You are on page 1of 10

International Journal of Fatigue 156 (2022) 106632

Contents lists available at ScienceDirect

International Journal of Fatigue


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

Cyclic deformation behavior of metastable austenitic stainless steel AISI


347 in the VHCF regime at ambient temperature and 300 ◦ C
Tobias Daniel, Marek Smaga *, Tilmann Beck
Institute of Materials Science and Engineering, TUK, 67663 Kaiserslautern, Germany

A R T I C L E I N F O A B S T R A C T

Keywords: Very high cycle fatigue (VHCF) tests on metastable austenitic stainless steels using an ultrasonic fatigue testing
Cyclic deformation behavior (USFT) system are challenging due to the transient material behavior and associated pronounced self-heating
VHCF effects. Because the conventional measurement of stress-strain hysteresis is not possible in USFT, the cyclic
Ultrasonic testing system
deformation behavior can’t be described in a conventional manner. Hence, an energy-based approach is pro­
Metastable austenite
Ambient and elevated temperature
posed for the characterization of cyclic deformation behavior of austenitic stainless steels in the VHCF regime at
α′ -martensite ambient temperature (AT) and elevated temperature. Therefore, in-situ dissipated energy and temperature
measurements were performed. At AT both values underwent a change during cyclic loading, while at 300 ◦ C
only the dissipated energy changed. The investigated metastable austenitic stainless steel AISI 347
(X6CrNiNb1810, 1.4550) showed cyclic softening in the high cycle fatigue (HCF) regime (Nf < 107) at both
testing temperatures, which led to specimen failure. In the VHCF regime, cyclic hardening was observed, which
resulted in reaching the limiting number of load cycles Nl = 2 × 109 without specimen failure. This change in the
cyclic deformation behavior by a small reduction of the stress amplitude led to a horizontal course of the Woehler
curve in the VHCF regime and resulted in the true endurance limit for the investigated material. At AT, ferro­
magnetic α′ -martensite was detected in all fatigued specimens using magnetic measurements. At 300 ◦ C only
specimens which reached the limiting number of cycles without failure exhibited a small amount of
α′ -martensite.

[7,9] which both allow loading frequencies up to f = 1 kHz, ultrasonic


1. Introduction fatigue testing (USFT) systems [7,10] operated at f = 20 kHz are widely
used. Resonance pulsators allow force- and displacement-controlled fa­
During service life, many cyclically loaded technical components tigue testing, whereby the load frequency is affected by the stiffness of
undergo 108 up to 1011 load cycles, i.e. very high cycle fatigue (VHCF) the testing system. Thus, by changing the specimen geometry and/or the
loadings. Conventional fatigue investigations assumed a “true” fatigue oscillating mass, the load frequency can be influenced [7,11]. In servo-
limit for materials that achieve a horizontal asymptote in the Woehler hydraulic testing systems, the load frequency can be adjusted almost
diagram between 106 and 107 load cycles, such that, below the fatigue continuously and also strain-controlled tests can be realized up to fre­
limit, no fatigue failure occurs [1]. However, VHCF research in the last quencies of nearly 100 Hz [12]. The operating principle of ultrasonic
decades has shown that for many materials a further decrease in fatigue fatigue testing systems is based on the inverse piezoelectric effect, which
strength can be found at load cycles beyond 107 [2–4]. Consequently, leads to the transformation of an electrical signal from the ultrasonic
investigations of the fatigue behavior in the VHCF regime up to 1011 generator into a high-frequency mechanical oscillation, which is
load cycles have gained increasing attention [5]. In order to achieve amplified and transmitted to the specimen via a booster [13,14]. All
such high numbers of load cycles within reasonable time, very high parts of the system are designed to assure that under resonance condi­
loading frequencies are necessary. Using conventional servo-hydraulic tions a standing wave develops along the major load axis with a
testing systems operating usually at f = 5 Hz, fatigue tests up to 2 × maximum stress amplitude in the center of the gauge length of the
109 load cycles would take more than ten years. For this reason, high- specimen (Fig. 1a) [15,16]. Because force measurement using a load cell
frequency systems are used for VHCF investigations. In addition to a isn’t possible in USFT systems, force-controlled fatigue tests can’t be
servo-hydraulic high-frequency system [6–8] and resonance pulsators performed, and only displacement amplitude controlled tests are viable.

* Corresponding author.
E-mail address: smaga@mv.uni-kl.de (M. Smaga).

https://doi.org/10.1016/j.ijfatigue.2021.106632
Received 10 August 2021; Received in revised form 30 September 2021; Accepted 26 October 2021
Available online 19 November 2021
0142-1123/© 2021 Elsevier Ltd. All rights reserved.
T. Daniel et al. International Journal of Fatigue 156 (2022) 106632

Nomenclature Pmax maximum power of ultrasonic generator


PUS power of ultrasonic generator
A elongation after fracture Q dissipated heat
AISI American Iron and Steel Institute Rp0.2 0.2% yield strength
AT ambient temperature RZ roughness
bcc body-centered cubic SFE stacking fault energy
c specific heat capacity sa displacement amplitude
Cr Chromium T temperature
Edis dissipated energy t time
f frequency U internal energy
fcc face-centered cubic USFT ultrasonic fatigue testing
feff effective frequency UTS ultimate tensile strength
HCF high cycle fatigue VHCF very high cycle fatigue
hdp hexagonal close-packed W deformation energy
I integral α′ bcc α′ -martensite phase
k calculation factor γ fcc γ-austenite phase
LCF low cycle fatigue ΔTpulse temperature change during pulse sequence
m mass ε hdp ε-martensite phase
MS Martensit start temperature εt total strain
N number of cycles μ thermal conductivity
Nf number of cycles to failure ξ ferromagnetic phase content in FE-%
Ni Nickel ρ density
Nl limiting number of cycles σ stress
P proportional σa stress amplitude

cooling is assured during the pause time (Fig. 1d). Note that possibilities
for characterizing the cyclic deformation behavior using USFT systems
are very restricted, while various measuring techniques have been
established for fatigue tests at lower frequencies, predominantly with
servo-hydraulic systems. Stanzl-Tschegg et al. [20] showed the usage of
strain gauges for total strain controlled fatigue tests and the indirect
determination of the plastic strain amplitude using relations with the
measured temperature increase. However, determining stress-strain
hysteresis is not possible due to the fact that load cells are not appli­
cable in USFT systems [2]. Therefore, other physically based data have
Fig. 1. Schematic diagrams showing: a) the operating principle of an ultrasonic to be taken into consideration to characterize the cyclic deformation
fatigue testing (USFT) system; b) the development of specimen displacement behavior. During ultrasonic fatigue tests, the specimen temperature can
(s); c) the development of the power of the ultrasonic generator (PUS); d) the be measured contactless by a pyrometer. Thus, the temperature differ­
development of specimen temperature (T) during one pulse sequence; e) the ence before and after a pulse sequence can be determined as ΔTpulse
relationship between the dissipate energy (Edis) and the strain energy density (Fig. 1d). The authors had already demonstrated in [13,14,17] that
(W) and change in temperature ΔTpulse [21]. ΔTpulse is a sensitive value for detecting specimen failure in cyclically
stable materials. In addition, they also used the dissipated energy (Edis)
Generally, the specimen displacement amplitude is correlated in pre­ which is defined as an integral of the generator power over the pulse
liminary tests by a linear relationship with the applied generator time (Fig. 1e) to describe the fatigue behavior during high-frequency
voltage, and checked during the VHCF test using a laser vibrometer [17] cyclic loading.
or a vibration gauge [18,19]. Using the measured displacement ampli­ In this study, measurement of the power of the ultrasonic generator
tude at the specimen end, the load amplitude in the center of the spec­ (PUS) and specimen temperature (T) versus time (t) were used to
imen can be calculated with a linear elastic correlation factor. Due to determine the dissipated energy (Edis) and the change in temperature
the, compared with other methods, significantly higher load frequency ΔTpulse as a result of cyclic deformation during a pulse load sequence.
of approximately 20 kHz, almost all materials undergo significant self- Both values correspond to the strain energy density determined
heating during ultrasonic fatigue testing, even at amplitudes close to, conventionally from stress-strain hysteresis loops [21–24] (Fig. 1e).
or at the fatigue limit. Hence, fatigue tests using USFT systems are Piotrowski et al. [21] defined the area of a stress-strain hysteresis as the
usually carried out in a pulse-pause mode [13,14,17] (Fig. 1b). As shown irreversible plastic deformation energy (W) per load cycle and unit
in Fig. 1c, at the beginning of the pulse sequence the power of the ul­ volume. W is composed of dissipated heat Q and deformation induced
trasonic generator (PUS) increases rapidly until a maximum value (Pmax) increase of the material internal energy U (eq. (1)), whereby about
is reached approximately at the time when the desired displacement 90–95% of the deformation energy is converted into heat.
amplitude of the specimen (sa) is established (Fig. 1b). Then the power is
W = Q+U (1)
regulated in such a way, that the displacement amplitude and, there­
with, the stress amplitude in the center region of the test piece, remain In USFT systems, at first electrical energy is converted into me­
constant until the end of the pulse sequence were the generator power chanical energy. A small part of the mechanical energy is dissipated in
and displacement amplitude decrease to zero. The pulse and pause time the test rig. The remaining mechanical energy leads to the oscillation of
must be adjusted in a suitable way, such that the specimen temperature the specimen, which is converted into deformation energy during cyclic
does not exceed a limit value during the cyclic loading and sufficient loading. As mentioned above, approximately 95% of the deformation

2
T. Daniel et al. International Journal of Fatigue 156 (2022) 106632

energy is dissipated to heat and can, hence, be measured continuously fatigue failure in the VHCF regime [2,15,16,26], resulting in a hori­
by observation of the sample temperature during each load pulse. The zontal asymptote in the Woehler diagram above 107 load cycles. Only
increase in temperature during the pulse time is a cumulative process. few research was published on VHCF of austenitic steels at elevated
Hence, the change in temperature as well as the dissipated energy temperatures. Smaga et al. [2,43] were able to demonstrate the exis­
continuously increase during a load pulse, which lasted between 40 and tence of a true fatigue limit in the VHCF regime for metastable austenites
72 ms (see 3.2) in the tests on the investigated material and contained AISI 347 with a fully austenitic microstructure in the initial state even at
approximately 600–1200 load cycles [13,14,17,25]. Therefore, both elevated temperatures by fatigue tests on a servo-hydraulic testing sys­
values offered the possibility of analyzing the cyclic deformation tem at f = 980 Hz and T = 300 ◦ C up to a load cycle limit of Nl = 5 × 108.
behavior and, therewith, provided implicit information about the Wagner et al. [44] investigated the fatigue behavior of the stable
change in microstructure during the fatigue process induced by ultra­ austenitic steel AISI 303 at 600 ◦ C as well as 700 ◦ C. They observed a
sonic loadings [2,17,26]. continuous negative slope in the Woehler line up to 109 load cycles. In
Austenitic CrNi steels, also called “austenitic stainless steels”, often the VHCF regime, both surface and volume crack initiation occurred.
shown a metastable state of the austenitic phase after quenching from The aim of this work was, on the one hand, to characterize the cyclic
solution annealing temperature [2,27]. This allows a deformation deformation behavior in the VHCF regime of metastable austenitic
induced phase transformation from paramagnetic face-centered cubic stainless steel using the dissipated energy determined from the power of
(fcc) γ-austenite into a more stable microstructure, i.e. paramagnetic the ultrasonic generator (AT and 300 ◦ C), as well as the temperature
hexagonal close-packed (hdp) ε-martensite and/or ferromagnetic body- change during a pulse sequence (only at AT). On the other hand, the aim
centered cubic (bcc) α′ -martensite which significantly enhances the fa­ was to increase the knowledge about fatigue life of this materials class in
tigue resistance of this materials class [2,24,28,29]. The metastability of the VHCF regime at AT and T = 300 ◦ C.
such austenites strongly depends on the chemical composition and can
be assessed by different parameters based on empirical equations (2–4) 2. Material
[2,30–32]. The Md30 temperature according to Angel describes the
temperature at which 50 vol% of α′ -martensite is achieved by a true 2.1. Investigated material
plastic deformation of 30%. At the MS temperature a thermally induced
phase transformation occurs without externally applied deformation The investigated material was the Niobium stabilized metastable
[30,33]. With the value of the stacking fault energy (SFE), the defor­ austenitic stainless steel AISI 347 (X6CrNiNb1810, 1.4550). The VHCF
mation mechanisms under cyclic loading can be identified. In particular, specimens were extracted from an original nuclear power plant surge
the Md30 value is often used to compare the austenite stability of line pipe with an outside diameter of 333 mm and a wall thickness of 36
different CrNi steels [34,35]. mm. The pipe was manufactured seamless, drilled from the inside,
◦ turned from the outside and delivered in a solution annealed state
Md30,Angel in C = 413 − 462x(C + N) − 13.7xCr − 9.5xNi − 8.1xMn
(2) (1050 ◦ C / 10 min), such that in the initial state no α′ -martensite was
− 18.5xMo − 9.2xSi detected by microscopic, x-ray and magnetic Feritscope™ measure­

ments [45]. The surge line pipe was not previously used for applications
MS,Eichelmann in C = 1350− 1665x(C +N)− 42xCr − 61xNi − 33xMn− 28xSi in nuclear power plants.
(3) The chemical composition of the investigated material determined
by spectral analysis is given in Table 1 and corresponds to the limits
mJ
SFE in = 25.7 + 2xNi − 0.9xCr − 1.2xMn + 410xC − 77xN − 13xSi given in the respective international standard [46]. Note, that this
m2 standard mainly ensures sufficient corrosion resistance and does not
(4)
focus on the metastability of the austenite. Consequently, the allowed
The concentrations of all alloying elements have to be inserted in wt difference in the chemical composition leads, for the same type of
%. austenitic stainless steel, to a significant difference in their metastability
Due to good mechanical and technological properties as well as [34]. Regarding the metastability of the investigated AISI 347 evaluated
excellent corrosion resistance, austenitic stainless steels are very by equations (2)-(4), it is shown that the investigated material is in a
important engineering materials, e.g. used for components with high metastable state at ambient temperature (Table 1).
safety requirements in reactor pressure vessels of nuclear power plants, The microstructure in the initial state was fully austenitic with
[26,32]. Typical for these applications are stresses in the LCF regime due typical annealing twins and relatively large grains (Fig. 2a). The cross
to reactor start-up and shut-down processes. These are superimposed by section showed a rather high chemical homogeneity compared to studies
high frequency loadings in the VHCF regime due to fast cyclic thermal on bar material, which often shows pronounced segregation [39].
fluctuations triggered by fluid dynamic processes [36]. VHCF in­ Quantitative microstructure analysis revealed a mean grain size of 120
vestigations at the stable austenitic stainless steel AISI 904 under µm without the consideration of solution annealing twins. All specimens
rotating bending showed a negative slope of the S-N curve in the VHCF of the investigated material for fatigue and tensile tests were mechani­
regime up to 1010 load cycles [37]. The crack initiation occurred on the cally and electrolytically polished to obtain a surface roughness of Rz =
specimen surface. Due to the high content of alloying elements in AISI 0.16 µm and to remove small amounts of α′ -martensite at the specimen
904L the microstructure was fully austenitic and no deformation surface that may have formed in the turning process. Afterwards, mag­
induced phase transformation was observed [38,39]. Investigations of netic Feritscope™ measurements on the specimens showed no ferro­
the metastable austenitic stainless steel AISI 304 with a high content of magnetic phase content ξ = 0.00 FE-%, confirming that no α′ -martensite
α′ -martensite (>30 vol%) produced by quasi static pre-deformation at was present in the initial state. As a safeguard, this measurement was
low temperature showed a second decrease in the S-N curve up to performed at four different locations in the gauge length, rotated against
highest load cycles [40–42]. Furthermore, for N > 3 × 106, a transition each other by 90◦ .
from crack initiation at the specimen surface to internal failure at oxide The stress-strain curve and the respective evolution of the
inclusions was observed. If an inclusion was located in a previously α′ -martensite content during monotonic tensile tests at AT and 300 ◦ C
formed α′ -martensite area, this led to local stress concentrations in a are shown in Fig. 2b. The results are summarized in Table 2. As ex­
region where the material no longer had the potential to withstand crack pected, the material strength decreased with increasing temperature.
initiation or crack propagation through cyclic hardening [41]. In Due to no presence of α′ -martensite formation at 300 ◦ C during mono­
contrast, metastable austenites with fully austenitic microstructure or tonic tensile loading, the ultimate strain to failure was lower at elevated
α′ -martensite contents up to 27 vol% in the initial state did not show any temperature. At ambient temperature the plastic deformation of 66% led

3
T. Daniel et al. International Journal of Fatigue 156 (2022) 106632

Table 1
Chemical composition in wt% and austenite stability parameters of the investigated metastable steel AISI 347.
C N Cr Ni Nb Mn Mo Si MS in ◦ C Md30 in ◦ C SFE in mJ/m2 at AT

0.04 0.007 17.6 10.64 0.62 1.83 0.29 0.41 − 189 25 42

Fig. 2. a) Optical micrograph of the investigated material in its initial state, etched with Beraha II and b) stress strain behavior at ambient temperature and T =
300 ◦ C as well as the formation of α′ -martensite during tensile tests [47].

Table 2
Mechanical properties, α′ -martensite content (ξ) after specimen failure, density (ρ), Poisson’s ratio as well as thermal conductivity (µ).
Rp0.2 in MPa Young’s Modulus in GPa UTS in MPa A in % ξ in FE-% ρ in g/cm3 Poisson’s ratio Thermal conductivity µ in W/(m⋅K)

AT 242 183 569 66 4.41 7.85 0.3 15 [46]


300 ◦ C 180 159 357 36 0.00 – –

to the formation of a 4.41 FE-% ferromagnetic phase, i.e. α′ -martensite. the specimens were designed for axial oscillation in the first Eigenmode
Note that the linear relationship between FE-% and vol% of at the operating frequency via dynamic FEM-simulation [8]. The
α′ -martensite of the investigated material was determined systemati­ Young’s Modulus of the investigated material determined in tensile
cally by XRD measurements and can be described as 1.57xFE-% = vol% tests, Poisson’s ratio as well as material density (s. Table 2) were used as
α′ -martensite [48]. However in the further text, the original measure­ input data for the simulation. Due to the temperature-dependent
ment data of the Feritscope™ without recalculation into vol% Young’s Modulus, specimens with different geometries were used for
α′ -martensite is given to keep comparability to results presented else­ the tests at ambient temperature and T = 300 ◦ C (Fig. 3a and d). For the
where [2,34,43]. fatigue tests at ambient temperature, a booster with an amplitude
transformation ratio of 1:2.5 was used, i.e. the axial displacement
3. Experimental setup and testing procedure amplitude at the converter was amplified by a factor of 2.5. Due to the
lower stress amplitudes required at T = 300 ◦ C, a booster with an
3.1. Ultrasonic fatigue testing (USFT) systems for tests at AT and T = amplitude transformation ratio of 1:1.5 was used for these tests. The
300 ◦ C maximum stress amplitude was calculated based on steady state FEM
simulations from the linear relationship between maximum stress
Fig. 3 shows the details of the USFT systems operating at AT and amplitude and maximum displacement amplitude at the lower end of
300 ◦ C as well as the respective specimen geometries. To ensure fully the specimen measured by a 1-axis laser vibrometer. A calculation factor
reversed loading with the maximum displacement at the specimen end of kAT = 26.3 MPa/µm was determined for the tests at AT and
and the maximum stress in the middle of the gauge length (see Fig. 1a), k300◦ C = 24.6 MPa/µm for the tests at T = 300 ◦ C. Accordingly, in the

Fig. 3. a) and d) specimen geometry designed by FEM-simulation in ABAQUS; b) and c) photo of experimental setup for fatigue tests at AT as well as 300 ◦ C.

4
T. Daniel et al. International Journal of Fatigue 156 (2022) 106632

results part only the stress amplitude is used to describe the fatigue re­
sults. In order to achieve a laser vibrometer signal with high quality and
stability over a complete VHCF experiment, an appropriate reflecting
foil was fixed to the measuring points (see Fig. 3b and c). With this
experimental setup, the fatigue test was continuously monitored and the
exact number of load cycles could be determined. To enable contactless
temperature measurement, the specimens were covered with
temperature-resistant black matt lacquer before starting the test to
achieve an emission value of approximately 1.
For the isothermal fatigue tests at 300 ◦ C, an induction heating
system was integrated into the USFT system (Fig. 3c) [49]. The tem­
perature measurement/control was contactless via a pyrometer. To Fig. 4. Influence of the readjustment of the load pulse, a) load pulse for the
purely austenitic microstructure at the beginning of the test, b) load pulse for
protect the converter from elevated temperatures, an extension rod, also
the austenitic-martensitic microstructure without readjustment, c) load pulse
designed via FEM simulation for longitudinal oscillation at 20 kHz with
for the austenitic-martensitic microstructure after readjustment.
an amplitude transformation ratio of 1:1, was inserted between the
booster and the specimen. Preliminary investigations were carried out to
successive adjustments of the control parameters (Fig. 4c). In the first
assure a symmetrical temperature distribution along the specimen [50],
case, i.e. for a fully austenitic sample, the desired amplitude was reached
such that a symmetrical stress state could be assured with the maximum
within a few ms, remained constant during the stationary phase of the
stress in the center of the gauge length. Before fatigue testing, the
pulse and dropped rapidly at the end of the pulse. Within the pulse
specimen was heated to 300 ◦ C within 30 min and then kept at constant
sequence, perfectly sinusoidal vibrations can be seen in the changed
temperature for further 30 min. In the fatigue tests at elevated tem­
scale of the time axis (detail view in Fig. 4a). In the stationary phase of
perature, the displacement measurement took place at the lower end of
this pulse, a constant displacement amplitude of 9.5 µm was achieved. In
the extension rod by using the laser vibrometer. Since the fatigue
contrast, the second exemplarily shown pulse showed a change in pulse
specimen are symmetrical, the same displacement occurred at the lower
shape due to deformation-induced α′ -martensite formation during cyclic
end of the specimen as at the lower end of the extension rod (comp.
loading (Fig. 4b) which resulted in a considerable variation of the
Fig. 1a) [51]. The accuracy of this method was confirmed in preliminary
oscillation amplitude in the stationary load period. Consequently, con­
tests with maximum deviations less than 2%. Throughout each load
stant amplitude fatigue testing wasn’t viable in this manner. After
pulse, continuous measurement of the power of the ultrasonic generator
adjusting the proportional and integral (PI) parameters of the load unit
and temperature in the specimen gauge length was performed. At AT,
control circuit, a nearly ideal pulse shape with constant displacement
both dissipated energy and temperature change were used to charac­
amplitude was recovered, even if further deformation-induced phase
terize the cyclic deformation behavior (see chapter 1). During fatigue
transformation occurred (Fig. 4c). Hence, in order to perform the VHCF
testing at 300 ◦ C only the dissipated energy was considered since the
test on metastable austenite completely with an USFT, namely from N =
temperature was kept constant via closed-loop-control. In selected fa­
0 up to the limiting number of cycles e.g. Nl = 2 × 109, a successive
tigue test at AT, short interruptions were realized at defined cycle
adjustment of the control parameters is essential, especially in the initial
numbers for quasi in-situ Feritscope™ measurements of the ferromag­
phase where pronounced α′ -martensite formation occurs [39]. The
netic α′ -martensite fraction. Feritscope™ measurements in tests at
recording of the exact number of load cycles per pulse is realized by
300 ◦ C were only carried out before and after the fatigue test.
means of a self-developed LabVIEW-based measuring program. Only
those load cycles were counted where at least 90% of the generator
3.2. Control and adjustment of load amplitude
amplitude is present. This guarantees that load cycles with insufficient
stress during the onset and decay processes are not counted [54].
To perform valid fatigue tests on an USFT system in pulse-pause
In addition to keeping the load amplitude constant, the deformation
mode, the following aspects must be considered: (i) a correct pulse
induced temperature change of the material during VHCF tests has to be
shape with short onset and decay times as well as a stable stationary
kept within allowable limits. Due to low thermal conductivity together
phase must be achieved; (ii) a constant displacement respectively stress
with pronounced local cyclic plasticity even in the VHCF regime
amplitude must be ensured throughout the complete fatigue test; (iii)
[38,42], austenitic stainless steels show a significant self-heating during
the pulse-pause ratio needs to be selected in such a way that specimen
high frequency loadings [8,11,55]. While pulse-pause ratios of 500 ms /
heating due to the high-frequency loading does not lead to excessive
3500 ms (corresponding to an effective frequency feff = 2500 Hz) are
temperatures. Since, compared to other materials, metastable austenitic
typically used for materials such as titanium alloys or martensitic steels,
stainless steels exhibit more significant self-heating during high fre­
this setting resulted in a specimen temperature of 195 ◦ C for the
quency fatigue combined with deformation induced phase trans­
investigated metastable austenitic steel within two load pulses at σa =
formation from fcc γ-austenite to the bcc α′ -martensite, those three
263 MPa [2,15,17,52,54]. In order to prevent such excessive tempera­
aspects are extremely important in case of ultrasonic fatigue testing of
ture increases in the VHCF tests on AISI 347 an initial pulse-pause ratio
this materials class [2,15].
of 40–60 ms / 2500 ms depending on the stress amplitude and the
Due to the highly transient material behavior of this material class
cooling performance had to be used. This resulted in a relatively low
during USFT, especially at ambient temperature, the tests presented very
effective frequency of feff = 320–400 Hz, which would lead to 58 days of
challenging conditions compared to stable metallic materials e.g.
testing to reach the limiting number of cycles Nl = 2 × 109. Because the
ferritic-martensitic steels [52]. Already a small increase of the stiffer
deformation-induced α′ -martensite formation strongly reduced the
α′ -martensitic phase in the softer austenite phase led to a pronounced
locally cyclic plasticity [2,56] and, therewith, the self-heating of the
change of the displacement amplitude as well as the pulse shape due to
material, a stepwise increase of the pulse time and a reduction of pause
reduced micro-plastic deformation and a change in the material reso­
time could be realized, resulting in higher effective frequencies up to
nance behavior [8,39,53]. Fig. 4 shows representative pulses with a
1650 Hz for tests at ambient temperature (see Figs. 6, 7 & 9) [6; 9]. The
constant pulse time of 60 ms for three different fatigue states; (i) at the
VHCF tests at 300 ◦ C required considerably smaller readjustments of the
beginning of the test where a fully austenitic microstructure was present
PI parameters due to the higher austenite stability at elevated temper­
(Fig. 4a), (ii) at N = 107 load cycles where the specimen had a mixed
ature. Furthermore, the cyclic loading could be started with a higher
austenitic / α′ -martensitic microstructure without adjusting the control
pulse-pause ratio of 60 ms / 1 500 ms (feff = 800 Hz), which resulted in
parameters (Fig. 4b) and (iii) at the same cycle number but after

5
T. Daniel et al. International Journal of Fatigue 156 (2022) 106632

small deviations in specimen temperature (max. 3% / max. 309 ◦ C). Also


at elevated temperature, the pulse time could be increased during cyclic
loading because cyclic hardening, albeit less pronounced, occurred at
elevated temperature, too (see Fig. 8)
The abovementioned changes in the pulse / pause ratio directly
influenced the estimated Edis defined as the integral of the generator
power over the pulse time:

Edis = Pdt (5)

Therefore, the dependence on the pulse time had to be taken into


Fig. 5. a) Beginning of an ultrasonic fatigue test with in-test readjustment of consideration in analyzing Edis and also ΔT. To illustrate this aspect,
the generator voltage and PI parameters as well as the transient response to Fig. 5a shows an example of the development of Edis in the early phase of
achieve the required displacement respectively stress amplitude; b) the stable the VHCF loading for two fatigue tests, performed with two different
shape of two pulses with different pulse times. pulse / pause ratios: (i) 40 ms / 2 500 ms and (ii) 60 ms / 2 500 ms. Due
to the different pulse times, there were different numbers of load cycles
per pulse for both experiments: (i) N ≈ 600 for 40 ms and (ii) N ≈ 1 000
for 60 ms. This led to different absolute values of Edis in both tests
despite the same load amplitude of 250 MPa, which corresponded with a
displacement amplitude of 9.5 µm.
As mentioned earlier, at the beginning of VHCF tests (N 〈1 0 4) an
adjustment of the control parameters was required. Due to slight dif­
ferences in the oscillation of individual specimens at the same control
settings, the first pulse of a fatigue test was deliberately started with a
low generator voltage to avoid overshoots in the stress amplitude (first
data points in Fig. 5a). To achieve the required stress amplitude, the
generator voltage was increased after the first load pulse. In the
following load period up to approximately 8 × 103 load cycles, a
decrease of the displacement amplitude as well as the Edis can be seen in
Fig. 6. Development of the in-situ measured data: maximum displacement
amplitude (sa), loading frequency (f), change in temperature during load pulse
both tests, which resulted from the initial increase of the specimen
(ΔTpulse) and maximum generator power (Pmax), the quasi in-situ magnetic temperature. At this point further in-test adjustment of PI parameters
Feritscope™ measured (ξ) as well as the post processing determinate dissipated was performed to keep the load pulse regular and constant. After N =
energy (Edis) during a single-step test with σa = 249 MPa at ambient 104 a stable oscillation was reached, as confirmed by the stable shape of
temperature.

Fig. 9. a) Fatigue life curves of the investigated metastable austenitic steel AISI
Fig. 7. a) Cyclic deformation behavior characterized by development of change 347 at ambient temperature (f = 20 kHz) and at T = 300 ◦ C (f = 20 kHz, 980 Hz
in temperature ΔTpulse during a pulse and b) the dissipated energy Edis versus & 20 Hz); b) development of the dissipate energy versus load cycles during five
load cycles during VHCF tests at ambient temperature in an USFT system. VHCF tests at ambient temperature with stress amplitudes of σa = 249 /
250 MPa.

Fig. 8. a) Development of the dissipated energy versus load cycles during VHCF tests with a test frequency of f = 20 kHz; b) development of plastic strain amplitude
during HCF tests with a test frequency of f = 20 Hz [43] at T = 300 ◦ C for the same batch of AISI 347.

6
T. Daniel et al. International Journal of Fatigue 156 (2022) 106632

load pulse #16 for a pulse time of 40 ms and load pulse #10 for a pulse While the specimen loaded with σa = 249 MPa reached the VHCF
time of 60 ms in the representative displacement-time graphs in Fig. 5b. regime without failure, even just slightly higher stress amplitudes led to
Note that this initial process only took up 0.0005% of the limiting specimen failure which generally occurred at cycle numbers below 106.
number of load cycles of the test series. Accordingly, the fatigue tests are The cyclic deformation behavior at different stress amplitudes (σa = 249
presented in the following from 104 load cycles onwards. to 283 MPa) was characterized by using ΔTpulse and Edis (see Fig. 7). All
tests were performed with the same pulse/pause ratio of 60 ms / 2500
4. Results and discussion ms, except σa = 249 MPa for N > 4 × 107, where a higher pulse/pause
ratio was possible after the cyclic hardening phase of the experiment
Using the methods outlined in chapter 3, ultrasonic fatigue tests were (see above). A direct correlation between ΔTpulse and Edis was clearly
performed on AISI 347 at ambient temperature with stress amplitudes in seen in all experiments. As shown in Fig. 6, the fatigue test at σa = 249
the range 243 MPa ≤ σa ≤ 283 MPa as well as at 300 ◦ C in the range 153 MPa showed, after stable behavior from about N = 106 load cycles,
MPa ≤ σa ≤ 192 MPa. To characterize the cyclic deformation behavior, cyclic hardening which led to a decrease in the temperature change
several data were obtained. Fig. 6 shows an example of a fatigue test associated with a reduction of the dissipated energy per pulse.
with σa = 249 MPa and the development of in-situ and quasi in-situ Increasing the stress amplitude by only 1 MPa to σa = 250 MPa led to an
determined data. The in-situ data were: maximum displacement entirely different cyclic deformation behavior (Fig. 7): In this case, after
amplitude (sa), loading frequency (f), change in temperature during an initially stable state, the material showed cyclic softening (N > 5 ×
each load pulse ΔTpulse and maximum generator power (Pmax) as well as 104) until specimen failure at Nf = 4 × 105. With further increase of the
the resulting dissipated energy (Edis). The quasi in-situ data involved the stress amplitude to σa = 283 MPa, the cyclic softening took place earlier
magnetic Feritscope™ signal, which was measured at defined cycle and was more pronounced.
numbers during the pause time. Due to closed-loop control of the temperature in the isothermal
The test started with a pulse-pause ratio of 60 ms / 2500 ms and a VHCF tests at T = 300 ◦ C only the dissipated energy offers the possibility
load frequency of 20.27 kHz. As already described in Section 2, the to characterize the cyclic deformation behavior. Fig. 8a shows the
specimen had at the beginning of the test a purely austenitic micro­ development of the dissipated energy during four fatigue tests, which
structure without α′ -martensite. After achieving the required loading were started with the same pulse-pause ratio, i.e. 60 ms /1500 ms. The
amplitude at 104 cycles, all measured values remained constant and the results showed a qualitatively similar cyclic deformation behavior as
material showed a stable state up to 106 load cycles. Starting from an observed at ambient temperature (Fig. 7). For σa = 157 MPa an initial
initial specimen temperature of 20 ◦ C, cyclic loading with σa = 249 MPa cyclic softening was followed by saturation between N = 105 to 106 load
during the pulse time of 60 ms increased the specimen temperature to cycles and subsequent cyclic hardening, until the limiting number of
43 ◦ C, which results in ΔTpulse of 23 K. In the pause time of 2500 ms the cycles was reached without specimen failure. Due to cyclic hardening,
specimen cooled down to the initial temperature before the next pulse the pulse-pause ratio was similarly increased as in the tests at ambient
followed. After 106 load cycles a strong decrease in ΔTpulse was detected, temperature. The longer pulse time resulted in a higher dissipated en­
which correlated to the cyclic hardening of the material. Alongside with ergy, which is seen in the abrupt increases of Edis at N = 4 × 107 and N =
this, formation of α′ -martensite was detected by Feritscope™ measure­ 2 × 108. After N = 109 load cycles, the cyclic deformation behavior
ments which resulted in the increasing strength of the austenitic/ stabilized until the limiting number of load cycles was reached.
martensitic microstructure and, hence, in cycling hardening [22,41]. The aforementioned ultrasonic fatigue tests at 20 kHz and 300 ◦ C
Because of the formation of α′ -martensite, also the amplitude evolution were compared with results from tests at 20 Hz and 300 ◦ C for the same
during the pulse changed, especially in the phase of the fatigue test with batch of AISI 347 using a servo-hydraulic test system [43,45]. In the
a high rate of α′ -martensite formation. Therefore, a permanent adjust­ latter case the stress strain hysteresis loops were recorded using data
ment of the PI parameters was performed as described in Section 3.2. It from continuous stress and strain measurement. Dissipated energy per
could be argued that, because α′ -martensite exhibits a higher damping load pulse Edis measured during USFT and εa,p measured at 20 Hz indi­
[53], the increase of the ferromagnetic content should lead to an in­ cated the same sequences in cyclic deformation depending on the
crease in the generator power required to maintain oscillation at con­ applied stress amplitude (compare Fig. 8a with 8b). Note that the two
stant displacement amplitude. However, a decrease in Pmax was specimens tested at f = 20 Hz and σa = 160 MPa showed very similar
observed during this phase. This can be explained by the fact, that the development of the plastic strain amplitude, with an increase from the
lower plastic deformation, as a result of cyclic hardening, induced less beginning up to N ~ 104, followed by a decrease, up to εa,p = 0% until
energy dissipation and, thus, less power from the generator was the load cycle limit (Nl = 107). A direct comparison of the data shown in
required. From these findings, it was concluded that a decrease of Fig. 8a and 8b clearly confirmed the relationship between the dissipated
generator power Pmax which is associated with a decrease of the dissi­ energy and the plastic strain amplitude or, in general, to the plastic
pated energy Edis indicates cyclic hardening in metastable austenitic deformation energy density. Furthermore, the comparison of the two-
stainless steels. Conversely, increasing Pmax and Edis were associated to test series with almost identical stress amplitudes showed good accor­
cyclic softening. In addition, the vibration frequency showed a reaction dance between the tests performed on an ultrasonic and a servo-
to the cyclic deformation behavior of the material. While it increased hydraulic system, respectively.
due to cyclic hardening (see Fig. 6), it decreased during cyclic softening. Fig. 9a presents S-N curves of the investigated AISI 347 at ambient
As stated above, the cyclic hardening and the resulting lower self- and elevated temperature. A total of 63 fatigue tests were performed, 19
heating of the specimen allowed an increase in the pulse-pause ratio. at ambient temperature at 20 kHz and 44 at 300 ◦ C at different test
For this reason, after N = 5 × 107 load cycles the pulse-pause ratio was frequencies. As expected and analogous to the results of quasi-static
increased to 68 ms / 1500 ms which resulted in feff2 = 870 Hz. The in­ loading, the fatigue strength decreased with increasing temperature.
crease of the pulse time led to a higher ΔTpulse, which explains the No failure occurred in the VHCF regime at both, ambient temperature
abrupt change of this value, as well as of the dissipated energy in Fig. 6. and 300 ◦ C. At ambient temperature, σa ≥ 250 MPa led to specimen
After N = 1.25 × 108 load cycles, the pulse-pause ratio was further failure only in the HCF regime (Nf 〈1 0 7). Because in the range between
increased to 72 ms / 800 ms, resulting in an effective frequency of feff3 = 246 MPa and 250 MPa both, failures and run-outs, were observed, seven
1650 Hz. Until the limiting number of cycles, a plateau in the measured additional tests were performed to further investigate this phenomenon.
parameters was reached, i.e. the α′ -martensite content assumed a stable The results confirmed a sharp transition from specimen failure at σa =
value of 2.4 FE-% which was measured at the end of the fatigue test. 250 MPa and reaching the load cycle limit at σa ≤ 249 MPa. Fig. 9b
Accordingly, ΔTpulse, Pmax und Edis showed no significant changes from shows the development of the dissipated energy during four exemplarily
N = 1.25 × 108 cycles until the end of the fatigue test. chosen tests in this stress amplitude regime. Note that the tests were

7
T. Daniel et al. International Journal of Fatigue 156 (2022) 106632

started with different pulse-pause ratios, resulting in different values of steel including characterization of the cyclic deformation behavior
dissipated energy per pulse sequence at the beginning, as explained based on measurements of specimen temperature and dissipated energy.
above (see Fig. 5). Two fatigue tests with σa = 249 MPa (dark grey and Fatigue tests on AISI 347 stainless steel were performed using these
black curve) showed cyclic hardening. The fatigue test with σa = 250 methods at both, AT and 300 ◦ C. For comparison, fatigue tests were
MPa (grey) and a pulse-pause ratio of 40 ms / 2500 ms initially dis­ performed on servohydraulic testing systems at 300 ◦ C and frequencies
played weak cyclic hardening before cyclic softening and subsequent of 20 Hz and 980 Hz. From this, the following conclusions are drawn:
specimen failure at Nf = 4 × 106 load cycles. Another test at σa = 250
MPa (lighter grey) and a pulse-pause ratio of 60 ms / 2500 ms indicated • Opposed to high-strength steels and further metallic materials like
only cyclic softening. The α′ -martensite content in FE-% measured by aluminum and titanium alloys, displacement-controlled VHCF tests
Feritscope™ after the end of the fatigue test is indicated in Fig. 9a. It can on metastable austenites using an ultrasonic fatigue testing system
be seen, that specimens loaded at AT which reached the limiting number imposes serious challenges in adjusting the closed loop control and
of load cycles reached distinctly higher α′ -martensite contents (1.3 ≤ ξ test parameters due to the phase transformation and pronounced
≤ 2.4 FE-%) than those, which had already failed in the HCF regime at self-heating of the material. The complex test procedure required
higher stress amplitudes (0.3 ≤ ξ ≤ 0.85 FE-%). The cyclic hardening extensive readjustments of several control parameters to keep the
due to the formation of α′ -martensite obviously reduced cyclic plastic loading amplitude constant.
deformation, and inhibited the formation and propagation of micro • Due to exceptionally strong self-heating of the material, ultrasonic
cracks [11,40,43]. Consequently, a true fatigue strength exists at σa = fatigue testing was only possible with rather small pulse/pause ra­
249 MPa, i.e. far above the endurance limit for 109 cycles of 200 MPa tios, which, even though using a loading frequency of f = 20 kHz, led
estimated by extrapolation of the S-N curve in the HCF regime. to long test times (Nl = 2 × 109 in ~ 25 days).
Notably, also at 300 ◦ C no fatigue failure occurred in the VHCF • The temperature change during a pulse sequence (ΔTpulse) and the
regime. Failure in the HCF regime occurred for σa > 163 MPa, while respective dissipated energy (Edis) were found to be a suitable means
lower stress amplitudes led to reaching the limiting number of cycles for detailed characterization of cyclic deformation behavior of
without failure. At a stress amplitude of σa = 163 MPa, both, a specimen metastable austenitic stainless steels in VHCF regime.
failure at Nf = 5⋅106 and a run-out were found. In the HCF regime, the S- • No specimen failure was detected in ultrasonic fatigue tests at
N curve showed a similar slope as at ambient temperature while, even at ambient temperature below stress amplitudes of 249 MPa due to
T = 300 ◦ C, cyclic hardening led to the formation of a horizontal S-N cyclic hardening associated with deformation induced formation of
curve up from N = 107 load cycles. Additionally, the HCF tests per­ α′ -martensite. At higher stress levels, cyclic softening occurred which
formed with a frequency of f = 20 Hz (19 fatigue tests) show a lower is increasingly pronounced with increasing amplitude and results in
number of cycles to failure than for f = 20 kHz (14 fatigue tests) as well failure in the HCF regime. The specimens that achieved the limiting
as tests performed on a servo-hydraulic high frequency system at f = number of cycles showed higher α′ -martensite contents (1.3 ≤ ξ ≤
980 Hz (11 fatigue tests). On the one hand, the different specimen ge­ 2.4 FE-%) compared to specimens that failed in the HCF regime (0.3
ometries have to be taken into account [43], on the other hand when the ≤ ξ ≤ 0.85 FE-%).
specimen heating is suppressed, an increase in test frequency and • Also at T = 300 ◦ C no VHCF failure was observed. The S-N curve
consequently the strain rate leads to reduced dislocation movement by assumed a horizontal shape at a stress amplitude of 163 MPa, which
increasing yield stress of the material, which ultimately results in a was also in this case associated with cyclic hardening. This complied
higher fatigue life [57]. with the fact that α′ -martensite was found in run-out samples despite
Due to the increased austenite stability at 300 ◦ C, underpinned by the an increased austenite stability at elevated temperatures.
results from monotonic (Fig. 2b) and cyclic loading in LCF regime • A comparison of the fatigue tests at 300 ◦ C with different frequencies
[2,45,47], the formation of α′ -martensite was not expected in VHCF tests showed a good qualitative correlation between the evolutions of the
at 300 ◦ C. However, all run-out specimens tested at this temperature plastic strain amplitude determined in tests at 20 Hz on a servohy­
exhibited small ferromagnetic α′ -martensite contents while in all failed draulic system with the dissipated energy during the single load
specimens no α′ -martensite (ξ = 0.0 FE-%) was detected. This effect did pulses applied in the USFT system.
not only occur in case of VHCF loadings at 20 kHz up to limiting cycle • The investigated AISI 347 steel showed a positive influence of test
number of 5 × 108 and beyond, but also in tests at 20 Hz, and 980 Hz frequency on the fatigue strength in the VHCF regime at T = 300 ◦ C,
[43,45], which were, due to lower test frequencies, stopped at Nl = i.e. at a test frequency of 20 kHz and 980 Hz a higher fatigue strength
107 and 5 × 108 load cycles, respectively. Note in this context that the was achieved than at f = 20 Hz.
run-outs loaded at 20 kHz had higher α′ -martensite contents of ξ =
0.6–0.9 FE-% than those loaded at f = 20 and 980 Hz which exhibited ξ
Declaration of Competing Interest
= 0.1–0.3 FE-%. Several studies on α′ -martensite formation as a function
of the test frequency showed that an increasing of the strain rate results
The authors declare that they have no known competing financial
in a higher number of shear bands and their interactions, which lead to a
interests or personal relationships that could have appeared to influence
higher α′ -martensite formation rate. However, this is only valid if
the work reported in this paper.
excessive specimen heating, as a result of high deformation rates, is
prevented, which causes an opposite effect [45,57,58].
Acknowledgement
In order to further characterize the microstructural deformation and
fatigue mechanisms of metastable austenitic stainless steels in the VHCF
The authors thank the Federal Ministry for Economic Affairs and
regime at ambient and elevated temperature, including planar and/or
wavy slip character, twinning, development of stacking faults, and for­ Energy (BMWi), Germany for the financial support.
mation of ε- and/or α′ -martensite, extensive microstructural in­
vestigations using high resolution electron microscopy would be References
required which will be performed in further research.
[1] Wöhler A. Über die Versuche zur Ermittlung der Festigkeit von Achsen, welche in
den Werkstätten der Niederschlesisch-Märkischen Eisenbahn zu Frankfurt a. d. O.
5. Summary and conclusions angestellt sind. Z. f. Bauwesen 1866;16:67–84.
[2] Smaga M, Boemke A, Daniel T, Skorupski R, Sorich A, Beck T. Fatigue Behavior of
Metastable Austenitic Stainless Steels in LCF, HCF and VHCF Regimes at Ambient
In the presented work, advanced experimental procedures were and Elevated Temperatures. Metals 2019;9. https://doi.org/10.3390/
developed for ultrasonic fatigue testing of metastable austenitic stainless MET9060704.

8
T. Daniel et al. International Journal of Fatigue 156 (2022) 106632

[3] Bathias C. There is no infinite fatigue life in metallic materials. Fatigue Fract Eng [29] Mayer P, Skorupski R, Smaga M, Eifler D, Aurich J. Deformation induced surface
Mater Struct 1999;22:559–65. https://doi.org/10.1046/j.1460-2695.1999.00183. hardening when turning metastable austenitic steel AISI 347 with different
x. cyrogenic cooling strategies. Proc CIRP 2014;14:101–6. https://doi.org/10.1016/j.
[4] Mughrabi H. On ’multi-stage’ fatigue life diagrams and the relevant life-controlling procir.2014.03.097.
mechanisms in ultrahigh-cycle fatigue. Fatigue Fract Eng Mater Struct 2002;25: [30] Eichelmann GH, Hull FC. The effect of composition in the temperature of
755–64. https://doi.org/10.1046/j.1460-2695.2002.00550.x. spontaneous transformation of austenite to martensite in 18–8 type stainless steel.
[5] Zimmermann M. Diversity of damage evolution during cyclic loading at very high Trans ASM 1952;45(45):77–104.
numbers of cycles. Int Mater Rev 2012;57:73–91. https://doi.org/10.1179/ [31] Angel T. Formation of martensite in austenitic stainless steels - Effects of
1743280411Y.0000000005. deformation, temperature and composition. J Iron Steel Inst 1954;177:165–74.
[6] Morgan JM, Milligan WWA. 1 kHz servohydraulic fatigue testing system. Proc HCF [32] Martin S, Fabrichnaya O, Rafaja D. Prediction of the local deformation mechanisms
Struct Mat 1997:305–12. in metastable austenitic steels from the local concentration of the main alloying
[7] Stanzl-Tschegg SE. Very high cycle fatigue measuring techniques. Int J Fatigue elements. Mater Lett 2015;159:484–8. https://doi.org/10.1016/j.
2014:2–17. https://doi.org/10.1016/j.ijfatigue.2012.11.016. matlet.2015.06.087.
[8] Daniel T, Boemke A, Smaga M, Beck T. Investigations of very high cycle fatigue [33] Talonen J. Effect of strain-induced α′ -martensite transformation on mechanical
behavior of metastable austenitic steels using servohydraulic and ultrasonic properties of metastable austenitic stainless steels. Dissertation Helsinki University
testings systems. Proc. ASME PVP Conf. 2018:PVP2018-84639. https://doi.org/ of Technology; 2007.
10.1115/PVP2018-84639. [34] Smaga M, Boemke A, Daniel T, Klein MW. Metastability and fatigue behavior of
[9] Berchtold M, Klopfer I. Fatigue testing at 1000Hz testing frequency. Procedia Struct austenitic stainless steels. MATEC Web Conf 2018;165. https://doi.org/10.1051/
Integrity 2019;18:532–7. https://doi.org/10.1016/j.prostr.2019.08.197. matecconf/201816504010.
[10] Bathias C. Piezoelectric fatigue testing machines and devices. Int J Fatigue 2006; [35] Hahnenberger F, Smaga M, Eifler D. Fatigue behavior and phase transformation in
28:1438–45. https://doi.org/10.1016/j.ijfatigue.2005.09.020. austenitic steels in the temperature range -60◦ C≤T≤25◦ C. Proc Eng 2011;10:
[11] Müller-Bollenhagen C. Verformungsinduzierte Martensitbildung bei mehrstufiger 625–30. https://doi.org/10.1016/j.proeng.2011.04.104.
Umformung und deren Nutzung zur Optimierung der HCF- und VHCF- [36] Chopra OK, Shack WJ. Effect of LWR Coolant Environments on the Fatigue Life of
Eigenschaften von austenitischem Edelstahlblech: Siegener werkstoffkundliche Reactor Materials. Final Report, NUREG/CR-6909; 2007.
Berichte. Dissertation Universität Siegen 2011;3. [37] Carstensen JV, Mayer H, Brondsted P. Very high cylce regime fatigue of thin walled
[12] Jost B, Klein M, Beck T, Eifler D. Temperature dependent cyclic deformation and tubes made from austenitic stainless steel. Fatigue Fract Eng Mater Struct 2002;25:
fatigue life of EN-GJS-600 (ASTM 80–55-06) ductile cast iron. Int J Fatigue 2017; 837–44. https://doi.org/10.1046/j.1460-2695.2002.00554.x.
96:102–13. https://doi.org/10.1016/j.ijfatigue.2016.11.010. [38] Grigorescu A, Hilgendorff P-M, Zimmermann M, Fritzen C-P, Christ H-J. Fatigue
[13] Koster M, Wagner D, Eifler D. Cyclic deformation behavior of a medium carbon behaviour of austenitic stainless steels in the VHCF regime. In: Christ H–J, editor.
steel in the VHCF regime. Proc Eng 2010;2:2189–97. https://doi.org/10.1016/j. Fatigue of Materials at Very High Numbers of Loading Cycles. Springer; 2018.
proeng.2010.03.235. p. 49–72.
[14] Heinz S, Eifler D. Crack initiation mechanisms of Ti6Al4V in the very high cycle [39] Boemke A, Smaga M, Beck T. Influence of surface morphology on the very high
fatigue regime. Int J Fatigue 2016;93:301–8. https://doi.org/10.1016/j. cycle fatigue behavior of metastable and stable austenitic Cr-Ni steels. MATEC Web
ijfatigue.2016.04.026. Conf 165 2018.. https://doi.org/10.1051/matecconf/201816520008.
[15] Daniel T, Smaga M, Beck T. Ermüdungsverhalten des metastabilen austenitischen [40] Grigorescu AC, Hilgendorff P-M, Zimmermann M, Fritzen C-P, Christ H-J. Cyclic
Stahls AISI 347 bei Raumtemperatur und 300 ◦ C im VHCF-Bereich. Tagungsband deformation behavior of austenitic Cr–Ni-steels in the VHCF regime: Part I -
Werkstoffprüfung 2019:81–6. Experimental study. Int J Fatigue 2016;93:250–60. https://doi.org/10.1016/j.
[16] Müller-Bollenhagen C, Zimmermann M, Christ H-J. Very high cycle fatigue ijfatigue.2016.05.005.
behaviour of austenitic stainless steel and the effect of strain-induced martensite. [41] Zimmermann M, Grigorescu A, Müller-Bollenhagen C, Christ H-J. Influence of
Int J Fatigue 2010;32:936–42. https://doi.org/10.1016/j.ijfatigue.2009.05.007. deformation-induced alpha prime martensite on the crack initiation mechanism in
[17] Ritz F, Beck T. Influence of mean stress and notches on the very high cycle fatigue a metastable austenitic steel in the HCF and VHCF regime. Proc 13th Int Conf on
behaviour and crack initiation of a low-pressure steam turbine steel. Fatigue Fract Fract 2013.
Eng Mater Struct 2017;11:1762–71. https://doi.org/10.1111/ffe.12666. [42] Müller-Bollenhagen C, Zimmermann M, Christ H-J. Adjusting the very high cycle
[18] Stanzl-Tschegg SE, Schönbauer B. Mechanisms of strain localization, crack fatigue properties of a metastable austenitic stainless steel by means of the
initiation and fracture of polycrystalline copper in the VHCF regime. Int J Fatigue martensite content. Proc Eng 2010;2:1663–72. https://doi.org/10.1016/j.
2010;32:886–93. https://doi.org/10.1016/j.ijfatigue.2009.03.016. proeng.2010.03.179.
[19] Mayer H, Fitzka M, Schuller R. Constant and variable amplitude ultrasonic fatigue [43] Smaga M, Sorich A, Eifler D, Beck T. Very high cycle fatigue behavior of metastable
of 2024–T351 aluminium alloy at different load ratios. Ultrasonics 2013;53: austenitic steel X6CrNiNb1810 at 300◦ C. Proc 7th Int Conf VHCF 2017:249–54.
1425–32. https://doi.org/10.1016/j.ultras.2013.02.012. [44] Wagner D, Cavalieri FJ, Bathias C, Ranc C. Ultrasonic fatigue tests at high
[20] Stanzl-Tschegg SE, Mughrabi H, Schönbauer B. Life time and cyclic slip of copper temperature on an austenitic steel. Propul Power Res 2012;1:29–35. https://doi.
in the VHCF regime. Int J Fatigue 2007;29:2050–9. https://doi.org/10.1016/j. org/10.1016/j.jppr.2012.10.008.
ijfatigue.2007.03.010. [45] Sorich A. Charakterisierung des Verformungs- und Umwandlungsverhaltens des
[21] Piotrowski A, Eifler D. Bewertung zyklischer Verformungsvorgänge metallischer zyklisch beanspruchten metastabilen austenitischen Stahls X6CrNiNb1810:
Werkstoffe mit Hilfe mechanischer, thermometrischer und elektrischer Werkstoffkundliche Berichte. Dissertation Technische Universität Kaiserslautern
Meßverfahren: Characterization of cyclic deformation behaviour by mechanical, 2017;40.
thermometrical and electrical methods. Materialwissenschaften und [46] Deutsches Institu für Normung: DIN. DIN EN 10088-1: Nichtrostende Stähle - Teil
Werkstofftechnik 1995;26(26):121–7. 1: Verzeichnis der nichtrostenden Stähle Deutsche Fassung EN 10088-1:2014;
[22] Xue H, Wagner D, Ranc N, Bayraktar E. Thermographic analysis in ultrasonic 2014.
fatigue tests. Fatigue Fract Eng Mater Struct 2006;29:573–80. https://doi.org/ [47] Sorich A, Smaga M, Eifler D. Influence of loading condition on the cyclic
10.1111/j.1460-2695.2006.01024.x. deformation behavior and phase transformation of the austenitic stainless steel
[23] Ranc N, Wagner D, Paris PC. Study of thermal effects associated with crack AISI 347 at ambient temperature and 300 ◦ C. 7th Int. Conf. on LCF; 2013.
propagation during very high cycle fatigue tests. Acta Mater 2008;56:4012–21. [48] Veile I, Szielasko K, Tschuncky R, Weber F, Thieltges S, Smaga M et al.
https://doi.org/10.1016/j.actamat.2008.04.023. Zustandsüberwachungssystem auf der Basis von elektromagnetisch angeregten
[24] Smaga M, Walther F, Eifler D. Deformation-induced martensitic transformation in Ultraschall-Wandlern: Phase 2. Abschlussbericht Reaktorsicherheitsforschung Nr.
metastable austenitic steels. Mater Sci Eng, A 2008;7:394–7. https://doi.org/ 1501554; 2019.
10.1016/j.msea.2006.09.140. [49] Cavalieri F, Bathias C, Ranc N, Cardona A, Riss J. Ultrasonic fatigue analysis on an
[25] Koster M, Nutz H, Freeden W, Eifler D. Measuring techniques for the very high austenitic steel at high temperature. Mecanica Computacional 2008;27:1205–24.
cycle fatigue behaviour of high strength steel at ultrasonic frequencies. Int J Mat [50] Schopf T, Stumpfrock L, Daniel T, Smaga M, Beck T, Rudolph J. Untersuchungen
Res 2012;103:106–12. https://doi.org/10.3139/146.110623. zum Ermüdungsverhalten austenitischer Werkstoffe und deren
[26] Daniel T, Smaga M, Beck T, Schopf T, Stumpfrock L, Weihe S et al. Investigation of Schweißverbindungen für RDB- und Kerneinbauten im HCF- und VHCF-Bereich:
the very high cycle fatigue (VHCF) behavior of austenitic stainless steels and their Phase I. Abschlussbericht Reaktorsicherheitsforschung Nr. 1501548; 2020.
welds for reactor internals at ambient temperature and 300 ◦ C. Proc. ASME PVP [51] Kumar A, Torbet CJ, Jones JW, Pollock TM. Nonlinear ultrasonics for in situ
Conf. 2020:PVP2020-21460. https://doi.org/10.1115/PVP2020-21460. damage detection during high frequency fatigue. J Appl Phys 2009;106. https://
[27] Man J, Smaga M, Kubena I, Eifler D, Polak J. Effect of metallurgical variables on doi.org/10.1063/1.3169520.
the austenite stability in fatigued AISI 304 type steels. Eng Fract Mech 2017;185: [52] Beck T, Kovacs SA, Ritz F. VHCF behavior and work hardening of a ferritic-
139–59. https://doi.org/10.1016/j.engfracmech.2017.04.041. martensitic steel at high mean stresses. Key Eng Mater 2015;664:246–54. https://
[28] Smaga M, Eifler D. Fatigue Life Calculation of Metastable Austenitic Stainless Steels doi.org/10.4028/www.scientific.net/KEM.664.246.
on the Basis of Magnetic Measurements. MP Mater Test 2009;51:370–5. https:// [53] Talonen J, Hänninen H. Damping Properties of Austenitic Stainless Steels
doi.org/10.3139/120.110046. Containing Strain-Induced Martensite. Metall Mater Trans A 2004;35A:
2401–20406. https://doi.org/10.1007/s11661-006-0220-x.

9
T. Daniel et al. International Journal of Fatigue 156 (2022) 106632

[54] Ritz F. Ermüdungs- und Schädigungsverhalten von X10CrNiMoV12-2-2 im VHCF- Fatigue Loading at Ambient and Lower Temperatures. Adv Eng Mater 2012;14(10):
Bereich unter dem Einfluss von Mittelspannungen und Kerben X10CrNiMoV12-2-2 853–8. https://doi.org/10.1002/adem.201100341.
im VHCF-Bereich unter dem Einfluss von Mittelspannungen und Kerben: [57] Talonen J, Nenonen P, Pape G, Hänninen H. Effect of Strain Rate on the Strain-
Werkstoffkundliche Berichte. Dissertation Technische Universität Kaiserslautern Induced gamma arrow alpha prime-Martensite Transformation and Mechanical
2019;45. Properties of Austenitic Stainless Steels. Metall Mater Trans A 2005;36A:421–32.
[55] Klein MW, Smaga M, Beck T. Surface Morphology and Its Influence on Cyclic [58] Hecker SS, Stout MG, Staudhammer KP, Smith JL. Effects of Strain State and Strain
Deformation Behavior of High-Mn TWIP Steel. Metals 8 2018;10:832. https://doi. Rate on Deformation-Induced Transformation in 304 Stainless Steel: Part I.
org/10.3390/met8100832. Magnetic Measurements and Mechanical Behavior. MTA 1982;13A:619–26.
[56] Hahnenberger F, Smaga M, Eifler D. Influence of γ-α′ -Phase Transformation in https://doi.org/10.1007/BF02644427.
Metastable Austenitic Steels on the Mechanical Behavior During Tensile and

10

You might also like