a maximum power of 5 kW at a frequency between 30
and 150 kHz. Both the heating coil and the crosshead clamps were
closed loop water cooled. The temperature of water was controlled
by a cooling unit. A digital universal processor controlled the balance of induction heating and forced air cooling so that the ramps
corresponded to the desired experimental parameters. The measurement and control of temperature was achieved with ribbon
S-type (Pt/PtRh) thermocouple wrapped around the sample at
the middle of the gauge length as recommended by Code-of-Practice for Strain-Controlled Thermo-Mechanical Fatigue Testing [7].
Further detailed tolerance limits achieved during testing can be obtained in our earlier work [28].
The fatigue tests were performed at a constant frequency of
0.03 Hz and at two different mechanical strain amplitudes each
generating constant mechanical strain rate for IF and TMF tests
(demech/dt = 3.6  104 s1 for Demech/2 = 0.3% and 6.06  104 s1
for 0.5%). It is quintessential to have a similar strain rate between
TMF and IF tests, as we are interested in understanding the timedependent damage mechanism. Since the fatigue experiments usually involve a scatter, three samples were tested for each of the
above described test conditions. A triangle wave form was used
for both thermal and mechanical cycling. During TMF tests, specimens were thermally cycled between 100 C and 340 C under zero
mechanical load in the course of which they were allowed for free
thermal expansion and contraction. With the establishment of
homogenous temperature cycling across the gauge length, the

thermal strain corresponding to expansion and contraction were


captured with an extensometer at zero force in order to compensate the thermal strain during the test. For this purpose the
mechanical strain was superimposed on captured thermal strain
proles. The following equation expresses the total applied strain.

etot etherm T emech t


Demech Detot  Detherm Detot  aTT max  T min

where Detot, Demech, Detherm are total, mechanical and thermal strain
ranges respectively, Tmax and Tmin are the maximum and minimum
TMF cycling temperatures respectively and a(T) is the coefcient of
thermal expansion. Finally, the TMF test results were compared and
analyzed with those obtained from IF tests at Tmax i.e. 340 C, in order to estimate the conservatism if any on the existing industrial
practice of performing isothermal tests at maximum temperature
of TMF for designing fatigue curves for piping systems in LWR systems. The end of fatigue life (Nf5) for TMF and IF tests was determined as the cycle number for which the maximum stress (rmax)
decreases by 5% below a tangent line drawn at the last point of zero
curvature on the curve of rmax vs. N in linear scaling.
3. Results
3.1. Fatigue life and cyclic stressstrain behavior
During TMF loading, the presence of difference in material
deformation behavior in relation to IF tests if any, can be well explained by cyclic stress response curves. Fig. 3a and b illustrates
the summary of maximum and minimum stress evolution during
TMF and IF at Tmax tests for two mechanical strain amplitudes of
0.3% and 0.5% in the two investigated steel grades. From their
stress response curves, it is quite evident that the fatigue lifetime
is similar in the three different loading conditions on a logarithmic
scale. Typical to the behavior of austenitic SS undergoing a cyclic
loading, all the above curves indicate an initial hardening regime
followed by softening and a distinct saturation stage until the
end of fatigue life. However, there is clear tendency in the TP
316L grade to have higher lifetime compared to the TP 347 grade
for identical test conditions. At lower strain amplitude (0.3%), both
materials (TP 347 and TP 316L) exhibit secondary hardening.
Between the two tested grades, TP 316L exhibits marginally higher

687

M. Ramesh et al. / International Journal of Fatigue 33 (2011) 683691

Fig. 3. Cyclic stress response curves during IP-TMF, OP-TMF and IF at 340 C for (a) TP 347 and (b) TP 316L.

Table 3
Summary of TMF and IF at Tmax test results for TP 347 under the temperature interval of 100340 C.
Mechanical strain
amplitude Demech/2 (%)

Test condition

Specimen name

Mean stress r at half


life Nf5/2 (MPa)

Fatigue life cycles, Nf5

0.3

In-phase TMF

IP1
IP2
IP3

38
32
37

26,358
24,122
21,270

23,916

Out-of-phase TMF

OP1
OP2
OP3

+40
+35.5
+38.5

17,614
20,268
22,152

20,011

IF at 340 C

IF1
IF2
IF3

+4.5
+0.5
3.5

18,228
19,342
20,178

19,249

In-phase TMF

IP1
IP2
IP3

39
33
40

12,278
12,684
11,482

12,148

Out-of-phase TMF

OP1
OP2
OP3

+38
+34
+34

7959
6427
8834

7740

IF1
IF2
IF3

+3.5
+4
1.5

8705
9267
7789

8587

0.5

IF at 340 C

secondary hardening compared to TP 347. It is also evident that the


TP 347 grade exhibits primary hardening (50 cycles) slightly earlier
than TP 316L grade (100 cycles) during all loading conditions and
strain amplitudes. Tables 3 and 4 illustrate the average fatigue lifetime and mean stress at saturation regime for the entire test
matrix.
In both TP 347 and TP 316L grades, in-phase TMF shows a higher
lifetime compared to out-of-phase TMF and IF tests in all tested
strain amplitudes (Tables 3 and 4). The individual tests also indicate a very low scatter between the three samples in a given test
condition (Fig. 3b, Tables 3 and 4). At 0.5% strain amplitude, in TP
316L IF test at Tmax shows the lowest fatigue lifetime of the investigated three test conditions. With decrease in strain amplitude
from 0.5% to 0.3%, a crossover was observed between out-of-phase
TMF and IF. In general, in both tested steel grades, the difference in
lifetime between the three test conditions decreases at 0.3% strain
amplitude. Out of the two tested materials, the non-stabilized
grade shows a higher lifetime in all the tested conditions and strain
amplitudes compared to the stabilized TP 347 grade. During a
typical IF test, a material experiences zero mean stress, but from
Tables 3 and 4 one can surmise a development of tensile mean
stress during an out-of-phase loading while it is in compressive
mode during in-phase. Finally, the obtained TMF and IF test results

Average fatigue
life cycles, N f 5

are compared with existing ASME Code fatigue curves (mean and
design curves) and ANL best t air model [29]. The ASME mean
curve was developed from small polished specimens of comparably high strength austenitic SS undergoing IF at room temperature
with a zero mean stress state in an air environment (Fig. 4). From
this experimentally tted ASME mean curve, described by the
Langer equation for austenitic SS

lnNf 6:954  2:0 lnDemech =2  0:167

the ASME design curve was then obtained by an adjusted factor of 2


on strain amplitude and 20 on fatigue life. The above described conservative approach in obtaining ASME design curve was made to include the effects of parameters, such as mean stress, surface nish,
size and geometry and loading history. It must be noted that this
conservatism does not include the effect of light water reactor coolant environments, which can lead to a much shorter fatigue lifetime
in reality. These two curves are based on experimental data obtained through the last three decades with comparably high
strength austenitic SS. Therefore, the experimental data (mean
curve) and the design curve are not a suitable representative for
the currently available piping austenitic SS. In air, at temperatures
up to 400 C, the isothermal fatigue data for types 304, 304L, 316,

688

M. Ramesh et al. / International Journal of Fatigue 33 (2011) 683691


Table 4
Summary of TMF and IF at Tmax test results for TP 316L under the temperature interval of 100340 C.
Mechanical strain
amplitude Demech/2 (%)

Test condition

Specimen name

Mean stress r at
half life Nf5/2 (MPa)

Fatigue life cycles, Nf5

0.3

In-phase TMF

IP1
IP2
IP3

42
36.5
40.5

27,039
39,780
36,000

34,273

Out-of-phase TMF

OP1
OP2
OP3

+39.5
+37
+40.5

23,926
26,770
26,382

25,692

IF at 340 C

IF1
IF2
IF3

0.5
1.5
0.5

25,570
24,060
29,414

26,348

In-phase TMF

IP1
IP2
IP3

36
38
36

14,807
12,548
14,071

13,808

Out-of-phase TMF

OP1
OP2
OP3

+36
+37
+39

11,314
9008
11,112

10,478

IF at 340 C

IF1
IF2
IF3

+2
3
3.5

8455
9802
10,077

9444

0.5

316L and 326NG SS are best represented by the ANL best t air
equation [29]:

lnNf 6:891  1:920 lnDemech =2  0:112

Mechanical strain amplitude, mech/2 [%]

It is a well known fact that TMF tests are quite complex to perform due to huge experimental efforts and large number of test
parameters compared to IF testing. Hence, lifetime estimation
using IF test data is the most efcient way of estimating the lifetime. IF data at 340 C were selected for lifetime prediction under
TMF loading. Three models were studied and compared in the present study. The standard deviation which gives an estimation of the
agreement between the model predicted and experimental results
was determined to evaluate the most appropriate model.

TP 316L TP 347
OP-TMF
IP-TMF
IF 340 C

0.9
0.7
0.5

0.3

Bes
t-F
ANL it Air
Mod
el
ASME
D

esign
C

ASME
Mean
C

urve

urve

0.1
5000

10000

20000

40000

Fatigue life cycles, Nf5


Fig. 4. Experimental data (with scatter band) of TP 347 and TP 316L with ASME
mean, design and ANL best t air model curve.



De p
lne0f C ln2Nf
2

where Dep is the plastic strain range; e0f and C are the material constants; Nf is the fatigue lifetime. From IF at Tmax test results, the
hardening coefcient and hardening exponent must be identied
to calculate the lifetime depending on total plastic strain (from isothermal tests it was calculated as: (a) in 316L, e0f = 2.5 and C = 0.6;
(b) in TP 347, e0f = 11.47 and C = 0.8). Based on total plastic strain
from TMF experiments and earlier identied material constants,
the lifetime is estimated using the above mentioned relation. This
is followed by comparing the obtained prediction with experimental TMF test results in a parity plot (Fig. 5). The standard deviation
between the experimental and predicted lifetimes was then calculated as 0.272.
3.2.2. Ostergrens frequency modied damage function
Ostergren developed a new approach for predicting strain controlled low cycle fatigue life at an elevated temperature by relating
it to net tensile hysteresis energy [30]. The net tensile hysteresis

Predicted fatigue life cycles

3.2. Lifetime estimation

life cycles, N f 5

3.2.1. CofnManson criterion


According to Cofn and Manson [23], there exists a linear relationship between the number of cycles to the end of fatigue lifetime and applied plastic strain range

ln
Fig. 4 shows that only out-of-phase and IF results of the TP 347
grade for the mechanical strain amplitude of 0.3% possess shorter
fatigue life times than predicted by the ANL best t air equation.

Average fatigue

TP
347

10000

1000
1000

TP
316L TMF mech/2
IP
OP
IP
OP

0.3 %
0.3 %
0.5 %
0.5 %

10000

Experimental fatigue life cycles


Fig. 5. Comparison between the experimental and predicted TMF lives using
CofnManson criterion.

689

TP
347

TP
316L TMF mech/2
IP
OP
IP
OP

10000

1000
1000

Predicted fatigue life cycles

Predicted fatigue life cycles

M. Ramesh et al. / International Journal of Fatigue 33 (2011) 683691

0.3 %
0.3 %
0.5 %
0.5 %

TP
347

mech/2

0.3 %
0.3 %
0.5 %
0.5 %

10000

1000
1000

10000

TP
316L TMF
IP
OP
IP
OP

10000

Experimental fatigue life cycles

Experimental fatigue life cycles


Fig. 6. Predicted fatigue life vs. experimental fatigue life by Ostergren frequency
modied damage function.

Fig. 7. Comparison of simulated and experimental life time for both TP 347 and TP
316L pipe steels with SmithWatsonTopper rule.

energy during fatigue damage can be calculated as the product of


plastic strain range and average maximum stress level. This relationship can be given by

out-of-phase TMF test conditions. In other words, in-phase TMF


develops a compressive mean stress while out-of-phase TMF develops a tensile one. The development of tensile mean stress in out-ofphase TMF leads to a lower lifetime when compared to in-phase
TMF test. In general, the IF tests at Tmax, of both the materials exhibit lower lifetime than that of TMF in the two tested strain amplitudes, thereby indicating more severe fatigue damage at higher
temperature isothermal tests. This can be attributed to decrease
in ow stress of the material with increase in temperature leading
to a shorter fatigue lifetime. The exception to the above observation is out-of-phase TMF at 0.5% strain amplitude in TP 347 which
can be attributed to inherent heterogeneity of this material as observed during our earlier investigations and also to scatter associated with such fatigue tests [28]. However, with the lower strain
amplitude (0.3%), in both the materials secondary hardening plays
a signicant role on the end of fatigue lifetime resulting in crossover with out-of-phase TMF. This conrms the earlier crossover
observations by Zamrik et al. [15] during his tests with TP 316L.
However, for the TP 347 grade, IF and out-of-phase TMF show similar lifetimes at 0.3% and 0.5% strain amplitudes. Hence, further
tests need to be carried out to understand the trend clearly. In
any case, from an engineering point of view, this does not have a
relevant signicance concerning the lifetimes.
Cyclic stress response behavior of austenitic SS has been studied
in great detail over the years. It is characterized by primary hardening at an early stage, which is due to: (a) the generation of dislocations; (b) their mutual interaction; and (c) the interaction of
interstitial solute atoms with the dislocations, also known as dynamic strain ageing (DSA). This stage is typically followed by a softening and a distinct saturation regime, characterized by a
rearrangement of dislocation structures favoring cyclic strain localization. At lower strain amplitudes (<0.5%), a secondary hardening
can be observed, which is most likely caused by an increase in dislocation density over lifetime and DSA at the relevant temperatures
and strain rate for austenitic SS. It is also important to note that
with decreasing strain amplitude and strain rate, a slightly stronger inuence of DSA comes into effect [26]. Here, it is also pertinent
to note that the tested temperature interval of 100340 C and applied strain rate of 104 s1 encompass a partially active DSA domain (250600 C), which plays an important role in the cyclic
stress response during the entire lifetime. It is also well established
in literature that DSA reduces the fatigue resistance of the material
during strain controlled fatigue testing, by inducing heterogeneous
deformation and DSA induced hardening. In both the niobium-stabilized TP 347 and non-stabilized TP 316L, there exists a difference
in concentration of interstitial atoms (C or N) in the solid solution.
The addition of Nb to the stabilized grade is primarily aimed at
lowering the concentration of C and N in the solid solution by its

Nf Lrmax Dep c

where rmax is the average maximum stress, L and C are the material
constants. By plotting Ostergren energy (rmax Dep ) vs. Nf for the IF
test data of both materials, we can obtain the parameters L and C
of the above Eq. (5) by multiple regression analysis. The obtained
values of material parameters of C = 1.51, L = 8103 in TP 347
grade; and C = 1.22, L = 14,764 for TP 316L grade, were comparable with the previous works given in literature [11]. The calculated
lifetime from the above function and experimental life times were
then plotted on a logarithmic scale for comparison (Fig. 6). The standard deviation between simulated and experimental values was
found to be 0.29.
3.2.3. SmithWatsonTopper criterion
The inuence of mean stress on the end of fatigue lifetime can
be well described by SmithWatsonTopper parameter.

Nf Lrmax  Demech =2c

where Demech/2 is the applied total mechanical strain amplitude.


Similar to earlier cases, material constants were calculated from
the isothermal test data as: (a) In TP 316L, ln(L) = 28.06 and
C = 3.11; (b) In TP 347, ln(L) = 24.84 and C = 2.6. Finally, similar
to the earlier two models, the estimated and experimental fatigue
lifetimes were plotted on a logarithmic scale with a scatter band
of 2 (Fig. 7) and standard deviation was determined as 0.232.
4. Discussion
In an austenitic SS subjected to TMF loading between 100 C and
340 C in air, fatigue is the only activated damage mechanism that
affects the lifetime of the material, with creep and environmental
effects being negligible. In our case, all the three tested conditions
have similar fatigue lifetime from an engineering point of view.
This can be explained by similar plastic strain range experienced
on cyclic loading from in-phase TMF, out-of-phase TMF and IF at
Tmax tests. However, the stress paths near the end of the fatigue life
during TMF tests differ slightly from typical isothermal tests owing
to the development of non-zero mean stress during TMF testing. In
TP 316L and TP 347, the higher lifetime of in-phase TMF compared
to the other two test conditions for both strain amplitudes is primarily due to the effect of mean stress. This can be explained by
dependency of yield stress on temperature, which results in the
development of different peak stress values for the in-phase and

690

M. Ramesh et al. / International Journal of Fatigue 33 (2011) 683691

Moreover, it is evident that in the tested temperature of 100


340 C, both TMF and IF test impart a similar plastic strain range
resulting in a good lifetime prediction. There exists a general tendency of estimated lifetime to be marginally lesser than the experimental TMF lifetime in some cases. However, they are less
signicant from an engineering perspective.
5. Conclusion
Based on the experimental results and above discussions, the
following conclusions can be drawn.

Fig. 8. Secondary electron image of a crack initiated near an austenite matrixcoarser carbide interface in TP 347 subjected to TMF loading.

precipitation as NbX (where X denotes C or N). Therefore, in the


tested material of TP 347, fewer interstitial solute atoms are available in solid solution compared to non-stabilized grade of TP 316L
due to formation of NbX precipitates (Fig. 2c). The transmission
electron image shows the dense homogeneous distribution of
NbX precipitates along {1 1 1} contributing to early initial hardening in TP 347 by Orowan mechanism. This was conrmed during
our earlier investigations on the same material [28]. According to
the ndings by Hong et al. [26] concerning the effect of DSA on fatigue life behavior of austenitic SS, TP 316L, which has a higher concentration of interstitial atoms in austenitic matrix should exhibit
a shorter lifetime when compared to Nb stabilized TP 347 owing to
stronger effect of DSA. This is also evident in slightly lower secondary hardening in TP 347 compared to TP 316L at lower strain amplitude of 0.3%. However, earlier microscopic investigations revealed
that TP 347 exhibits a lower lifetime compared to TP 316L due to
the presence of microscopic voids near the NbXmatrix interface
in the as supplied condition (Fig. 8). This gure illustrates the secondary electron image of a crack initiated near an austenite matrix-coarser carbide interface during SEM investigation on TP 347
material subjected to TMF loading until the end of fatigue life. This
was also conrmed during earlier investigations by Lee et al. [31].
These voids can act as heterogeneous crack nucleation zones leading to shorter fatigue lifetime. In addition, the absence of such heterogeneous nucleation sites delays the onset of crack initiation
resulting in a higher lifetime. The onset of primary hardening in
TP 347 is mainly attributed to the presence of ne nano-metric
NbC precipitates on the active slip planes which contributes to
early hardening by Orowan mechanism (Fig. 2c).
It is quite evident from Fig. 4 that all our fatigue test results are
within the limits of the ASME design curve. The obtained fatigue
test results of both the materials are also inside the ANL best t
air model curve. However, there exists a slight deviation when it
comes to TP 347 fatigue data at low strain amplitudes which can
be attributed to the earlier mentioned effect of material variability
in the form of micro-cavities. In general, the three models analyzed
here show good prediction capabilities for both the tested strain
amplitudes from IF data. The SmithWatsonTopper criterion
prediction is more precise than the other two, which is evident
from the corresponding standard deviation values. This can be explained by the effect of mean stress taken into account during the
lifetime estimation by this model. The observed error on the estimation of lifetime in both the materials for this loading condition
is well inside the scatter of factor 2. This is primarily attributed
to the fact that, in all the above damage criterions, plastic strain
range is incorporated as a main constituent of damage parameter.

 In-phase TMF showed higher fatigue lifetime than IF at Tmax and


out-of-phase TMF tests in the tested temperature interval and
strain range for TP 347 and TP 316L.
 In TP 316L, a crossover between out-of-phase TMF and IF at Tmax
was observed with decreasing strain amplitude.
 TP 316L showed a higher lifetime than TP 347, between the two
tested grades of austenitic SS under all test conditions.
 In general, all the three models give a satisfactory prediction by
using isothermal data, with lifetime estimation by SmithWatsonTopper criterion being more precise than estimates by the
other two models.
 In TP 347 grade, out-of-phase TMF and IF at Tmax fatigue lifetimes are similar in both strain amplitudes.

Acknowledgements
Financial support by the Swiss Federal Nuclear Safety Inspectorate (ENSI) under Contract numbers H-100397 and H-100625 is
gratefully acknowledged. The authors would like to express their
thanks to Dr. Christian Solenthaler for the TEM observations and
to EMEZ Electron Microscopy ETH Zurich for SEM investigations.
References
[1] IAEA. Assessment and management of ageing of major nuclear power plant
components important to safety. Primary piping in PWRs. In: IAEA-TECDOC1361, Vienna; 2003.
[2] Metzner KJ, Wilke U. European THERFAT project thermal fatigue evaluation
of piping system Tee-connections. Nucl Eng Des 2005;235(24):47384.
[3] Faidy C. Thermal fatigue of reactor components in OECD-NEA countries: a
threefold program to enhance cooperation. In: 18th Int conf struct mech
reactor technol, Beijing; 2005.
[4] ASTM E2368-04 Standard practice for strain controlled thermomechanical
fatigue testing. ASTM Int, West Conshohoken; 2005.
[5] Hhner P, Rinaldi C, Bicego V, Affeldt E, Brendel T, Andersson H, et al. Research
and development into a European code-of-practice for strain-controlled
thermo-mechanical fatigue testing. Int J Fatigue 2008;30(2):37281.
[6] Loveday MS, Bicego V, Hhner P, Klingelhffer H, Khn HJ, Roebuck B. Analysis
of a European TMF inter-comparison exercise. Int J Fatigue 2008;30(2):38290.
[7] Hhner P, Rinaldi C, Affeldt E, Beck T, Klingelhffer H, Loveday MS. Validated
code-of-practice for strain-controlled thermo-mechanical fatigue testing. EUR
22281 EN. European Communities, Luxembourg; 2005.
[8] Suresh S. Fatigue of materials. Cambridge University Press; 1998.
[9] Christ H-J. Effect of environment on thermomechanical fatigue life. Mater Sci
Eng A 2007;468470:98108.
[10] Majumdar S. Thermomechanical fatigue of type 304 stainless steel. In: ASME
PVP, vol. 123; 1987. p. 316.
[11] Nagesha A, Valsan M, Kannan R, Bhanu Sankara Rao K, Bauer V, Christ HJ, et al.
Thermomechanical fatigue evaluation and life prediction of 316L(N) stainless
steel. Int J Fatigue 2009;31(4):63643.
[12] Kuwabara K, Nitta K. Thermalmechanical low cycle fatigue under creep
fatigue interaction on type 304 stainless steel. In: Proc ICM 3; 1979. p. 6978.
[13] Remy L. Fatigue and thermomechanical fatigue at high temperature.
Encyclopedia Mater: Sci Technol 2001;3:286879.
[14] Shi HJ, Wang ZG, Su HH. Thermomechanical fatigue of a 316L austenitic steel
at two different temperature intervals. Scripta Mater 1996;35(9):110713.
[15] Zamrik SY, Davis DC, Firth LC. Isothermal and thermomechanical fatigue of
type 316 stainless steel. ASTM Special Tech Pub 1996;1263:96116.
[16] Zauter R, Christ HJ, Mughrabi H. Some aspects of thermomechanical fatigue of
AISI 304L stainless steel: part I. Creep fatigue damage. Metall Mater Trans A
1994;25(2):4016.

M. Ramesh et al. / International Journal of Fatigue 33 (2011) 683691


[17] Zauter R, Petry F, Christ HJ, Mughrabi H. Thermomechanical fatigue of the
austenitic stainless steel AISI 304L. ASTM Special Technical Publication; 1993,
p. 7090.
[18] Ellison EG, Al-Zamily A. Stress response under thermalmechanical strain
cycling for a 1 CrMoV ferritic steel and two 316 stainless steels. Fatigue Fract
Eng Mater Struct 1994;17(1):3951.
[19] Fujino M, Taira S. Effect of thermal cycling on low cycle fatigue life of steels
and grain boundary sliding characteristics. Proc ICM3 1979;2:4958.
[20] Boismier DA, Sehitoglu H. Thermo-mechanical fatigue of Mar-M247. Part 1.
Experiments. J Eng Mater Technol Trans ASME 1990;112(1):6879.
[21] Neu RW, Sehitoglu H. Thermomechanical fatigue, oxidation, and creep: part i.
Damage mechanisms. Metall Trans A 1989;20(9):175567.
[22] Huang ZW, Wang ZG, Zhu SJ, Yuan FH, Wang FG. Thermomechanical fatigue
behavior and life prediction of a cast nickel-based superalloy. Mater Sci Eng A
2006;432(12):30816.
[23] Cofn LF. A study of the effects of cyclic thermal stresses on a ductile metal.
Trans ASME 1954;76:93150.
[24] Smith KN, Watson P, Topper TH. Stressstrain function for the fatigue of
metals. J Mater 1970;5(4):76778.

691

[25] Christ HJ, Jung A, Maier HJ, Teteruk R. Thermomechanical fatigue damage
mechanisms and mechanism-based life prediction methods. Sadhana Acad
Proc Eng Sci 2003;28(12):14765.
[26] Hong SG, Lee KO, Lee SB. Dynamic strain aging effect on the fatigue resistance
of type 316L stainless steel. Int J Fatigue 2005;27(1012):14204.
[27] Chauvot C, Sester M. Fatigue crack initiation and crystallographic crack growth
in an austenitic stainless steel. Comput Mater Sci 2000;19(14):8796.
[28] Ramesh M, Thermomechanical fatigue behavior of austenitic stainless tube
and pipe steels under light water reactor relevant temperature conditions and
associated microstructural evolution. Dissertation No 19137 ETH Zurich; 2010.
[29] Chopra, OK, Shack WJ. Effect of LWR coolant environments on the fatigue life
of reactor materials. NUREG/CR6909, ANL06/08; 2007.
[30] Ostergren WJ. Damage function and associated failure equations for predicting
hold time and frequency effects in elevated temperature low cycle fatigue. J
Test Eval 1976;4(5):32739.
[31] Lee BS, Oh YJ, Yoon JH, Kuk IH, Hong JH. JR fracture properties of SA508-1a
ferritic steels and SA312-TP347 austenitic steels for pressurized water
reactors (PWR) primary coolant piping. Nucl Eng Des 2000;199(12):11323.

S-ar putea să vă placă și