Está en la página 1de 27

INVITED REVIEW ARTICLE

doi: 10.1111/ffe.12365

Recent developments in ultrasonic fatigue


H. MAYER
Institute of Physics and Materials Science, BOKU, Peter-Jordan-Str. 82, A-1190 Vienna, Austria
Received Date: 21 April 2015; Accepted Date: 28 September 2015

A B S T R A C T The recently increased interest in very high cycle fatigue properties of materials has led to

extended use and further development of the ultrasonic fatigue testing technique. Specimens
are stimulated to resonance vibrations at ultrasonic frequency, where the high frequency allows collecting lifetime data of up to 1010 cycles and measuring crack propagation rates down
to 1012 m per cycle within reasonable testing times. New capabilities and methods of ultrasonic testing and outstanding results obtained since the year 1999 are reviewed. Ultrasonic
tests at load ratios other than R = 1, variable amplitude tests, cyclic torsion tests and
methods for in situ observation of fatigue damage are described. Advances in testing at very
high temperatures or in corrosive environments and experiments with other than bulk metallic materials are summarized. Fundamental studies with copper and duplex steel became
possible and allowed new insights into the process of very high cycle fatigue damage. Higher
cyclic strength of mild steels measured at ultrasonic frequency because of plastic strain rate
effects are described. High-strength steels and high-alloy steels are less prone to frequency
inuences. Environmental effects that can lead to prolonged lifetimes in some aluminium
alloys and possible frequency effects in titanium and nickel and their alloys are reviewed.
Keywords

NOMENCLATURE

A1
A2

c
pl
N
R

=
=
=
=
=
=
=
=
=
=

high frequency; ultrasonic fatigue; very high cycle fatigue.

logarithmic amplitude of rst-order harmonic


logarithmic amplitude of second-order harmonic
nonlinearity parameter
specic heat capacity
plastic strain amplitude
number of cycles
load ratio
mass density
temperature
stress amplitude

ABBREVIATION

VA = variable amplitude
CA = constant amplitude

INTRODUCTION

Load-bearing components of vehicles, including engine


components, axles, wheels and chassis components, can
be loaded with more than 108 cycles during service. Railway components, bridges, offshore structures and components of wind turbines need to withstand more than 109
cycles. The requirements of safe operation without failure
and economically efcient use of materials require knowlCorrespondence: H. Mayer. E-mail: herwig.mayer@boku.ac.at

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

edge about the fatigue behaviour in the very high cycle


fatigue (VHCF), that is, in the regime beyond the classical
fatigue limit. VHCF strength of materials and mechanisms
leading to fracture can be signicantly different from the
high cycle fatigue (HCF) regime. Fatigue crack initiation
in high-strength steels, for example, preferentially occurs
at the surface in the HCF regime. In the VHCF regime,
internal inclusions act as preferential crack initiation locations and cause a further drop of the SN curve and the
elimination of a fatigue limit.1,2 Most aluminium alloys
do not show a fatigue limit. Surface slip or fractured

H. MAYER

constituent particles are typical sources of fatigue cracks


that can lead to failures up into the VHCF regime.3,4 In
some cast aluminium alloys, however, gas or shrinkage porosities are favourable starting places of fatigue cracks, and
fatigue limits are found at stress amplitudes that are too low
to propagate these cracks to rupture.5,6 Some titanium alloys show SN curves that are approximately parallel to
the abscissa in the VHCF regime,7 whereas the SN curves
of others decrease by 300 MPa when lifetimes increase
from the HCF to the VHCF regime.8 Ti6Al4V hardly
shows VHCF failures at low stress ratios, whereas strongly
decreasing SN curves are found beyond the HCF regime if
stressed at high stress ratios.9,10 These examples show that
the VHCF properties of materials can hardly be extrapolated from HCF data. A survey about specics of VHCF
and the differences to HCF can be found for materials containing secondary-phase crack-initiating particles in Ref.
[11]. Crack-initiating mechanisms other than inclusions
and progress of fatigue damage leading to VHCF failures
for virtually defect-free materials are reviewed in Ref. [12].
Scientic interest and the need for sound material characterization have put forward the interest to test materials at
very high numbers of load cycles, in the regime where they
are actually loaded in technical components. However, testing in the VHCF regime rst of all requires techniques that
work at high frequency. Testing of a single specimen up to
109 cycles using conventional servo-hydraulic equipment
working at 50 Hz, for example, would take 8 months. Considering the typically greater number of specimens required
for characterizing a material, accelerated testing methods are
needed for experiments in the VHCF regime. Servohydraulic testing machines operating at 1000 Hz, resonant
and forced-vibration machines, rotating bending equipment
and ultrasonic equipment are used in VHCF studies. Evaluating the advantages and disadvantages of these techniques,
ultrasonic testing has proved to be the most suitable method
for investigations in the VHCF regime.13 Ultrasonic testing
works at very high frequency, it is practically maintenance
free, it can be easily installed as it does not require a cooling
circuit and, nally, it is by a factor of about 10 000 more
energy efcient than a servo-hydraulic testing system.
Moreover, the technique has been employed since about
60 years, and thus, a lot of experimental setups and applications are already described in literature.
The history of ultrasonic fatigue testing starts in about
1950,14 and older literature is reviewed in several papers.1518 The rst ultrasonic fatigue experiments were
endurance tests under fully reversed loading conditions.19 Ultrasonic fatigue crack growth and threshold
measurements started in about 1973.20 Variable amplitude (VA) fatigue tests,21 ultrasonic frequency cycling
with superimposed static tension forces (load ratios
R -1)22 and cycling with superimposed torque loads
to generate mixed-mode loading22,23 were further steps

to extend the capabilities of this testing technique.


Ultrasonic torsion fatigue experiments with ceramic and
metallic materials24,25 have been performed with and
without superimposed static axial preloads. Fatigue lifetime and fatigue crack growth have been studied in ultrasonic experiments at low and elevated temperatures, in
inert and corrosive environments, mainly with metallic
materials but also with ceramics and composites.18
The objective of this paper is to review recent developments in ultrasonic fatigue testing. Interesting new
ultrasonic testing procedures will be reviewed, such as
testing at load ratios other than R = 1, VA testing, cyclic
torsion testing and methods, for in situ observation of
fatigue damage. Advances of the method to perform
investigations at very high temperatures, in corrosive
environments and testing other than bulk metallic
materials, will be described. Fundamental fatigue investigations in the VHCF regime with copper and duplex
steel that became feasible through ultrasonic testing are
included. Finally, investigations on frequency inuences,
which is probably the most important question involved
with high frequency testing, will be reviewed. The
reviewed time period dates back until 1999. Earlier
ultrasonic literature and investigations about frequency
inuences are included in earlier work of the author.18

ULTRASONIC FATIGUE TESTING METHOD

Different ultrasonic fatigue testing equipment exists from


commercial manufacturers and research institutions.
While there are differences in process control, obtainable
accuracy and possible testing applications, they share the
same basic mechanical principle for the load train. In the
following, the equipment developed at the Institute of
Physics and Materials Science at University of Natural
Resources and Life Sciences, Vienna (Physics BOKU
Vienna), is described. Figure 1 shows a picture of the
ultrasonic equipment from Physics BOKU Vienna.
Experimental procedure
Figure 2 shows the mechanical components of the ultrasonic load train for performing ultrasonic fatigue tests at
load ratios R 1. Additionally, the calculated strain
and vibration amplitudes along the load train using nite
element method are shown.
In an ultrasonic fatigue test, appropriately designed
specimens are stimulated to resonance vibrations at frequencies close to 20 kHz. Vibrations are generated by a
piezoelectric ultrasonic converter and are magnied with
an amplifying horn. Both ends of the specimen vibrate
with maximum displacement amplitude in opposite directions, and the movement of one end of the specimen

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

RECENT DEVELOPMENTS IN ULTRASONIC FATIGUE

frames, as shown in Fig. 1. More details about the design


of ultrasonic components can be found elsewhere.18
The working principle of the ultrasonic equipment is
shown in Fig. 3. The displacement amplitude measured
at one end of the specimen is proportional to the strain
amplitude in its centre and can therefore be used for control purposes in the test generator. The linear proportionality between displacement and strain amplitude is
calibrated prior to actual tests using strain gauges along
with a strain gauge conditioner.
The tests are controlled by a test generator
performing several tasks:

Fig. 1 Ultrasonic fatigue testing equipment used to perform fatigue


tests at load ratio R 1: servo-hydraulic load frame with built-in
ultrasonic load train (left side of the gure) and electronic equipment
(right side of the gure). Additional computer control is used in variable amplitude tests (not shown).

is measured with a vibration gauge. A vibration node with


maximum strain amplitude is formed in the centre of the
specimen, as indicated in the nite element analysis simulation shown in Fig. 2. The nite element simulation in
AUTODESK Simulation Mechanical is based on the linear elastic model, using the modal vibration analysis type.
No degrees of freedom were constrained; all nodes are
allowed to displace in all directions. The components of
the load train feature the exact resonance lengths, along
with their correct mechanical properties.
In fully reversed fatigue tests (load ratio R = 1), one end
of the specimen is allowed to vibrate freely. In superimposed
loading tests (R 1), the specimen is mounted on both
sides into the load train containing rods of length of half
or full wavelength. Mounting devices (Mounting 1 and
Mounting 2 in Fig. 2) serve to apply static tensile or
compressive forces at vibration nodes without damping
the resonance vibration. Superimposed loads may be
generated with electromechanical or servo-hydraulic load

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

1 Setting the desired vibration amplitude, pulse and


pause duration: The vibration amplitude is selected
after calibration. Pulse and pause lengths are chosen
appropriately to avoid heating of the specimen.
Additional forced air cooling is applied if necessary.
2 Control of vibration amplitude: Cyclic hardening or
softening of the material or the initiation and growth
of cracks, for example, change the power requirements
to excite the specimen to the desired amplitude. The
signal of the vibration gauge is processed in a
closed-loop control circuit, which guarantees that the
pre-selected amplitude and the actual vibration
amplitude coincide accurately within 1%, by comparing the pre-selected (nominal) amplitude with the
measured amplitude and adjusting the output of the
power ampliers accordingly. Additionally, the power
at the beginning of a pulse is increased to reach rapidly
the nominal vibration amplitude.
3 Control of resonance frequency: The excitation frequency must match the actual resonance frequency of
the ultrasonic load train including the specimen. A
voltage-controlled oscillator and a phase-locked loop circuit are used for this purpose, where the phase shift of the
ultrasonic power signal and the specimens displacement
signal are kept constant. This permits the cycling frequency and the resonance frequency to coincide better
than 0.1 Hz. Additionally, the cycling frequency is
monitored and used to detect specimen failure. Initiation
of a crack increases the specimens compliance and
reduces resonance frequency. Frequency limits serve to
stop the experiment automatically.
Power output to the ultrasonic converter is provided
by power ampliers with a maximum output power of
600 W. A power control can be used to adjust output
power and output voltage. An oscilloscope serves to display various signals (Fig. 3). Computer-based data acquisition and control applications allow external control of
the ultrasonic test. It serves for measuring and classifying
all load cycles, analyzing the vibration signal and utilizing
external command signal options. A more detailed

H. MAYER

Fig. 2 Load train used to perform ultrasonic fatigue tests with preloads (load ratio R 1): Mechanical components, strain and displacement
amplitudes along the load train are shown; for testing at load ratio R = 1, both Mounting 1 and Mounting 2 are omitted.

Fig. 3 Working principle of ultrasonic fatigue testing equipment.26

description of the electrical components and of the computer system may be found in Ref. [26].
Fatigue data measured at different load ratios
Ultrasonic fatigue tests with superimposed mean load
are used to study the inuences of load ratio on fatigue
lifetimes, fatigue crack initiation and crack propagation.
Karsch et al.27 investigated fatigue lifetimes of highcarbon chromium bearing steel 100Cr6 (52100) in
bainitic condition (tensile strength 2480 MPa) and in
tempered martensitic condition (tensile strength
2150 MPa). Experiments were performed at load ratio
R = 1 and R = 0.1 in the HCF and VHCF regimes.
The VHCF strength in martensitic condition was about

20% higher than the strength in bainitic condition for


fully reversed cycling. Comparable VHCF strength for
both material conditions was found in experiments at load
ratio R = 0.1. This indicates a stronger mean stress
sensitivity of the martensitic microstructure in the VHCF
regime. Experiments with hydrogen-charged specimens
showed a strong deterioration of cyclic properties for
both microstructures at both load ratios.
Sander et al.28 investigated the effects of mean stress
on the cyclic properties of low-alloy high-strength steel
34CrNiMo6 (tensile strength 1200 MPa). Experiments
were performed in HCF and VHCF regimes at load ratio
R = 1 and R = 0. Internal crack initiation at inclusions
was found to be the most frequent source of fatigue
cracks for both load ratios. Failures above 108 cycles

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

RECENT DEVELOPMENTS IN ULTRASONIC FATIGUE

occurred after loading at load ratio R = 0 but not at


R = 1. VA loads were used to produce arrest marks on
the fracture surface close to the internal crack-initiating
inclusion. With this, the very slow growth rates of initial
cracks in specimens that failed in the VHCF regime
could be calculated.
Kovac et al.29 studied martensitic 12%Cr turbine steel
(X10CrNiMoV12-2-2, tensile strength 1000 MPa) at
load ratios between R = 1 and R = 0.7. Change of crack
initiation from the surface in the HCF regime to the subsurface in the VHCF regime was found for all load ratios.
No failures occurred above 108 cycles at load ratio
R = 1, whereas specimens failed even above 109 cycles at
load ratios R = 0.5 and R = 0.7. Optically dark areas30,31 were
found at internal crack-initiating inclusions after VHCF failure at R = 1. The most interesting is that optically dark areas
could not be found at load ratios greater or equal R = 0.1,
which was reported for the rst time in this investigation.
In the authors laboratory, the cyclic properties of the
aluminium alloy 2024-T351 were investigated at load
ratios R = 1, R = 0.1 and R = 0.5. Figure 4 shows the
measured SN data at the three load ratios. A clear inuence of load ratio is visible. That is, with the increasing
load ratio, the SN curves are shifted towards lower cyclic stress amplitudes. Slope exponents of approximation
lines increase with the increasing numbers of cycles for
all load ratios. Failures can still occur above 5 109 cycles
(R = 1 and R = 0.1) or 1010 cycles (R = 0.5), and no

fatigue limit is found. Fatigue cracks leading to failures


above 109 cycles are initiated at the surface or slightly
below at broken Al7Cu2(Fe, Mn) particles or at
agglomerations of fractured particles. Cracks at fractured
secondary-phase particles were also found in runout specimens. Figure 5 shows the surface of a specimen that was
loaded with 1.6 1010 cycles at R = 1 without failure.
Survival of very high numbers of cycles is therefore
associated with a non-propagating condition of short
cracks rather than with a non-initiating condition.
Liu et al.10 measured fatigue lifetimes of Ti6Al4V at
load ratios in the range from R = 1 to R = 0.5. A strong
inuence of load ratio on the slope of the SN curves
and on the mechanism of fatigue crack initiation is found.
At load ratios R = 1 and R = 0.5, the SN curve shows
a horizontal asymptote and a fatigue limit. Specimen failures occur solely below about 107 cycles, and fatigue
cracks are initiated exclusively at the surface. In contrast,
stepwise SN curves are found for load ratios R = 0.1,
0.1 and 0.5. The SN curves show a sharp decrease above
approximately 107 cycles with crack initiation at the
surface as well as in the interior.

Fig. 4 Constant amplitude ultrasonic fatigue data of the aluminium


alloy 2024-T351 for load ratio R = 1 (squares), R = 0.1 (circles)
and R = 0.5 (triangles), respectively.4

Fig. 5 Surface of a 2024-T351 specimen that survived 1.6 10 cycles


at /2 = 100 MPa and load ratio R = 1: fatigue cracks starting at a
fractured constituent particle. Crack tips are indicated with arrows.4

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

Fatigue testing of thin sheets


Ultrasonic fatigue experiments are typically performed
using specimens of appropriate length to vibrate in
resonance at ultrasonic frequency. Thin sheets, however,
cannot be tested in that way because of buckling
problems and problems with the sound transfer.
Liu et al.32 developed a method to test thin sheets at
ultrasonic frequency. Rather than vibrating in resonance,
the sheet specimens are xed to an hourglass-shaped carrier specimen and are forced to joint resonance vibration.
Additionally, the carrier specimen and the mounted sheet

10

H. MAYER

specimen are preloaded with a tensile force. With this


method, single-crystal Ni-based superalloy sheets could
be tested at a positive load ratio.
An alternative method to test thin sheets at positive
load ratios has been developed at the authors laboratory.33,34 The sheets are xed to a dumbbell-shaped carrier specimen. The static preload is realised through
bending the carrier out of vertical alignment prior to xing the sheet specimen and re-establishing vertical alignment. Cyclic and static strains for the consecutive test are
measured with strain gauges on both sides of the thin
sheet specimen. This serves to guarantee that no undesirable bending occurs. This method has been successfully
applied to investigate cyclic properties of thin sheets of
nitrided high-strength maraging steel. Experiments in
the range between 107 and 109 cycles to failure at load ratio R = 0.1 showed internal crack initiation in all specimens. TiN inclusion in Ti containing maraging steel
sheets was found to be preferential sources of VHCF
failures.33,34 Elimination of Ti content and increase of
Co content led to preferential crack initiation at Al2O3
inclusions.34 Oxide inclusions were found to be less
harmful than TiN inclusions, and the VHCF strength
was therefore improved.
Investigations of new materials and bonds, reversed
bending testing
Most ultrasonic tests are performed with bulk metallic
materials. Zettl et al.35,36 used the ultrasonic fatigue testing method to investigate the cyclic properties of foams.
He tested four aluminium foams with a mean density of
500 kg m3. Cylindrical rods with a continuous surface
layer and a closed-cell foam structure inside were studied.
All foams showed a pronounced fatigue limit, and no failures occurred above 1.4 107 cycles. Fatigue cracks are
initiated in the cell structure at pre-existing cracks or
holes followed by fracture of the surface layer. Below
the fatigue limit, fatigue cracks can be initiated, but they
are trapped at nodes of cells.
Cremer et al.37 investigated the fatigue properties of
welds in AlMgSi1 aluminium alloy (EN AW-6082).
Hourglass-shaped ultrasonic specimens with a welldened maximum of the stress amplitude in their centres
were used. The specimens were cut out of the welded
plate with their centres in certain distances to the weld.
Thus, the cyclic properties of the weld as well as of the
heat-affected zone could be measured. The seam weld
was found to be the weakest region, where incomplete
fusion or gas pores are preferential places for crack
initiation.
Xue et al.38 developed a method to perform threepoint bending tests at ultrasonic frequency. They used a
longitudinally vibrating load train to stimulate a

rectangular-shaped bar specimen to bending vibrations.


A static force was superimposed to the vibration. This
served to investigate the cyclic bending fatigue properties
of TiAl at a load ratio of R = 0.4. Lifetimes could be measured in the regime between 106 and 1010 cycles, which
demonstrated the feasibility and applicability of the
developed method.
Backe et al.39 designed a load train to perform ultrasonic three-point bending fatigue tests with carbon
bre-reinforced polymer. Their experiments demonstrated the applicability of the ultrasonic testing method
for testing polymer material. A rectangular bar was stimulated to transversal resonance vibrations superimposed
to a static bending load, which led to load ratios between
0.29 and 0.49. The experiments were performed with
infrared control of the specimens temperature and
measurement of specimens movement using laser
vibrometry. Delamination of samples was used as failure
criterion, and lifetimes were measured up to 109 cycles.

FATIGUE TESTING IN DIFFERENT


ENVIRONMENTS

The fatigue properties of materials are strongly affected


by the temperature and by possible chemical processes
initiated by the environment. Even humid air as the most
frequently used environment to perform fatigue tests
must be considered as chemically active for several materials. It has already been attempted in earlier ultrasonic
literature to perform measurements under different
environmental conditions. Corrosion fatigue studies40
and studies in liquid nitrogen41 and at elevated temperatures42 have been performed several decades ago. In
recent works, it has been pursued to extend further
ultrasonic testing to very high temperatures under wellcontrolled environmental conditions and using advanced
experimental setups.

Testing at very high temperatures


Yi et al.43 performed ultrasonic fatigue tests with a
Ni-based superalloy single crystal at 1000 C. The
specimens were heated with an induction coil using an
infrared pyrometer to monitor their temperature. SN
data in the regime from 5 105 to 5 108 cycles could
be generated in this way. Failures were found above
108 cycles. The crack was initiated preferentially subsurface at carbides. The cracks propagated on octahedral
slip planes in the ultrasonic tests, which are different
from crack growth perpendicular to the maximum
tensile stress found in high-temperature tests at lower
frequencies.

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

RECENT DEVELOPMENTS IN ULTRASONIC FATIGUE

Furuya et al.44 performed ultrasonic fatigue tests with a


single-crystal superalloy at 1000 C and with heatresistant steel 12Cr-2W at 650 C. Induction heating of
the specimen and control of temperature via infrared
thermography are used. Single crystals were tested at load
ratio R = 0 in the regime between 107 and 5 108 cycles.
Measured lifetimes agreed closely with the data obtained
from conventional fatigue tests. Tests of 12Cr-2W at load
ratio R = 1 similarly showed good coincidence to measurements performed at lower frequencies. Both materials
do not show a fatigue limit, indicating the need to perform
fatigue tests to obtain resilient VHCF data.
Inuence of humidity
Fatigue crack growth tests with the aluminium alloys
2024-T3 and 7075-OA (i.e. 7075 in overaged condition)
have been performed by Holper et al.45,46 with servohydraulic equipment at 20 Hz and ultrasonic equipment
at 20 kHz. Experiments were performed in ambient air

and in vacuum at load ratios R = 1, R = 0.1 and R = 0.5.


The two materials had been chosen to represent aluminium alloys with different slip characteristics. 2024-T3
contains shearable precipitates that promote a
crystallographic crack path. Precipitates in 7075-OA are
non-shearable, which favours a more homogeneous
deformation leading to a more planar fracture surface.
Figure 6 shows the measured crack propagation rates
for both materials in vacuum (a, b) and in ambient air
(c, d).45 Near-threshold fatigue crack growth rates in
vacuum agree closely at 20 Hz and 20 kHz. Threshold
stress intensities at a limiting growth rate of 1010 m
per cycle coincide within the range of scatter at the three
load ratios for both materials. Mean growth rates of
1012 m per cycle could be measured testing 7075-OA
in vacuum at ultrasonic frequency. Such low growth rates
are hardly accessible in conventional fracture mechanics
fatigue tests, which underlines a strong advantage of ultrasonic testing. In ambient air, humidity causes chemical
processes with the newly formed surfaces at the crack tip

(a)

(c)

(b)

(d)

Fig. 6 Fatigue crack growth at load ratios R = 1, R = 0.05 and R = 0.5 in 2024-T3 in vacuum (a), in 7075-OA in vacuum (b), in 2024-T3 in
ambient air (c) and in 7075-OA in ambient air (d). Solid symbols refer to servo-hydraulic experiments at 20 Hz, and open symbols refer to ultrasonic experiments at 20 kHz cycling frequency.45

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

10

H. MAYER

that accelerate crack growth and decrease the threshold


stress intensity. The time governing process is the time
for water molecules to diffuse to the crack tip.46 Near
threshold fatigue crack growth in aluminium alloys at
ultrasonic frequency is affected by air humidity.47 However, at ultrasonic frequency, the crack is not open sufciently long to enable a full environmental effect.45,4850
This explains the lower crack propagation rates at
20 kHz compared with 20 Hz in ambient air for both alloys. The largest difference is found for load ratio
R = 1. With the increasing load ratio, the crack stays
open for an extended portion of the load cycle, water
vapour diffusion is possible for a longer time and, thus,
the extrinsic frequency effect becomes smaller.
The experimental particularity of one end of the specimen vibrating freely in fully reversed ultrasonic tests allows inserting the specimen into an electron microscope.
Geathers et al.51 performed fatigue tests with Ti6242 in
an environmental scanning electron microscope. A load
train with the specimen mounted at one end was
designed to be placed into the environmental scanning
electron microscope chamber. Crack initiation at small
notches and successive fatigue crack growth in different
contents of water vapour and in vacuum were studied.
Crack initiation was the earlier and crack growth rates
were the faster the higher the water vapour pressure.
Additionally, the microstructure showed a strong inuence
on the progress of fatigue damage. Unfavourably oriented
grains for basal slip at the notch tip retard crack initiation,
and grain boundaries decelerated or arrested cracks.
Corrosion fatigue testing
Wang et al.52 studied the effect of pitting corrosion on
VHCF properties of aluminium alloy 7075-T6. Exposition of the material to salt water led to an increasing
number and depth of corrosion pits with longer exposure
periods. Cycling at low stress amplitudes showed that the
corrosion pits drastically reduce the crack initiation time
and thus lead to earlier failures.
Schnbauer et al.53,54 investigated environmental inuences on crack initiation, small and long fatigue crack
growth and fatigue lifetime in a 12%Cr martensitic steam
turbine blade steel (403/410 steel, tensile strength
723767 MPa). Fatigue limits were lower in the presence
of chlorides and oxygen compared to air and deaerated
pure water. Fatigue crack growth rates in the near threshold regime, however, decreased in aqueous solutions
because of crack closure effects. The inuence of environmental surface degradation was investigated with articially generated corrosion pits. Endurance tests and
studies of crack initiation and small crack growth showed
that corrosion pits can be treated as small cracks. A
method was developed that enables the prediction of

the fatigue limit in the presence of corrosion pits. The


model is applicable for different environments, stress
ratios and pit sizes.

VARIABLE AMPLITUDE LOADING

Cyclic loading of technical components occurs in many


cases with VA rather than with constant amplitude
(CA). In contrast to CA loading, information about the
fatigue behaviour in the VHCF regime under VA loading
conditions is very limited. Requirements for the reliability of machines and structures make it necessary to
perform tests with VA and to accumulate information
on how to predict VA lifetimes from CA data. Studies
of lifetimes and fatigue crack growth under VA cycling
at load ratio R = 1 applying ultrasonic fatigue technique
have been performed since several decades.21,55 The
method to perform ultrasonic VA tests is to vary the
vibration amplitudes of successive pulses according to a
pre-determined VA sequence. Recently, a method to
superimpose varying static loads to a varying ultrasonic
vibration has been developed that allows to perform VA
tests at constant load ratios other than R = 1.56

Testing principle
The concept of ultrasonic CA and VA tests is illustrated
in Fig. 7.
In CA tests, the vibration amplitude of successive
pulses is kept constant. If experiments are performed
with mean loads (load ratios other than R = 1), the
superimposed force is adjusted at the servo-hydraulic or
electromechanical load frame prior to the test and stays
constant throughout the experiment (Fig. 7a).
In VA tests without mean loads (at load ratio R = 1)
or with a constant mean load, the vibration amplitude
of successive pulses is adjusted according to a predetermined repeat sequence (Fig. 7b). Computer control
is needed to determine the load sequence and to measure,
classify and store all load amplitudes during the test.
In VA tests at constant load ratio, the vibration amplitude of successive pulses as well as the preload is adjusted
according to a pre-determined repeat sequence (Fig. 7c).
During a pulse, both the nominal vibration amplitude
and the preload are kept constant, whereas both change
in successive pulses. For this purpose, computer control
is needed, which performs the following procedure illustrated in Fig. 7c.
1 The load frame is set to the corresponding preload calculated from the ultrasonic amplitude of the upcoming
pulse, the load ratio and the specimens cross-sectional
area, in a ramped movement within a pre-congured

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

RECENT DEVELOPMENTS IN ULTRASONIC FATIGUE

11

(a)

(b)

(c)
Fig. 7 Principle of pulsepause loading in an ultrasonic constant amplitude test (a), in a variable amplitude test with constant mean stress (b)
and in a variable amplitude test at constant load ratio (c).

transition time (tTransition). A settling period (tSettle) is


allowed to avoid overshoot, after which the actual preload level and the frames operational state are checked
for successful completion.
2 The specimen is excited to ultrasonic vibrations over
the time period tPulse after which the vibration amplitude diminishes over tDecay. Vibration amplitudes are
registered in real time with computer-based data acquisition over tPulse + tDecay. The acquired vibration
amplitudes are counted and classied into classes of
width of 0.1% of the maximum amplitude. This allows
damage accumulation calculations with the actually
measured rather than with a theoretical distribution
of load amplitudes.
3 The procedure starts over at step (1). It is ensured that
the correct preloading force is reached before the
pulse is initiated and that no changes in preload occur
as long as load cycles are counted.
Experimental results
Very high cycle fatigue properties of bainitic high-carbon
chromium steel 100Cr6 (SAE 52100, JIS SUJ2) were
studied under VA loading conditions with a cumulative
frequency distribution of stress amplitudes similar to a

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

Gauss distribution.57 Ultrasonic VA fatigue tests were


performed at different maximum stress amplitudes of
the random sequence yielding mean fatigue lifetimes
between 107 and 1010 cycles. Surface crack initiation at
grinding marks and internal crack initiation at Al2O3 inclusions were observed. In the HCF regime, specimens
with surface crack initiation showed lower lifetimes than
those with internal initiation, whereas the opposite was
found in the VHCF regime. CA experiments have been
performed in the regime of mean lifetimes between 105
and 109 cycles.58 SN data were approximated with two
power law functions of stress amplitudes and cycles to
failure in the HCF regime and the VHCF regime. The
stress amplitude with 50% probability to survive 109
cycles (i.e. the mean endurance limit at 109 cycles) was
used to extrapolate the SN curve above 109 cycles for
the purpose of linear damage accumulation calculations
(Miner calculations). A damage sum of S = 0.44 was found
evaluating VA experiments in the HCF regime. Linear
damage accumulation calculations delivered damage
sums as low as S = 0.016 in the VHCF regime. When
mean VA fatigue lifetimes are above 109 cycles, more than
99% of the stress amplitudes are at stress levels below the
mean endurance limit at 109 cycles. Their damaging effect
is not appropriately considered in the Miner calculation.

12

H. MAYER

Numerous low load cycles at stress amplitudes that are


too low to cause failure in a CA test can lead to failure at
low damage sums. This could be demonstrated in two-step
VA tests with cast aluminium alloy 319-T759 and 0.15%C
steel.60 One detrimental effect of a great number of low load
cycles is to accelerate fatigue crack growth.59,60 However,
under certain conditions for the maximum load of a VA sequence, benecial inuences of low load cycles, such as
prolonged lifetimes and stop of crack growth, could be documented.59,60 Because of their great number, deleterious
and benecial inuences of low load cycles are especially important in the regime of very long lifetimes, which underlines the necessity to perform VA tests in the VHCF regime.
Fitzka et al.61 investigated the VHCF fatigue properties
of spray-formed hypereutectic aluminium silicon alloy
DISPAL S232-T6x under VA loading conditions. Experiments were performed at load ratio R = 1 and compared
with CA tests.62 Figure 8 shows the measured fatigue lifetimes of DISPAL S232-T6x for CA and VA loading. Prediction of lifetimes using Miner linear damage accumulation
calculation and assuming fracture at damage sum S = 1 is
shown in addition. The mean value of calculated fatigue
lifetimes correlates reasonably well with measured mean
lifetimes. However, some early fractures occurred where
the crack was initiated at large voids or inclusions, which
are not adequately captured in the Miner calculation.
A fracture mechanics model was developed considering defects at the crack initiation location (voids and inclusions) as initial cracks. Lifetimes in CA and VA tests
were assumed to be the number of cycles necessary to
propagate the crack to fracture. An adapted Paris law
for the propagation of (small) cracks was used, and

constant and exponent were tted for the best possible


description of lifetimes in CA tests. In Fig. 9, the predicted mean lifetime using the fracture mechanics model
for an interior defect with areaCI1/2 = 91 m, which was
the largest found defect in the material, is shown with a
dash-dotted line. Additionally, predicted lifetime for an
interior defect with areaCI1/2 = 45 m, which is used to
represent the mean lifetime of all specimens, is shown
with a dashed line. Similar calculations were performed
for surface defects. Actually measured lifetimes were a
factor of 0.4 lower than predicted for the highest
maximum stress amplitude, whereas prediction and measurement coincide for the lowest maximum stress amplitude. Large crack-initiating defects strongly reduce the
fatigue lifetime. Early failures are most relevant for safe
component design, and the crack propagation model
can successfully predict the inuence of large defects.
Variable amplitude ultrasonic fatigue tests with the
aluminium alloy 2024-T351 at load ratios R = 1,
R = 0.1 and R = 0.5 in the HCF and VHCF regimes were
performed at the authors laboratory.56 Figure 10 shows
the measured lifetimes for the three load ratios. Additionally, predicted lifetimes with Miner calculation assuming
fracture at damage sum S = 1 are shown. At R = 1, a
decreasing damage sum was found when mean lifetimes increased from the HCF regime to the VHCF regime. At
mean lifetimes beyond 1010 cycles, the damage sum increases again. At load ratios R = 0.1 and R = 0.5, mean
damage sums decrease with increasing lifetime. Damage
sums are inuenced by the load ratio and can be signicantly different in the HCF and VHCF regimes. This

Fig. 8 Fatigue lifetimes of DISPAL S232-T6x measured in constant amplitude (CA, squares) and variable amplitude (VA, circles)
tests, respectively. The solid lines approximate measured data; the
dashed line shows the predicted lifetimes with Miner calculation assuming fracture at damage sum S = 1.61

Fig. 9 Fatigue lifetimes of DISPAL S232-T6x under variable amplitude loading conditions: The solid line approximates all measured
data; the dashed line shows the predicted mean lifetimes for all specimens using a fracture mechanics model assuming a defect with
1/2
areaCI = 45 m; the dash-dotted line shows the predicted lifetime
1/2
for a specimen with a large interior defect with areaCI = 91 m.61

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

RECENT DEVELOPMENTS IN ULTRASONIC FATIGUE

13

rings in bearings are subjected to very high numbers of


shear stress amplitudes in service. No reliable method exists for predicting the VHCF properties under cyclic torsion loading from the materials axial cyclic properties.
Moreover, actual components may be surface treated,
which may affect cyclic torsion and cyclic tension in a different way. It is necessary for engineering applications as
well as for material development to gather fatigue data
and to study the failure mechanisms for high numbers
of shear loads in an efcient way. Accelerated torsion
fatigue testing with the ultrasonic testing method is
therefore of great interest.
Fully reversed cyclic torsion loading

Test method

Fig. 10 Variable amplitude fatigue data of Al 2024-T351 for load ratio R = 1 (squares), R = 0.1 (circles) and R = 0.5 (triangles), respectively: The dashed lines show the predicted lifetimes with Miner
calculation assuming fracture at damage sum S = 1.56

underlines the need for the accelerated ultrasonic fatigue


testing method for proper fatigue characterization of
materials intended for use in VHCF applications.
Mller and Sander63 investigated cyclic properties of
the low-alloy steel 34CrNiMo6 in two-step block loading
tests. The high loads were by a factor of 1.1 or 1.2, respectively higher than the VHCF endurance limit, and
the low loads were lower by a factor of 0.9. The number
of low load cycles in the two-step sequence was higher by
a factor of 1, 10 and 100 than the number of high load cycles. Measured lifetimes were compared with predictions
using the Miner rule. For two-step loadings with the high
load being factor 1.1 above the VHCF endurance limit,
the Miner rule was found to be an appropriate approach
to predict lifetimes. For the high load at a factor 1.2
above the VHCF endurance limit, high number of low
load cycles led to shorter lifetimes than predicted.

CYCLIC TORSION LOADING

Fatigue studies, notably in the VHCF regime, are mainly


performed under cyclic tensioncompression, cyclic tension or rotatingbending loading. Knowledge about
VHCF under cyclic torsion loading is very limited, although several technical components are predominantly
loaded in this way. Drive shafts, coil springs or balls and

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

The ultrasonic torsion fatigue testing method has been developed at the authors institute.24,64 First experiments
were performed with 2024-T3 aluminium alloy.64 In a second work, the cyclic properties of zirkonia were tested.24
The ceramic tubes were fatigued under cyclic torsion loading in one series and with static compression stress
superimposed to cyclic torsion in another test series.
The working principle of the ultrasonic torsion fatigue testing equipment employed is as follows: An ultrasonic torsion converter generates twist oscillations and
couples torsion waves into one end of the load train. All
components of the load train must be designed appropriately to allow the formation of a standing wave at 20 kHz,
which is the resonance frequency of the ultrasonic torsion converter. Besides the ultrasonic converter, the load
train consists of a mounting part, an ultrasonic horn and
the specimen. The mounting part is an approximately
cylindrical component, where the vibration node in its
centre serves to mount the whole load train. The ultrasonic torsion horn serves to magnify the rotational oscillation. The specimen is hourglass shaped to increase the
cyclic shear strain amplitude to a magnitude high enough
to initiate cracks and to propagate them to fracture.
Design principles for the components of the load train
are based on the propagation of shear waves along the bars
of different polar momentum. The lengths of all components of an ultrasonic torsion load train are signicantly
smaller than in a cyclic tensioncompression or cyclic tension load train, because shear waves are shorter and more
sensitive to changes of cross section than longitudinal waves.
The design method and a layout capable of performing ultrasonic torsion fatigue tests are described in Refs. [64,65].
Control of vibration amplitude and resonance frequency and the way to perform tests in pulsepause mode
are similar in ultrasonic torsion and ultrasonic axial loading fatigue tests. The same test generator and ultrasonic
ampliers are used. However, different vibration gauges
are necessary for measuring rotational and axial

14

H. MAYER

movement, respectively. Additionally, the power control


of the ultrasonic system has to be adjusted to the different
(torsional or axial) converters.65

Experimental results
Akiniwa et al.66 performed ultrasonic torsion fatigue tests
with an oil-tempered SiCr spring steel (JIS G3561,
SWOSC-V). Fatigue data were measured in the regime
between 105 and 109 load cycles and were compared to
cyclic tensioncompression data. The ratio of fatigue
strength under torsional and axial loading at the same
number of cycles was about 0.68 and was similar in the
HCF and VHCF regimes. Cracks were initiated
solely at the surface and were initially growing in mode II
(i.e. in maximum shear direction, either parallel or perpendicular to the specimens length axis). At greater lengths,
cracks propagated in mode I (i.e. 45 to the specimens
length axis, perpendicular to the maximum tensile stress).
No internal inclusion-induced failure was observed. This
was attributed to the very small maximum inclusion size
in the investigated steel. Additionally, the specimens had
been electropolished to remove surface residual stresses,
which facilitated surface crack initiation.
Cyclic torsion and cyclic tensioncompression fatigue
properties of mild steel (0.15%C, C15E, 1.1141) were investigated in the authors laboratory.60,67 SN curves under both loading conditions showed a pronounced
change of slope at about 107 cycles. Above 108 cycles,
failures are very rare, indicating the existence of a fatigue
limit. With 50% fracture probability at 109 cycles, the ratio of cyclic torsion and cyclic tensioncompression fatigue strength was 0.60. Surfaces of specimens that did
not fail in the fatigue tests were studied in a scanning
electron microscope. Figure 11 shows a fatigue crack

Fig. 11 Non-propagating fatigue crack visible on the surface of a


0.15% C specimen after cyclic torsion loading close to the fatigue
9
limit (a = 159 MPa): no failure after 1.0 10 cycles. Specimens
length direction is from top to bottom, that is, maximum shear
stresses from top to bottom and from left to right.60

visible on the surface of a runout specimen. Pronounced


slip activity in the direction of maximum shear stress initiated a crack. However, the crack did not propagate to
failure within 109 cycles. The fatigue limit in mild steel
is therefore associated with a non-propagating condition
for fatigue cracks rather than with a non-initiating
condition.
Xue and Bathias68 developed an interesting alternative
way to generate torsional ultrasonic vibrations. They used
an axial vibrating ultrasonic fatigue system to stimulate an
appropriately designed load train to rotational resonance
vibrations. With this method, they studied two steels:
ferriticpearlitic steel D38MSV5S (tensile strength
878 MPa) and martensitic bearing steel 100C6 (tensile
strength 2300 MPa). Both steels did not show a fatigue
limit, and several failures were found even above 109 cycles. Fatigue cracks were initiated at the surface, at internal
inclusions and internally without the presence of inclusions. Ultrasonic cyclic torsion tests with D38MSV5S
were compared with similar tests performed at 35 Hz,
and comparable lifetimes were found in the HCF regime.
Schuller et al.69 investigated VDSiCr spring steel
(tensile strength 1750 MPa) with a shot-peened surface
(residual surface compression stresses prior to testing between 670 and 710 MPa). They performed cyclic torsion
tests at load ratio R = 1 and axial loading fatigue tests at
load ratios R = 1, R = 0.1 and R = 0.5. No VHCF failures
were found in cyclic torsion experiments, but specimens
either failed below 106 cycles or survived 109 cycles or
more. In contrast, VHCF failures from internal inclusions as well as from the surface were found for axial
loading at the three load ratios. The ratio of VHCF
strength under cyclic torsion and cyclic tension
compression was 0.86. Surface compression residual
stresses were found to be stable under cyclic axial loading,
whereas they were reduced under cyclic torsion loading.
Considering the actual residual stresses after cycling and
the mean stress sensitivity of the material, cyclic torsion
fatigue strength could be well calculated from the cyclic
axial loading fatigue strength.
Shimamura et al.70 investigated the cyclic torsion fatigue properties of carburized and non-carburized
SCM420H steel (tensile strength 1200 MPa, carburized
surface layer with hardness 817 HV). Fatigue cracks were
initiated at the surface, forming initially mode II cracks
for both surface conditions. At greater lengths, the cracks
propagated in mode I. Carburizing led to longer mode II
cracks and an increase of cyclic torsion strength by more
than 50%.
The inuences of inclusion type (i.e. oxide type inclusions or MnS inclusions) on cyclic torsion fatigue
properties of high-carbon chromium steel JIS-SUJ2
were studied by Sandaiji.71 Surface crack initiation as
well as interior inclusion crack initiation was found

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

RECENT DEVELOPMENTS IN ULTRASONIC FATIGUE

15

testing shot-peened specimens in the VHCF regime. If


the fatigue crack initiated at an oxide inclusion,
debonding between inclusion and matrix was observed.
The initial crack starting at these inclusions grew in
mode I. In contrast, fracture of MnS inclusions started
with shear mode fracture of the particle forming an initial crack parallel to the maximum shear stress. Outside
the MnS inclusion, the fatigue crack further grew in
mode I.
Cyclic torsion loading at positive load ratios
Fully reversed cyclic torsion testing does not adequately
reproduce actual in-service loading of several technical
components. Drive shafts, coil springs and bearing components are loaded predominantly with cyclic shear
stresses superimposed to a static shear load. Valve springs
in combustion engines, for example, are preloaded. This
leads to a static torsion load superimposed to cyclic
torsion and typical load ratios between R = 0.3 and 0.5.
An existing method that was recently used to study the
VHCF properties of coil springs with preloads is to test
a great number of springs in parallel in one test
frame.7274 The cycling frequency of maximum
43 Hz73 however means that VHCF tests require long
testing times. Ultrasonic torsion fatigue testing at high
load ratios offers an interesting time-saving alternative
to this conventional test method.

Fig. 12 Mechanical setup and load train for ultrasonic cyclic torsion tests with superimposed static torsion load: (1) ultrasonic
converter, (2) upper mounting part, (3) and (6) ultrasonic horns,
(4) vibration gauge, (5) specimen and (7) rotating disc and lower
mounting part.75

Test method
A method to perform ultrasonic torsion fatigue tests at
different load ratios was developed in the authors
laboratory.75 Static torsion loads are superimposed to
the ultrasonic torsion vibration, which allows testing
up to high load ratios. The mechanical setup and the ultrasonic load train are shown in Fig. 12. An ultrasonic
converter (1) generates circumferential vibrations. All
components of the load train are designed to have resonance length for circumferential vibrations at the testing
frequency of about 20 kHz.64,65 Both ends of each component twist in opposite directions, and a vibration node
is formed in the centre. The node in the centre of the
upper mounting part (2) serves for rigid mounting of
the load train without damping the vibration. Ultrasonic
horns (3) with decreasing cross sections along their
lengths serve to increase the vibration amplitude. The
vibration amplitude at the coupling of ultrasonic horn
and specimen is measured with a vibration gauge (4),
which is an adapted induction coil. The specimen (5)
has two threads for attaching it on both sides to the
ultrasonic horns (3 and 6). The rotating disc, which is
guided by a ball bearing, and the lower mounting part
(7) serve to introduce a static torque into the load train

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

without inuencing the circumferential vibrations. A


static force is applied to the rotating disc by means of
weights, which generate the correct static torque to
perform the cyclic torsion experiment at the desired load
ratio.
Static as well as cyclic shear strains are measured with
a strain gauge in the centre of the specimen. The vibration amplitude of one specimens end is measured and
serves to control the experiments. The proportional
factor between vibration amplitude and strain amplitude
is calibrated prior to actual tests.

Experimental results
VDSiCr spring steel in patented and oil-quenched condition (tensile strength 1980 MPa) was investigated under
ultrasonic torsion fatigue loading at load ratios R = 0.1,
R = 0.35 and R = 0.5.75 The material is, for example, used
in the form of wires for coil valve springs in engines. Very
high numbers of torsional load cycles are superimposed
to a static torsional load because of the preloading of
the spring. Specimens were tested with shot-peened
surfaces to reproduce the surface conditions of actual

16

H. MAYER

valve springs (residual surface compression stresses prior


to testing between 620 and 730 MPa).
Figure 13 shows the measured fatigue lifetimes at the
three load ratios as a function of shear stress amplitude.
Dashed lines indicate 50% fracture probability at 5 109
cycles. A pronounced inuence of load ratio on the measured fatigue data is apparent. That is, the SN curves are
shifted towards lower torsion stress amplitudes with the
increasing load ratio. Mean cyclic strength at load ratio
R = 0.5 is 63% of the cyclic strength at R = 0.1 and 77%
at R = 0.35, respectively.
Fatigue cracks leading to failures in the VHCF regime
are initiated internally in the matrix or, less frequently, at
the surface. Figure 14 shows the surface and the fracture
surface of a specimen that failed in the VHCF regime. A
sh-eye fracture surface is visible with a rough area at the
starting place of the fatigue crack. The crack is initiated
in the matrix, and no second-phase particle is found at
the crack initiation location.
Failures in the VHCF regime are found after cyclic
torsion loading at load ratio R = 0.1 and R = 0.35, whereas
specimens either fail below 107 cycles or do not fail at

load ratio R = 0.5. Measurements of surface residual compression stresses in runout specimens showed that they
are reduced to about 60% of their initial value after
loading with R = 0.1 and R = 0.35 and to about 30% after
loading with R = 0.5. The considerably reduced compression residual stresses at R = 0.5 facilitate surface crack initiation, and specimens either fail from the surface in the
HCF regime or do not fail at all.

DAMAGE MONITORING

Fatigue lifetime is determined by an initiation period,


where cyclic plasticity leads to the formation of
microcracks followed by a crack growth period, where
the rst short and then long cracks propagate to
fracture. Crack propagation leads to an increase of
specimens compliance and to a reduction of its
resonance frequency, which can be used as a criterion
for specimen failure. However, a pronounced drop of
the resonance frequency occurs relatively late in the
fatigue lifetime, when long cracks are present. Monitoring methods delivering indications for fatigue damage
earlier in the fatigue life have been developed recently.
One attempt is to measure (local or global) specimen
temperature. Cyclic plastic deformation produces heat,
and temperature increase can be correlated to the cyclic
plastic deformation. Another approach is to analyze the
vibration properties of the specimen, that is, the secondorder harmonics, the eigenmodes or the extremely small
variations of resonance frequency due to forming of
initial cracks. These vibration properties are affected
by plasticity and by the formation of cracks and can
therefore be used to detect microstructural changes.
Thermal analysis

Fig. 13 SN data of shot-peened VDSiCr spring steel for cyclic torsion loading at load ratios R = 0.1 (black symbols), R = 0.35 (red symbols) and R = 0.5 (blue symbols), respectively. Triangles indicate
surface crack initiation, circles refer to interior crack initiation in
the matrix and open circles with arrows show runout specimens.
9
Dashed lines indicate 50% fracture probability at 5 10 cycles at
the respective load ratios.75

Cyclic plastic deformation in an ultrasonic fatigue test


leads to an increase of the specimens temperature.
Temperature measurements had rst been used by
Papakyriacou et al.76 to quantitatively determine the cyclic plastic strain amplitude over the fatigue lifetime in
an ultrasonic experiment. Temperature increase during
a certain number of load cycles was measured with
thermocouples. They were embedded in thermopaste
and mounted at the surface of the specimen in the area
of maximum stress. Resolution of temperature measurement with thermocouples was 0.1 C. Supplemental
measurements with a high-speed infrared camera showed
specimen surface temperatures similar to those measured
with thermocouples. The cyclic plastic strain amplitude,
pl, was calculated with the temperature increase T
during one ultrasonic pulse consisting of N cycles.
Assuming that cyclic plastic deformation is the sole

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

RECENT DEVELOPMENTS IN ULTRASONIC FATIGUE

17

Fig. 14 Interior matrix crack initiation after cyclic torsion loading at a = 572 MPa and load ratio R = 0.1, 3.6 10 cycles to failure.75

reason for temperature increase of the specimen, pl can


be evaluated using the following Eq. (1):
pl

cT
;
N

(1)

where is the mass density of the material, c is the specic


heat capacity and is the stress amplitude. Plastic strain amplitudes versus the numbers of cycles were determined for
annealed and cold worked commercially pure (c.p.) niobium
and tantalum.76 Cyclic softening, followed by cyclic

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

hardening, was found for c.p. Ta, whereas c.p. Nb solely


showed cyclic softening. Cyclic plastic strain amplitudes were
about one decade larger in c.p. Ta than in c.p. Nb. Greater
plastic strain amplitudes and consequently higher plastic
strain rates served as argument why frequency inuences on
fatigue lifetimes were found in c.p. Ta but not in c.p. Nb.77
Zettl et al.78 investigated HCF and VHCF properties of
0.6%C carbon steel using the method described earlier.76
Endurance tests showed a pronounced fatigue limit of this
material with no failures above 1.5 107 cycles. Figure 15
shows a specic behaviour of the cyclic plastic strain

18

H. MAYER

Wagner et al.79 studied the change of specimens


temperature during fatigue cycling with a thermocamera.
Temperatures at the surface of cast aluminium and of
low-alloy steel specimens were investigated while cycling
specimens in the HCF regime. A sudden increase of temperature was found at the very end of fatigue life. This is
interpreted as the initiation of a fatigue crack. Based on
this observation, it is concluded that more than 92% of
the fatigue life is spent to initiate the crack.
Krewerth et al.80 used in situ thermography to monitor
the progress of fatigue damage in AlSi7Mg-T6 cast
aluminium alloy (same as A356-T6). Using a highresolution thermocamera, surface and internal crack
initiations predominantly at porosities could be monitored. Crack initiation location and crack path could be
visualized because of local temperature changes. Additional fatigue tests in the HCF regime were performed
at 105 Hz cycling frequency, which showed lifetimes
comparable with the ultrasonic tests.
Analysis of vibration properties
Fig. 15 Plastic strain amplitude versus number of load cycles in 0.6%
C steel at stresses above and below the endurance limit of 340 MPa.78

amplitude cycling at stress amplitudes close to the endurance limit of 340 MPa. Cyclic softening is observed at low
numbers of cycles, which starts the earlier the higher the
stress amplitude. After the cyclic plastic strain has reached
a maximum, cyclic hardening is observed. Cyclic hardening
may continue to up to 109 cycles at stress amplitudes below
the endurance limit. At a cyclic stress slightly above the endurance limit, the specimen failed during cyclic hardening.

It has been shown by Kumar et al.81,82 that analysis of the


nonlinearity in the signal of the ultrasonic vibration
gauge may be used to detect fatigue damage. Considering
these results, a computer system and a signal processing
electronic component were developed at Physics BOKU
Vienna to analyze the vibration and to determine the
resonance frequency with high precision.26
A portion of each pulse of about 100 ms (Fig. 16, tShot)
is fed to computer-based data acquisition at a rate that is
sufciently high for true representation of second-order
harmonics at around 40 kHz in subsequent fast Fourier

Fig. 16 Fast Fourier transform-based monitoring of fatigue damage and resonance frequency.26

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

RECENT DEVELOPMENTS IN ULTRASONIC FATIGUE

transform. The nonlinearity parameter relates the


changes in power spectral density A2 of the second
harmonic oscillation to the respective value A1 of the
resonance vibration as shown in Fig. 16 (bottom panel):
A2 2 A1 :

(2)

While A2 changes over the course of an experiment


because of the progress of fatigue damage in the specimen, the amplitude A1 of the resonance vibration at
around 20 kHz remains unchanged, as it is proportional
to the process control variable of the ultrasonic fatigue
testing system. The relative measure rel = /0 quanties
Eq. (2) at a specic number of cycles to the nonlinearity
parameter determined during the rst few pulses 0
(i.e. nonlinearity parameter of the virgin specimen).
Additional means for in situ analysis of the specimens
vibration properties are provided by monitoring its resonance frequency. The high control accuracy of resonance
frequency together with the high resolution of the power
spectrum allows detecting changes of resonance frequency with an accuracy better than 0.1 Hz. Changes of
the resonance frequency can then be related to microstructural changes. That is, decrease of resonance frequency indicates an increase of specimens compliance.
Because resonance frequency is also sensitive to the
specimens temperature, the specimen must be kept at
constant temperature within relatively close limits
(typically within 0.4 C).
Analysis of vibration properties has been performed
with the spray-formed hypereutectic aluminium silicon
alloy DISPAL S232-T6x.61 Figure 17 shows the evolution of the vibration properties versus load cycles for
a specimen that was cycled at /2 = 162 MPa and load
ratio R = 1 and failed in the VHCF regime after

19

1.3 109 cycles. From the very beginning, a slight but


monotonic increase in rel is visible. While the increase
is small at rst, after around 50% of the lifetime, the
slope steepens notably. It is only a few thousand cycles
before the end of the experiment that rel increases dramatically and the specimen fails. Decrease of resonance
frequency starts from the beginning of the experiment.
A continuously decreasing frequency is visible up to
VHCF regime, when specimen failure occurs. With the
temperature of the ultrasonic system remaining constant,
decrease of resonance frequency can be attributed to an
increase of compliance because of cracking. Progress of
fatigue damage in the VHCF regime in this sprayformed material may therefore be interpreted as an
increase in number and length of fatigue cracks.
Kumar et al. were the rst to use the nonlinearity
parameter for correlation with the fatigue damage in a
wrought aluminium and a nickel alloy81 and in a cast
aluminium alloy.82 They found that rel is more sensitive
to the initiation and growth of a fatigue crack than resonance frequency. The nonlinearity parameter was used to
detect the proportions of the fatigue life spent for crack
initiation and small and long crack growth.
Heinz et al.83 performed ultrasonic fatigue tests with
Ti6Al4V up to 1010 cycles and studied the progress of
fatigue damage with 3D laser scanning vibrometry.
The oscillation velocity in three dimensions at several
points within a grid on the specimens surface was
measured. Cyclic strain could be determined contact
free and was found to be comparable with strain gauge
measurements and FEM simulations. Eigenmodes and
eigenfrequencies of the virgin specimen were determined and compared with similar measurements after
cyclic loading of the specimen. Local changes in the
eigenmodes could be used to localize the area where
fatigue failure occurred.

FUNDAMENTAL STUDIES IN THE VERY HIGH


CYCLE FATIGUE REGIME

Fig. 17 Nonlinearity parameter rel and change of resonance

frequency fr versus number of cycles for a DISPAL S232-T6x


9
specimen that failed after 1.3 10 cycles.61

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

Fatigue testing at ultrasonic frequency allows investigating materials at numbers of cycles that are hardly accessible with conventional testing techniques because of
excessive testing times. This gave recently new insights
into the fatigue process of materials when materials are
stressed with very high numbers of low load amplitudes.
Stanzl-Tschegg et al.84,85 performed fundamental
investigations on the progress of fatigue damage in polycrystalline copper in the VHCF regime. Using cylindrical samples, the sinusoidal stress distribution along the
specimens length allowed studying the generation of
slip lines and persistent slip bands (PSBs) at different
cyclic strain amplitudes in one specimen. A minimum

20

H. MAYER

stress amplitude of about 63 MPa is necessary to generate PSBs in polycrystalline copper within 2 106 cycles
(in the conventional sense of PSB threshold). Continuing cyclic loading beyond 1010 cycles, it was found that
PSBs are formed already at about 50% of the conventional PSB threshold.84,85 Weidner et al.86 investigated
the fatigue damage produced by more than 1010 load cycles below the conventional PSB threshold in more detail.
They characterize the damage by lamella of localized shear
containing elongated dislocation cells, a pronounced surface roughening, the initiation of stage I cracks not only
at the surface but also in interior grains and substantial
grain boundary displacements.
Stanzl-Tschegg et al.85 also showed that the stress amplitudes necessary to rupture a specimen are much higher
than those necessary to form PSBs or to initiate cracks.
Stress amplitudes higher than 93 MPa (about 50% above
the conventional PSB threshold) are necessary for failure
of polycrystalline copper within 1010 cycles. Otherwise,
possibly initiated short cracks stop propagating, which
can be well interpreted using a Kitagawa diagram.
Zimmermann and Jones87 observed surface roughening and the evolution of intergranular as well as
transgranular microcracks in runout specimens of pure
nickel. The cracks reached a length of approximately
30 m but did not propagate to fracture within more than
109 cycles. Higher stresses are necessary to propagate a
crack to fracture than to initiate cracks in pure Ni.
Moderate weight, limited space requirements and the
capability to work without cooling circuits make it possible to set up an ultrasonic system quickly. Liu et al.32 designed a portable ultrasonic fatigue instrument, which
was installed at a high-brilliance X-ray beamline. Thin
sheets of a single crystal Ni-based superalloy were
mounted on a carrier specimen where a hole served to allow the X-radiation to pass through. A small load frame
was designed that served to introduce a static preload.
With this method, crack initiation and propagation could
be studied in situ at load ratio R = 0.1. Information regarding the inuences of microstructure, compositional segregation and strain on initiation and successive crack growth
could be acquired by in situ X-ray imaging.
Dnges et al.88 investigated crack initiation and early
crack growth in the HCF and VHCF regime in stainless
steel X2CrNiMoN22-5-3 (equal volume fraction of
50% of austenitic and ferritic phase). Plastic deformation was found to start in the austenitic phase by activating discrete slip bands. Slip can be transferred to the
harder ferrite phase if well-oriented slip planes are present, and this transition often led to microcrack initiation
in the ferrite phase. Initiation of a microcrack in the rst
grain and overcoming the rst structural barrier (grain
or phase boundary) essentially determines HCF and
VHCF lifetime.

Istomin et al.89 investigated the evolution of


fatigue damage in the same duplex stainless steel
X2Cr2NiMoN22-5-3 using high-energy X-rays in a
beamline. The diffracted X-ray radiation after transmission through an hourglass-shaped ultrasonic fatigue
specimen was analyzed after different numbers of cycles
and at different locations of the specimen, which were
subjected to different cyclic strain amplitudes. It was
shown that fatigue damage started by forming slip bands
and subgrains in a few austenite grains, followed by
dislocation pileup against phase boundaries, which triggered transgranular fracture in ferrite grains.
Krupp et al.90,91 investigated the role of the
microstructural barriers in duplex stainless steel
X2CrNiMoN22-5-3 in more detail. Ultrasonic fatigue
tests revealed the existence of a real fatigue limit for this
material through experimental evidence. A model for
the number of cycles required to form a crack was
developed. Once fatigue cracks are initiated, they do not
necessarily propagate to failure. Whether these cracks do
or do not lead to nal fracture depends on the barrier
efciency to the following grain or phase boundary, that
is, the strength of the rst microstructural barrier.
COMPARISON OF ULTRASONIC AND
CONVENTIONAL FATIGUE DATA

A main question involved with accelerated testing is


whether the measured fatigue properties would be
similarly obtained using conventional testing methods
working at much lower frequencies. Fatigue tests with
servo-hydraulic equipment or rotating bending equipment,
for example, are typically performed at cycling frequencies
in the range from 10 to 100 Hz. Moreover, technical
components are typically stressed at cycling frequencies far
below the ultrasonic range, and cycling frequencies used in
conventional tests are much more comparable with those
present in actual applications.
Ultrasonic fatigue data may be different from the results collected with conventional (servo-hydraulic, rotating bending, reversed bending, resonant testing, etc.)
equipment because of frequency effects as well as because of the testing technique. Frequency inuences
can be divided into intrinsic strain rate effects and extrinsic time-dependent inuences of the testing environment. Inuences of the testing technique may be
caused by the way to control the experiments and by
specimen dimension requirements. Ultrasonic tests are
typically displacement controlled, where an increase of
compliance after crack initiation leads to a decrease of
nominal stress.92 Specimens used in ultrasonic fatigue
tests often have relatively small stressed volumes. A
smaller testing volume reduces the probability of encountering large defects resulting in higher measured

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

RECENT DEVELOPMENTS IN ULTRASONIC FATIGUE

cyclic strengths.9395 Differences between stress and


strain control and the size effects are well treated in
fatigue literature and will therefore not be further
discussed. Rather, the overview will summarize possible
differences between ultrasonic and conventional frequency data because of strain rate and environmental inuences. Data published after 1999 are included. Earlier
work is reviewed in Ref. [18].
Steels and cast iron

Mild steel
Tsutsumi et al.96 performed fatigue tests at 10 and 20 kHz
with annealed and cold worked 0.13%C steel (tensile
strength 370 MPa) containing small holes. The fatigue
limit under fully reversed loading conditions is 23 to
26% higher if measured at ultrasonic frequency rather
than at 10 Hz. Near threshold fatigue crack propagation
rates are lower, and plastic zone sizes are smaller in ultrasonic frequency tests. Crack paths show many slip bands
after cycling in the conventional test and only very
few bands after ultrasonic loading. An increase of yield
stress with increasing strain rate and reduced crack propagation rates serves to explain the higher cyclic strength
measured in the ultrasonic fatigue tests.
A comprehensive study of frequency inuences in
0.15%C steel (JIS S15C, tensile strength 441 MPa) was
performed by Guennec et al.97,98 Figure 18 shows SN
curves that were measured at four different frequencies
in the range between 0.2 and 140 Hz and at ultrasonic
frequency.97 The number of cycles to failure and the fatigue limit stress were found to increase with increasing
frequency. In the regime from 0.2 and 140 Hz, the increase of cyclic strength can be explained with an increase
of yield stress with increasing strain rate, whereas the frequency inuence on ultrasonic tests is stronger. Studies

21

of dislocation structures and crack initiation mechanisms


were performed.98 Clear differences, such as transgranular crack initiation in the ferrite grains at cycling
frequencies below 140 Hz and intergranular crack initiation at ultrasonic frequency, are found.
Fully reversed fatigue tests with 0.38%C steel (JIS
S38C, tensile strength 603 MPa) were performed at
10 Hz, 400 Hz and 19.8 kHz by Nonaka et al.99 Fatigue
limits at 10 and 400 Hz were comparable, whereas significantly higher cyclic strength was measured at ultrasonic
frequency. Ductile fracture surfaces with striations were
observed after fracture at 10 and 400 Hz. Fracture surfaces obtained with ultrasonic cycling appear more brittle, striations are hardly visible and areas with cleavage
fracture are present.
Notched (notch factor Kt = 2) and smooth specimens
of ultrane grained steel with 0.15%C (15C-P, tensile
strength 926 MPa) and 0.30%C (30C, tensile strength
971 MPa) were tested by Furuya et al.100 at 150 Hz and
20 kHz. The notched specimens showed similar lifetimes
at both frequencies. Smooth specimens showed higher
cyclic strength in the ultrasonic tests. Additionally, fatigue cracks in these specimens were exclusively initiated
at the surface at 150 Hz, whereas internal crack initiation
in the VHCF regime was found in ultrasonic tests.

Austenitic steel
Austenitic stainless steel tubes (AISI 904L, tensile
strength 657 MPa) have been studied in rotating bending
tests at 160200 Hz and in ultrasonic tests at 20 kHz by
Carstensen et al.101 In the range from 3 106 to 108 cycles, rotating bending and ultrasonic fatigue data are
available, which show no inuence of cycling frequency
on measured lifetimes. Ultrasonic data had been measured up to 1010 cycles. The slope of the SN curve
strongly changes at approximately 108 cycles, and failures
at greater numbers of cycles are very rare. This indicates
the presence of an endurance limit, which is remarkable
considering the face-centred cubic (fcc) lattice structure
of the investigated metal.

High-strength steel

Fig. 18 SN diagram of S15C steel tested at load ratio R = 1 for


several loading frequencies.97

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

Furuya et al.102 investigated fatigue lifetimes of a lowalloy high-strength steel (JIS SNCM439, tensile strength
1960 MPa) at 100 Hz, 600 Hz and 20 kHz at load ratio
R = 1. Fatigue properties were found to be independent
of cycling frequency. Fatigue crack initiation was mainly
found at interior inclusions. Fine granular areas were
found at the crack initiation locations, and their sizes
were independent of cycling frequency. A spring steel
(JIS-SUP7, tensile strength 1730 MPa) was tested by
Furuya et al.103 at 30 and 100 Hz under rotating bending,

22

H. MAYER

at 120 and 600 Hz under cyclic axial loading and at


20 kHz with ultrasonic equipment. The different volumes subjected to high cyclic stresses in the respective
tests had a pronounced inuence on the measured SN
curve, whereas the inuence of the cycling frequency
was found negligible.
Kovacs et al.29 performed fatigue tests with martensitic
12%Cr steel (X10CrNiMoV12-2-2, tensile strength
1000 MPa) with an electromagnetic resonance testing
machine at 100 Hz and with ultrasonic equipment at
20 kHz. Comparing fatigue lifetimes measured with both
methods at load ratio R = 1, similar results are reported.
Both series of experiments showed preferential crack initiation at the surface for failures below about 107 cycles,
whereas failures at higher numbers of cycles are caused
by cracks initiating at internal inclusions.
Notched specimens of a high-carbon martensitic
stainless steel (X90CrMoV18, tensile strength
2220 MPa) were investigated by Schmid et al.104 Fatigue
tests were performed in ambient air at load ratio R = 0.1
with an electromagnetic testing system at 150 Hz and
with ultrasonic equipment at 20 kHz. In the regime between 105 and 108 cycles, similar fatigue lifetimes were
measured with both methods. Moreover, both testing
methods showed crack initiation preferentially at surface
carbides in the HCF regime and at internal carbides in
the VHCF regime.
Li et al.105 performed rotating bending tests, conventional axial loading tests at 95 Hz and ultrasonic tests at
20 kHz with high-carbon chromium bearing steel
(GCr15, tensile strength 2300 MPa). Lifetimes found at
low and ultrasonic frequency under axial loading coincided well. In contrast, cyclic strength measured in rotating bending tests was signicantly higher because of the
effect of testing volume. A comparable steel was also
tested by Marines et al.106 and by Akiniwa and Tanaka107
at low and ultrasonic frequency. Investigations with
smooth specimens,106,107 but also with notched specimens,107 did not reveal any inuence of testing frequency
on lifetimes.
Zhao et al.108 investigated fatigue lifetimes of highcarbon chromium bearing steel (GCr15) in rotating
bending tests at 52.5 Hz and in ultrasonic fatigue tests
at 20 kHz. The material had been tempered at 150, 300,
450 and 600 C, respectively, which led to a tensile
strength between 2370 (150 C) and 1040 MPa (600 C).
Fatigue lifetimes measured in both testing series coincided with the hardest material tempered at 150 C,
whereas prolonged lifetimes and a higher cyclic strength
were found for the other three heat treatments. Loading
frequency did not affect the failure mode. Change from
surface crack initiation in the HCF regime to internal
crack initiation in the VHCF regime was found for tempering at 150 and 300 C, whereas the softer versions

solely showed surface failures. An explanation for the frequency inuence is proposed in terms of dislocation
movement, with shorter travelling distances and smaller
accumulated damage at higher loading frequencies.

Cast iron
Wang and Bathias109 investigated fatigue lifetimes of
spheroidal graphite cast iron (tensile strength 510 MPa)
at load ratios R = 1 and R = 0 at cycling frequencies of
25 Hz and 20 kHz. No fatigue limit was observed, but
specimens continued to fail above 107 cycles. Crack initiation in the VHCF regime was preferentially at interior
microshrinkage cavities. They concluded that a very
good agreement was found between ultrasonic and conventional fatigue test data.
The inuence of cycling frequency on fatigue lifetimes of high-strength austempered ductile iron (tensile
strength 1140 MPa) was investigated by Zhang et al.110
Fatigue tests were performed at load ratio R = 1 with a
resonance testing equipment working at 90 Hz and with
ultrasonic equipment at 20 kHz. In the HCF regime,
similar SN curves were measured at low and ultrasonic
frequency. Similar crack initiation locations at graphite
nodules or at shrinkage cavities were found for both
frequencies, indicating no inuence of cycling frequency
on fatigue damage in this material.

Summary on steels
Based on the investigations described earlier, it can be
concluded that fatigue properties of low-carbon and
medium-carbon steels are sensitive to cycling frequency.
Higher cyclic strengths are measured at ultrasonic rather
than conventional frequency. Mild steel contains ferrite
and pearlite, and cyclic plastic deformation in the HCF
regime takes place in the softer ferrite. Cyclic plastic deformation is smaller because of the increase of yield
strength at high strain rates in ultrasonic tests. The dislocation structure in the ferrite grains as well as the crack
initiation is affected by frequency. Surface crack initiation
occurs in the ferrite grains at low frequency and preferentially at grain boundaries at ultrasonic frequency.
Laird and Charsley111 have very well predicted these
strong inuences of cycling frequency on fatigue damage
in body-centred cubic metals. Shape changes of grains
due to asymmetric gliding of screw dislocations in tension and compression, which promotes intergranular
crack initiation and cleavage crack planes on fracture
surfaces, are the consequences of the high strain rates in
ultrasonic testing. In contrast, cyclic deformation of fcc
metals, crack initiation and propagation and the associated fatigue lifetime is assumed to be only weakly dependent on the strain rate.111 Consequently, measured

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

RECENT DEVELOPMENTS IN ULTRASONIC FATIGUE

lifetimes of austenitic stainless steel were similar at low


and ultrasonic frequency.
High-strength steels (martensitic steels and martensitic stainless steels) often do not show frequency inuences on measured fatigue lifetimes. A possible reason is
the very small plastic strain rate involved in cycling these
materials in the HCF and VHCF regimes. The fatigue
limits of steels with hardness below about 400 HV increase approximately proportional to their hardness and
consequently proportional to their tensile strength.112
Steels with greater hardness show only a slight increase
of the cyclic strength with the static strength, if at all.112
Thus, the ratio of cyclic stresses leading to HCF and
VHCF failures and static strength decreases with increasing strength. Very small cyclic plastic deformations and
consequently low plastic strain rates even at ultrasonic
frequencies are involved in cycling martensitic steels tempered at low temperatures, martensitic stainless steels or
other high-strength steels in the HCF and VHCF regimes. VHCF cracks in high-strength steels often start
at interior inclusions. This failure mode is hardly inuenced by cycling frequency, which may be attributed to
the very small cyclic plastic deformation involved.
Aluminium alloys
Fatigue lifetimes of AlZnMgCu1.5 aluminium alloy
(similar to aluminium alloy 7075) with three different
heat treatments (AlZnMgCu1.5-T6, tensile strength
645 MPa; AlZnMgCu1.5-T66, tensile strength 666 MPa;
AlZnMgCu1.5-T64, tensile strength 570 MPa) were measured with ultrasonic equipment and with a resonant testing machine working at 100 Hz.113 Fatigue data at load
ratio R = 1 are available in the regime between 105 and
107 cycles for both frequencies. Similar lifetimes were
found testing at 100 Hz and 20 kHz, indicating the absence
of a frequency inuence on fatigue damage for this aluminium alloy. Ultrasonic tests up to 109 cycles showed that
no fatigue limit exists for the material with the investigated
heat treatments. Cyclic tension fatigue tests at load ratio
R = 0.1 were performed with 2024-T351 aluminium
alloy.4 Fatigue tests in the range between 105 and 107 cycles
to failure delivered very similar mean lifetimes for tests
with a servo-hydraulic test frame at 810 Hz and with
ultrasonic equipment at 20 kHz.
Caton et al.5 investigated cyclic properties of W319-T7
cast aluminium alloy under cyclic tension compression
loading with ultrasonic and servo-hydraulic equipment.
Three different microstructures referring to different
cooling rates with low, medium and high secondary dendrite arm spacing and tensile strengths of 327, 225 and
166 MPa, respectively, were investigated. Fatigue data in
the range from 105 to 107 cycles are available at 40 Hz
and 20 kHz cycling frequency and show no frequency

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

23

inuence on fatigue lifetime. Moreover, all three SN


curves clearly show an endurance limit. No failures above
108 cycles were found although ultrasonic experiments
were continued until 109 cycles were reached.
Fatigue properties of the hypereutectic aluminium
silicon alloy DISPAL S232-T6x in the HCF and VHCF
regimes have been studied at ultrasonic frequency by
Stanzl-Tschegg et al.62 Experiments have been performed
at room temperature and at 150 C. The same material has
been tested at IWK Aachen, Germany, at 50 Hz (20 C)
and at 140 Hz (150 C) in the HCF regime. Measured lifetimes in the HCF regime at low and ultrasonic frequency
are well comparable at both temperatures, and no inuence of cycling frequency on lifetime is found.
Zhu et al.49,50 investigated frequency inuences on fatigue lifetime of the cast aluminium alloy E319-T7. Fully
reversed fatigue experiments were performed with servohydraulic equipment at 75 Hz and with ultrasonic equipment at 20 kHz at 20, 150 and 250 C. The numbers of
cycles to failure were by factor 5 to 10 higher in ultrasonic
tests at all three temperatures.50 Cyclic loading led to very
early crack initiation at porosities for both cycling frequencies. The difference in lifetimes may therefore be attributed to a prolonged fatigue crack propagation period in
the ultrasonic experiment. An environmental superposition
model for fatigue crack growth in humid environments was
developed on the basis of experiments at low and ultrasonic
frequency in environments of different humidity. This
served to explain both the inuence of frequency and the
water vapour partial pressure on fatigue crack growth.49

Summary on aluminium alloys


Face-centered cubic lattice structures are relatively insensitive to strain rate inuences,111 which minimizes potential strain rate inuences on fatigue damage in
aluminium alloys. Frequency inuences on fatigue damage of aluminium alloys are therefore unlikely in vacuum
or in inert environments. In ambient air, humidity brings
forward chemical processes that accelerate crack growth
and deteriorate the cyclic properties. Diffusion of water
vapour to the crack tip is too slow to enable the full environmental effect during an ultrasonic cycle.49,50 Retarded crack growth compared with conventional
frequency loading is found, especially at low load ratios.45,48 This leads to prolonged fatigue lifetimes for
some aluminium alloys, whereas others are less sensitive
and similar lifetimes are found in ultrasonic and conventional frequency tests.
Titanium alloys
Papakyriacou et al.77 tested c.p. titanium (tensile
strength 522 MPa) and Ti6Al7Nb (tensile strength

24

H. MAYER

953 MPa) with rotating bending equipment at 100 Hz


and with ultrasonic equipment at 20 kHz. Fatigue lifetimes between 105 and 2 108 cycles were measured.
Ultrasonic tests of c.p. Ti showed 15% higher mean cyclic strength at 2 108 cycles compared with the rotating
bending tests. In contrast, endurance tests with
Ti6Al7Nb did not show frequency inuences on fatigue
lifetimes. Mean cyclic strength at 2 108 cycles is 81%
of the yield stress for c.p. Ti and 64% of the yield stress
for Ti6Al7Nb. Greater cyclic plastic deformations of c.p.
Ti promote strain rate effects and are used to explain why
frequency inuences are found in c.p. Ti but not in the
titanium alloy.
Morrissey and Nicholas7,114 performed fully reversed
fatigue tests with Ti6Al4V (tensile strength 968 MPa)
with servo-hydraulic equipment at 60 Hz and with ultrasonic equipment at 20 kHz. In a wide regime between
3 105 and 108 cycles, SN data are available from both
testing methods. Very consistent lifetimes were found at
both frequencies, indicating that no frequency inuence
is present under the investigated conditions.
Takeuchi et al.115 tested Ti6Al4V from three different
manufacturers (tensile strengths between 906 and
967 MPa) at cycling frequencies of 120 Hz, 600 Hz and
20 kHz at load ratio R = 1. All three heats satised
American Society for Testing and Materials specications. For smooth specimens, two of the three heats
showed internal fatigue crack initiation above approximately 5 106 cycles. Fatigue lifetimes measured for
these two heats showing internal crack initiation in the
long lifetime regime were similar testing at the three different cycling frequencies. The third heat showed solely
surface crack initiation and no failures above 5 106 cycles. This heat showed a pronounced inuence of the
testing method and signicantly higher cyclic strength
if tested at 600 Hz and 20 kHz instead of 120 Hz. Testing
notched specimens at 120 Hz and 20 kHz delivered comparable lifetimes at both frequencies.
Furuya and Takeuchi9 studied Ti6Al4V from the
same three suppliers at stress ratios of R = 0 and R = 0.3
as well as with xed maximum stress and cycling frequencies of 120 Hz and 20 kHz. With superimposed tensile
mean loads, all three heats showed failures above 5 106
cycles with crack initiation in the interior. No frequency
effect on lifetimes in the HCF and VHCF regime was
found in overall nine testing series with smooth
specimens.
Ultrasonic fatigue crack propagation studies with
Ti6Al4V in solution treated and overaged condition were
performed in the authors laboratory.116 Fatigue crack
growth rates in the regime below 10-8 m per cycle were
investigated at load ratios between R = 1 and R = 0.8.
The same material was used in a comprehensive HCF
programme in the United States, of which fatigue crack

growth measurements performed at 50 Hz and load ratios R = 0.1, R = 0.5 and R = 0.8 are available. Threshold
stress intensities at a limiting crack propagation rate of
10-10 m per cycle were comparable at low and ultrasonic
frequency for the three load ratios. At growth rates in
the range of 10-9 to 10-8 m per cycle; however, fatigue
cracks grow by mean by a factor 3 faster in ultrasonic experiments than in tests at 50 Hz. The smaller crack
length evaluated in ultrasonic experiments as well as differences in the experimental procedure leading to load
history effects is possible explanations for the differences.

Summary on titanium alloys


Titanium alloys may fracture from cracks initiated in the interior. Internal crack initiation is promoted for high numbers of cycles and high load ratios. Frequency inuences
on lifetimes are small if cracks start in the interior. If surface
failure occurs, some investigations report higher cyclic
strengths measured at ultrasonic frequency. This may be a
consequence of strain rate effects but also of environmental
effect of humidity.51 More investigations are necessary to
better understand frequency inuences in titanium alloys.
Nickel alloys
Morrisey and Golden117 performed fully reversed ultrasonic fatigue tests with Ni-based single crystals (PWA
1484, a precipitation-strengthened cast nickel alloy based
on the NiCrAl system). Fatigue tests were performed at
593 C, and lifetimes were compared with tests at 60 Hz.
They concluded that PWA 1484 does not exhibit a strong
frequency effect. At both testing frequencies, mainly subsurface crack initiation at carbide inclusions was found.
Stcker et al.118,119 investigated pure nickel, the polycrystalline nickel-base superalloys Nimonic 80A (in peakaged and overaged condition) and Nimonic 75. They investigated fatigue lifetimes over an extremely wide range from
103 to more than 1010 cycles under fully reversed loading
conditions at room temperature. Experiments in the HCF
and VHCF regimes were performed at a cycling frequency
of 135 Hz with an electromechanical resonance testing system, of 760 Hz with a servo-hydraulic system and of 20 kHz
with ultrasonic equipment. Testing all four materials, they
conclude that no signicant and systematic inuence of
the frequency on fatigue behaviour was found. Slip lines
were found to initiate even in 1010 cycle regime; however,
they did not necessarily lead to failure. This points to a considerable difference in stress amplitudes necessary to initiate
cracks and to propagate them to failure. Further tests with
Nimonic 80A in peak-aged and overaged condition at high
temperatures were performed by Zimmermann et al.120 A
moderate decrease of cyclic strength was found at
temperatures of 400 and 600 C compared with at room

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

RECENT DEVELOPMENTS IN ULTRASONIC FATIGUE

temperature. Further increase to 800 C had a strong inuence and a pronounced decrease of cyclic strength resulted,
more pronounced for the overaged condition. This was
attributed to the formation of brittle oxide layers and
microcracks at the grain boundaries.

Summary on nickel alloys


Based on a rather limited number of investigations, nickel
alloys seem not to be unduly frequency sensitive. This is
well in accordance with the expectation of a weak strain
rate sensitivity of fcc crystal systems.111
SUMMARY

Developments of ultrasonic testing since the year 1999


are reviewed. It is shown that signicant progress using
the ultrasonic fatigue testing technique has been made
in several areas.
The capabilities of the ultrasonic fatigue testing method
have been extended. Studies of fatigue lifetimes, crack initiation and propagation have been performed at different
load ratios. Superimposing mean stresses to the ultrasonic
vibration was realized, introducing forces into the load train
at vibration nodes. Cyclic loading of thin sheets can be performed utilizing a carrier specimen. Reversed bending
loading was possible stimulating bars to bending resonance
oscillations. Most interesting results are documented in
tests at different load ratios: The mean stress sensitivity in
the VHCF regime can be greatly affected by the microstructure, slopes of SN curves and mechanisms of fatigue
failures can be strongly inuenced by the load ratio and optically dark areas were found at internal crack-initiating inclusions solely at low load ratios. Investigating foams, welds
or bre-reinforced composites showed that ultrasonic
testing can be applied to a broad range of materials.
Fatigue testing in different environments served to investigate chemical inuences and inuences of the temperature
on VHCF lifetimes and very slowly growing fatigue cracks.
Ultrasonic fatigue tests have been performed with singlecrystal superalloys up to very high temperatures of 1000 C.
Increased growth rates and decreased threshold stress intensity ranges were found cycling aluminium alloys at different
load ratios in humid air rather than in vacuum. The effect
of microstructure and humidity was studied mounting the
ultrasonic specimen in an environmental scanning electron
microscope. Strong deterioration of cyclic properties due to
corrosive inuences could be documented cycling plain or
pre-corroded specimens in different uids.
Variable amplitude loading tests, notably with varying
quasi-static forces superimposed to the vibration, have
been performed. Signicant changes of the damage sums
with increasing mean lifetimes were found. Predictions of
VHCF VA lifetimes based on HCF data therefore cannot

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

25

be considered robust. Inclusions strongly deteriorate VA


lifetimes especially in the VHCF regime. Numerous low
load cycles at stresses that would not lead to failure under
CA loading can contribute to damage in a VA sequence.
VA tests at different load ratios show a strong inuence
of the mean load on lifetimes and damage sums.
A method for ultrasonic cyclic torsion loading at different
load ratios has been developed. Cyclic torsion and cyclic
tensioncompression fatigue properties have been compared. Ratio of VHCF strength under torsion and
tensioncompression loading can vary signicantly
because of several inuences that affect both loading
conditions differently (e.g. stress gradient, evolution of surface residual stresses, distribution of inclusion sizes and
stressed volume). Cyclic torsion loading with superimposed
static torsion load shows strong inuences of the load ratio
on SN curves and crack-initiating mechanisms. VHCF
failures from the surface and the interior can frequently occur at some load ratios, whereas they are absent at others.
Promising fatigue damage monitoring techniques have
been introduced based on temperature measurements
and on analyses of the vibration properties of specimens.
Cyclic softening and hardening of a material could be
demonstrated with temperature changes of the specimen
during its lifetime. High-resolution temperature measurements of specimen surfaces were able to identify
localized plasticity and the locations of crack initiation.
Based on the occurrence of second-order harmonics
and slight changes in the resonance frequency, the
progress of fatigue damage could be estimated.
Fundamental studies in the VHCF regime have been performed with copper, where a strong difference in cyclic
stress amplitudes necessary to initiate PSBs and much
higher stress amplitudes necessary for fracture were found.
Progress of VHCF damage in duplex steel starting with
slip bands in austenite grains that trigger the initiation of
microcracks in the ferrite phase has been documented.
Studies on the progress of fatigue damage in beamlines
have extended the knowledge about VHCF damage.
Comparison of ultrasonic and conventional fatigue data has
been the subject of several recent works. Higher cyclic
strength measured at ultrasonic frequency must be expected for mild steel, where the ferrite grains are sensitive
to strain rate inuences. High-strength steels and highalloy steels seem less prone to frequency inuences. Some
aluminium alloys show frequency inuences because of
time-dependent chemical processes of humidity. An ultrasonic cycle is too short to allow a full environmental effect,
which can lead to prolonged lifetimes. Titanium alloys
seem not to be frequency sensitive if failures occur from
the interior in the VHCF regime, whereas surface failures
can be associated with higher cyclic strength measured in
the ultrasonic test. Nickel tested at low temperatures seems
not to be sensitive to testing frequency as well.

26

H. MAYER

REFERENCES
1 Naito, T., Ueda, H. and Kikuchui, M. (1984) Fatigue behavior of carburized steel with internal oxides and nonmartensitic micro-structure near the surface. Metall. Trans., 15A,
14311436.
2 Asami, K. and Sugiyama, Y. (1985) Fatigue strength of
various surface hardened steels. J. Heat. Treat. Technol. Assoc.,
25, 147150.
3 Stanzl-Tschegg, S. E. and Mayer, H. R. (2001) Fatigue and fatigue crack growth of aluminium alloys at very high numbers of
cycles. Int. J. Fatigue, 23, 231237.
4 Mayer, H., Schuller, R. and Fitzka, M. (2013) Fatigue of 202410
T351 aluminium alloy at different load ratios up to 10 cycles.
Int. J. Fatigue, 57, 113119.
5 Caton, M. J., Jones, J. W., Mayer, H., Stanzl-Tschegg, S. and
Allison, J. E. (2003) Demonstration of an endurance limit in
cast 319 aluminum. Metall. Mater. Trans. A, 34A, 3341.
6 Mayer, H., Papakyriacou, M., Zettl, B. and Stanzl-Tschegg,
S. E. (2003) Inuence of porosity on the fatigue limit of die
cast magnesium and aluminium alloys. Int. J. Fatigue, 25,
245256.
7 Morrissey, R. J. and Nicholas, T. (2005) Fatigue strength of
Ti-6Al-4V at very long lives. Int. J. Fatigue, 27, 16081612.
8 Szczepanski, C. J., Jha, S. K., Larsen, J. M. and Jones, J. W.
(2008) Microstructural inuences on very-high-cycle fatiguecrack initiation in Ti-6246. Metall. Mater. Trans. A., 39A,
28412851.
9 Furuya, Y. and Takeuchi, E. (2014) Gigacycle fatigue properties of Ti-6Al-4V alloy under tensile mean stress. Mater. Sci.
Eng. A., 598, 135140.
10 Liu, X., Sun, C. and Hong, Y. (2015) Effects of stress ratio on
high-cycle and very-high-cycle fatigue behavior of a Ti-6Al-4V
alloy. Mater. Sci. Eng. A., 622, 228235.
11 Li, S. X. (2012) Effects of inclusions on very high cycle fatigue
properties of high strength steels. Int. Mater. Rev., 57, 92114.
12 Zimmermann, M. (2012) Diversity of damage evolution during
cyclic loading at very high numbers of cycles. Int. Mater. Rev.,
57, 7391.
13 Stanzl-Tschegg, S. (2014) Very high cycle fatigue measuring
techniques. Int. J. Fatigue, 60, 217.
14 Mason, W. P. (1956) Internal friction and fatigue in metals at
large strain amplitudes. J. Acoust. Soc. Am., 28, 12071218.
15 Willertz, L. E. (1980) Ultrasonic fatigue. Int. Mater. Rev., 2,
6578.
16 Wells, J. M., Buck, O., Roth, L. D. and Tien, J. K. (1982)
Ultrasonic fatigue. In: 1st International Conf on Fatigue and Corrosion Fatigue up to Ultrasonic Frequencies, The Metall. Soc. of
AIME: Philadelphia.
17 Roth, L. D. (1992) Ultrasonic fatigue testing. In: ASM Handbook, (Edited by J. R. Newby, J. R. Davis, S. K. Refsnes and
D. A. Dietrich), ASTM, Philadelphia, 240258.
18 Mayer, H. (1999) Fatigue crack growth and threshold measurements at very high frequencies. Int. Mater. Rev., 44, 136.
19 Neppiras, E. A. (1959) Techniques and equipment for fatigue
testing at very high frequencies. 62nd Annual Meeting of ASTM,
ASTM: Philadelphia, pp. 691710.
20 Stanzl, S. and Tschegg, E. (1980) Fatigue crack growth and
threshold measured at very high frequencies (20 kHz). Metall.
Sci., April, 137143.
21 Stanzl, S. E., Tschegg, E. K. and Mayer, H. R. (1986) Lifetime
measurements for random loading in the very high cycle
fatigue range. Int. J. Fatigue, 8, 195200.

22 Stanzl, S. E., Czegley, M., Mayer, H. R. and Tschegg, E. K.


(1989) Fatigue crack growth under combined mode I and mode
II loading. In: Fracture Mechanics: Perspectives and Directions
(Edited by R. P. Wei and R. P. Gangloff), ASTM,
Philadelphia, 479496.
23 Tschegg, E. K., Mayer, H. R., Czegley, M. and Stanzl S. E.
(1991) Inuence of a constant mode III load on mode I fatigue
crack growth thresholds. In: Fatigue under Biaxial and Multiaxial Loading Edited by K. Kussmaul, D. McDiarmid and D.
Socie), Mech. Engng. Pub., London, 213222.
24 Mayer, H. R., Tschegg, E. K. and Stanzl-Tschegg, S. E. (1996)
High cycle torsion fatigue of ceramic materials under combined loading conditions (cyclic torsion + static compression).
In: Multiaxial Fatigue and Design (Edited by A. Pineau, G.
Cailletaud and T. C. Lindley), Mech. Engng. Publ., London,
pp. 411421.
25 Mayer, H. R. and Stanzl-Tschegg, S. E. (1996) High frequency
torsional and axial fatigue loading of al 2024-t351. In: 6th Int
Fatigue Congress (Edited by G. Ltjering and H. Nowack),
Elsevier Sci. Ltd., Berlin, pp. 691696.
26 Mayer, H., Fitzka, M. and Schuller, R. (2013) Constant and
variable amplitude ultrasonic fatigue of 2024 T351 aluminium
alloy at different load ratios. Ultrasonics, 53, 14251432.
27 Karsch, T., Bomas, H., Zoch, H. W. and Mndl, S. (2014) Inuence of hydrogen content and microstructure on the fatigue
behaviour of steel SAE 52100 in the VHCF regime. Int. J.
Fatigue, 60, 7489.
28 Sander, M., Mller, T. and Lebahn, J. (2014) Inuence of
mean stress and variable amplitude loading on the fatigue
behaviour of a high-strength steel in the VHCF regime. Int.
J. Fatigue, 62, 1020.
29 Kovacs, S., Beck, T. and Singheiser, L. (2013) Inuence of
mean stresses on fatigue life and damage of a turbine blade steel
in the VHCF-regime. Int. J. Fatigue, 49, 9099.
30 Murakami, Y., Nomotomo, T., Ueda, T. and Murakami, Y.
(2000) On the mechanism of fatigue failure in the superlong
7
life regime (>10 cycles). Part I: inuence of hydrogen trapped
by inclusions. Fatigue. Fract. Eng. Mater. Struct., 23, 893902.
31 Murakami, Y., Nomotomo, T., Ueda, T. and Murakami, Y.
(2000) On the mechanism of fatigue failure in the superlong
7
life regime (>10 cycles). Part II: a fractographic investigation.
Fatigue. Fract. Eng. Mater. Struct., 23, 903910.
32 Liu, L., Husseini, N. S., Torbet, C. J., Kumah, D. P., Clarke,
R., Pollock, T. M. and Jones, J. W. (2008) In situ imaging of
high cycle fatigue crack growth in single crystal nickel-base superalloys by synchrotron X-radiation. J. Eng. Mater. Technol.,
130, 16.
33 Mayer, H., Schuller, R., Fitzka, M., Tran, D. and Pennings, B.
(2014) Very high cycle fatigue of nitrided 18Ni maraging steel
sheet. Int. J. Fatigue, 64, 140146.
34 Schuller, R., Fitzka, M., Irrasch, D., Tran, D., Pennings, B. and
Mayer, H. (2015) VHCF properties of nitrided 18Ni maraging
steel thin sheets with different Co and Ti content. Fatigue.
Fract. Eng. Mater. Struct., 38, 518527.
35 Zettl, B., Mayer, H., Stanzl-Tschegg, S. E. and Degischer, H.
P. (2000) Fatigue properties of aluminium foams at high numbers of cycles. Mater. Sci. Eng. A, A229, 17.
36 Zettl, B., Mayer, H. and Stanzl-Tschegg, S. E. (2001) Fatigue
properties of Al1Mg0.6Si foam at low and ultrasonic frequencies. Int. J. Fatigue, 23, 565573.
37 Cremer, M., Zimmermann, M. and Christ, H. J. (2013) Highfrequency cyclic testing of welded aluminium alloy joints in the
region of very high cycle fatigue (VHCF). Int. J. Fatigue, 57,
120130.

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

RECENT DEVELOPMENTS IN ULTRASONIC FATIGUE

38 Xue, H. Q., Tao, H., Montembault, F., Wang, Q. Y. and


Bathias, C. (2007) Development of a three-point bending fatigue testing methodology at 20 kHz frequency. Int. J. Fatigue,
29, 20852093.
39 Backe, D., Balle, F. and Eier, D. (2015) Fatigue testing of
CFRP in the very high cycle fatigue (VHCF) regime at ultrasonic frequencies. Compos. Sci. Technol., 106, 9399.
40 Ebara, R., Yamada, Y. (1982) Corrosion fatigue behaviour of
13Cr stainless steel and Ti-6Al-4V at ultrasonic frequency.
In: Ultrasonic Fatigue (Edited by J. M. Wells, O. Buck, L. D.
Roth and J. K. Tien), The Metall. Soc. of AIME, Philadelphia,
349364.
41 Tschegg, E. K. and Stanzl, S. E. (1981) Fatigue crack propagation and threshold in B.C.C. And F.C.C. metals at 77 and
293 K. Acta Metall., 29, 3340.
42 Puskr, A. and Vrkoly, L. (1986) Inuence of temperature on
fatigue crack growth behaviour of steels at ultrasonic frequency. Fatigue. Fract. Eng. Mater. Struct., 9, 143150.
43 Yi, J. Z., Torbet, C. J., Feng, Q., Pollock, T. M. and Jones, J. W.
(2007) Ultrasonic fatigue of a single crystal Ni-base superalloy at
1000 C. Mater. Sci. Eng. A, 443, 142149.
44 Furuya, Y., Kobayashi, K., Hayakawa, M., Sakamoto, M.,
Koizumi, Y. and Harada, H. (2012) High-temperature ultrasonic
fatigue testing of single-crystal superalloys. Mater. Lett., 69, 13.
45 Holper, B., Mayer, H., Vasudevan, A. K. and Stanzl-Tschegg,
S. E. (2004) Near threshold fatigue crack growth at positive
load ratio in aluminium alloys at low and ultrasonic frequency:
inuences of strain rate, slip behaviour and air humidity. Int. J.
Fatigue, 26, 2738.
46 Wei, R. P. (1970) Some aspects of environment-enhanced
fatigue-crack growth. Eng. Fract. Mech., 1, 633651.
47 Stanzl-Tschegg, S. E., Mayer, H. R. and Tschegg, E. K. (1991)
The inuence of Air humidity on near-threshold fatigue crack
growth of 2024-T3 aluminium alloy. Mater. Sci. Eng. A., 147,
4554.
48 Holper, B., Mayer, H., Vasudevan, A. K. and Stanzl-Tschegg,
S. E. (2003) Near threshold fatigue crack growth in aluminium
alloys at low and ultrasonic frequency: inuences of specimen
thickness, strain rate, slip behaviour and air humidity. Int. J.
Fatigue, 25, 397411.
49 Zhu, X., Jones, J. W. and Allison, J. E. (2008) Effect of frequency, environment, and temperature on fatigue behavior of
E319 cast aluminum alloy: small crack propagation. Metall.
Mater. Trans. A, 39A, 26662680.
50 Zhu, X., Jones, J. W. and Allison, J. E. (2008) Effect of frequency, environment, and temperature on fatigue behavior of
E319 cast aluminum alloy: stress-controlled fatigue life response. Metall. Mater. Trans. A, 39A, 26812688.
51 Geathers, J., Torbet, C. J., Jones, J. W. and Daly, S. (2015) Investigating environmental effects on small fatigue crack growth
in Ti-6242S using combined ultrasonic fatigue and scanning
electron microscopy. Int. J. Fatigue, 70, 154162.
52 Wang, Q. Y., Kawagoishi, N. and Chen, Q. (2003) Effect of
pitting corrosion on very high cycle fatigue behavior. Scr.
Mater., 49, 711716.
53 Schnbauer, B. M., Stanzl-Tschegg, S. E., Perlega, A.,
Salzman, R. N., Rieger, N. F., Zhou, S., Turnbull, A. and
Gandy, D. (2014) Fatigue life estimation of pitted 12% Cr
steam turbine blade steel in different environments and at different stress ratios. Int. J. Fatigue, 65, 3343.
54 Schnbauer, B. M., Perlega, A., Karr, U. P., Gandy, D. and
Stanzl-Tschegg, S. E. (2015) Pit-to-crack transition under
cyclic loading in 12% Cr steam turbine blade steel. Int. J.
Fatigue, 76, 1932.

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

27

55 Stanzl, S. E., Tschegg, E. K. and Mayer, H. R. (1986) Slow


fatigue crack growth under step and random loading. Z.
Metallkde., 77, 588594.
56 Mayer, H., Fitzka, M. and Schuller, R. (2014) Variable
amplitude loading of Al 2024-T351 at different load ratios
using ultrasonic equipment. Int. J. Fatigue, 60, 3442.
57 Mayer, H., Haydn, W., Schuller, R., Issler, S. and BacherHchst, M. (2009) Very high cycle fatigue properties of
bainitic high-carbon-chromium steel under variable amplitude
condition. Int. J. Fatigue, 31, 13001308.
58 Mayer, H., Haydn, W., Schuller, R., Issler, S., Furtner, B. and
Bacher-Hchst, M. (2009) Very high cycle fatigue properties
of bainitic high-carbon-chromium steel. Int. J. Fatigue, 31,
242249.
59 Mayer, H., Ede, C. and Allison, J. E. (2005) Inuence of cyclic
loads below endurance limit and threshold stress intensity on
fatigue damage in cast aluminium alloy 319T7. Int. J. Fatigue,
27, 129141.
60 Mayer, H. (2009) Fatigue damage of low amplitude cycles in
low carbon steel. J. Mater. Sci., 44, 49194929.
61 Fitzka, M., Mayer, H., Schuller, R., Stanzl-Tschegg, S. E.,
Przeorski, T. and Krug, P. (2014) Variable amplitude loading
of spray formed hypereutectic aluminium silicon alloy

DISPAL S232 in the VHCF regime. Fatigue. Fract. Eng.


Mater. Struct., 37, 945957.
62 Stanzl-Tschegg, S. E., Mayer, H., Schuller, R., Przeorski, T.
and Krug, P. (2012) Fatigue properties of spray formed

hypereutectic aluminium silicon alloy DISPAL S232 at high


and very high numbers of cycles. Mater. Sci. Eng. A., 538,
327334.
63 Mller, T. and Sander, M. (2013) On the use of ultrasonic fatigue testing technique variable amplitude loadings and crack
growth monitoring. Ultrasonics, 53, 14171424.
64 Stanzl-Tschegg, S. E., Mayer, H. R. and Tschegg, E. K. (1993)
High frequency method for torsion fatigue testing. Ultrasonics,
31, 275280.
65 Mayer, H. (2006) Ultrasonic torsion and tension
compression fatigue testing: measuring principles and investigations on 2024-T351 aluminium alloy. Int. J. Fatigue, 28,
14461455.
66 Akiniwa, Y., Stanzl-Tschegg, S., Mayer, H., Wakita, M. and
Tanaka, K. (2008) Fatigue strength of spring steel under axial
and torsional loading in the very high cycle regime. Int. J.
Fatigue, 30, 20572063.
67 Mayer, H., Stojanovic, S., Stanzl-Tschegg, S. and Zettl, B.
(2008) High cycle fatigue behavior of normalized 0.15%C steel
under tensioncompression and torsion loading. Key. Eng.
Mater., 378-379, 2938.
68 Xue, H. Q. and Bathias, C. (2010) Crack path in torsion loading in very high cycle fatigue regime. Eng. Fract. Mech., 77,
18661873.
69 Schuller, R., Mayer, H., Fayard, A., Hahn, M. and BacherHchst, M. (2013) Very high cycle fatigue of VDSiCr spring
steel under torsional and axial loading. Mat. Wiss.
Werkstofftech., 44, 282289.
70 Shimamura, Y., Narita, K., Ishii, H., Tohgo, K., Fujii, T.,
Yagasaki, T. and Harada, M. (2014) Fatigue properties of
carburized alloy steel in very high cycle regime under torsional
loading. Int. J. Fatigue, 60, 5762.
71 Sandaiji, Y., Tamura, E. and Tsuchida, T. (2014) Inuence of
inclusion type on internal fatigue fracture under cyclic shear
stress. Proc. Mater. Sci., 3l, 894899.
72 Kaiser, B. and Berger, C. (2005) Fatigue behaviour of technical
springs. Mat. Wiss. Werkstofftechn., 36, 685696.

28

H. MAYER

73 Kaiser, B., Pyttel, B. and Berger, C. (2011) VHCF-behavior of


helical compression springs made of different materials. Int. J.
Fatigue, 33, 2332.
74 Pyttel, B., Brunner, I., Kaiser, B., Berger, C. and Mahendran,
M. (2014) Fatigue behaviour of helical compression springs at
a very high number of cycles investigation of various inuences. Int. J. Fatigue, 60, 101109.
75 Mayer, H., Schuller, R., Karr, U., Irrasch, D., Fitzka, M.,
Hahn, M. and Bacher-Hchst, M. (2015) Cyclic torsion very
high cycle fatigue of VDSiCr spring steel at different load ratios. Int. J. Fatigue, 70, 322327.
76 Papakyriacou, M., Mayer, H., Plenk, H. and Tschegg, S.
(2002) Cyclic plastic deformation of tantalum and niobium at
very high numbers of cycles. Mater. Sci. Eng., A325, 520524.
77 Papakyriacou, M., Mayer, H., Pypen, C., Jr H. P. and StanzlTschegg, S. (2001) Inuence of loading frequency on highcycle fatigue properties of b.c.c. And h.c.p. Metals. Mater. Sci.
Eng. A, 308, 143152.
78 Zettl, B., Mayer, H., Ede, C. and Stanzl-Tschegg, S. (2006)
Very high cycle fatigue of normalized carbon steels. Int. J.
Fatigue, 28, 15831589.
79 Wagner, D., Ranc, N., Bathias, C. and Paris, P. C. (2009)
Fatigue crack initiation detection by an infrared thermography
method. Fatigue Fract. Eng. Mater. Struct., 33, 1221.
80 Krewerth, D., Weidner, A. and Biermann, H. (2013) Application of in situ thermography for evaluating the high-cycle and
very high-cycle fatigue behaviour of cast aluminium alloy
AlSi7Mg (T6). Ultrasonics, 53, 14411449.
81 Kumar, A., Torbet, C. J., Jones, J. W. and Pollock, T. (2009)
Nonlinear ultrasonics for in situ damage detection during high
frequency fatigue. J. Appl. Phys., 106, 02490110249019.
82 Kumar, A., Torbet, C., Pollock, T. and Jones, J. W. (2010) In
situ characterization of fatigue damage evolution in a cast Al
alloy via nonlinear ultrasonic measurements. Acta Mater., 58,
21432154.
83 Heinz, S., Balle, F., Wagner, G. and Eier, D. (2014) Analysis of fatigue properties and failure mechanisms of Ti6Al4V
in the very high cycle fatigue regime using ultrasonic technology and 3D laser scanning vibrometry. Ultrasonics, 53,
14331440.
84 Stanzl-Tschegg, S., Mughrabi, H. and Schoenbauer, B. (2007)
Life time and cyclic slip of copper in the VHCF regime. Int. J.
Faigue, 29, 20502059.
85 Stanzl-Tschegg, S. E. and Schnbauer, B. (2010) Mechanisms of strain localization, crack initiation and fracture of
polycrystalline copper in the VHCF regime. Int. J. Fatigue,
32, 886893.
86 Weidner, A., Amberger, D., Pyczak, F., Schnbauer, B.,
Stanzl-Tschegg, S. and Mughrabi, H. (2010) Fatigue damage
in copper polycrystals subjected to ultrahigh-cycle fatigue
below the PSB threshold. Int. J. Fatigue, 32.
87 Zimmermann, M., Jones, J. W. (2012) Microstructure and
early crack growth of non-ferrous metals in the very high cycle
fatigue range. In: VHCF5 (Edited by C. Berger and H.-J.
Christ), DVM, Berlin, 121126.
88 Dnges, B., Giertler, A., Krupp, U., Fritzen, C. P. and Christ,
H. J. (2014) Signicance of crystallographic misorientation at
phase boundaries for fatigue crack initiation in a duplex
stainless steel during high and very high cycle fatigue loading.
Mater. Sci. Eng. A, 589, 146152.
89 Istomin, K., Dnges, B., Schell, N., Christ, H. J. and Pietsch,
U. (2014) Analysis of VHCF damage in a duplex stainless steel
using hard X-ray diffraction techniques. Int. J. Fatigue, 66,
177182.

90 Krupp, U., Knobbe, H., Christ, H. J., Kster, P. and Fritzen,


C. P. (2010) The signicance of microstructural barriers during fatigue of a duplex steel in the high- and very-high-cyclefatigue (HCF/VHCF) regime. Int. J. Fatigue, 32, 914920.
91 Krupp, U., Giertler, A., Sker, M., Fu, H., Dnges, B., Christ,
H. J., Istominc, K., Hsecken, A., Pietsch, U., Fritzen, C. P.
and Ludwig, W. (2014) Signicance and mechanism of the
crack initiation process during very high cycle fatigue of duplex
stainless steel. Proc. Eng., 74, 143146.
92 Schoeck, G. (1982) Calculation of the stress intensity in
ultrasonic resonance. Z. Metallkde., 9, 576578.
93 Murakami, Y., Yokoyama, N. N. and Nagata, J. (2002) Mechanism of fatigue failure in ultralong life regime. Fatigue Fract.
Eng. Mater. Struct., 25, 735746.
94 Furuya, Y. (2008) Specimen size effects on gigacycle fatigue
properties of high-strength steel under ultrasonic fatigue testing. Scr. Mater., 58, 10141017.
95 Furuya, Y. (2011) Notable size effects on very high cycle
fatigue properties of high-strength steel. Mater. Sci. Eng. A,
528, 52345240.
96 Tsutsumi, N., Murakami, Y. and Doquet, V. (2009) Effect of
test frequency on fatigue strength of low carbon steel. Fatigue
Fract. Eng. Mater. Struct., 32, 473483.
97 Guennec, B., Ueno, A., Sakai, T., Takanashi, M. and Itabashi,
Y. (2014) Effect of the loading frequency on fatigue properties
of JIS S15C low carbon steel and some discussions based on
micro-plasticity behavior. Int. J. Fatigue, 66, 2938.
98 Guennec, B., Ueno, A., Sakai, T., Takanashi, M., Itabashi, Y.
and Ota, M. (2015) Dislocation-based interpretation on the
effect of the loading frequency on the fatigue properties of
JIS S15C low carbon steel. Int. J. Fatigue, 70, 328341.
99 Nonaka, I., Setowaki, S. and Ichikawa, Y. (2014) Effect of load
frequency on high cycle fatigue strength of bullet train axle
steel. Int. J. Fatigue, 60, 4347.
100 Furuya, Y., Torizuka, S., Takeuchi, E., Bacher-Hchst, M. and
Kuntz, M. (2012) Ultrasonic fatigue testing on notched and
smooth specimens of ultrane-grained steel. Mater. Des., 37,
515520.
101 Carstensen, J., Mayer, H. and Bronsted, P. (2002) Very high
cycle regime fatigue of thin walled tubes made from austenitic stainless steel. Fatigue Fract. Eng. Mater. Struct., 25, 837844.
102 Furuya, Y., Matsuoka, S., Abe, T. and Yamaguchi, K. (2002)
Gigacycle fatigue properties for high-strength low alloy steel
at 100 Hz, 600 Hz, and 20 kHz. Scr. Mater., 46, 157162.
10
103 Furuya, Y., Abe, T. and Matsuoka, S. (2003) 10 -cycle fatigue
properties of 1800 MPa-class JIS-SUP7 spring steel. Fatigue
Fract. Eng. Mater. Struct., 26, 641645.
104 Schmid, S., Hahn, M., Issler, S., Bacher-Hoechst, M., Furuya,
Y., Mehner, A., Bomas, H. and Zoch, H. W. (2014) Effect of
frequency and biofuel E85 on very high cycle fatigue behaviour
of the high strength steel X90CrMoV18. Int. J. Fatigue, 60,
90100.
105 Li, W., Sakai, T., Li, Q., Luc, L. T. and Wang, P. (2011) Effect of
loading type on fatigue properties of high strength bearing steel
in very high cycle regime. Mater. Sci. Eng. A, 528, 50445052.
106 Marines, I., Dominguez, G., Baudry, G., Vittori, J. -F.,
Rathery, S., Doucet, J. -P. A. N. D. and Bathias, C. (2003)
Ultrasonic fatigue tests on bearing steel AISI-SAE 52100 at
frequency of 20 and 30 kHz. Int. J. Fatigue, 25, 10371046.
107 Akiniwa, Y. and Tanaka, K. (2007) Very high cycle fatigue
strength of bearing steel with notch. In: 4th Int Conf Very
High Cycle Fatigue (Edited by J. E. Allison, J. W. Jones,
Larsen JM and R. O. Ritchie, TMS, Warrendale, PA, Ann
Arbor, MI.

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

RECENT DEVELOPMENTS IN ULTRASONIC FATIGUE

108 Zhao, A., Xie, J., Sun, C., Lei, Z. and Hong, Y. (2012) Effects
of strength level and loading frequency on very-high-cycle fatigue behavior for a bearing steel. Int. J. Fatigue, 38, 4656.
109 Wang, Q. Y. and Bathias, C. (2004) Fatigue characterization of
a spheroidal graphite cast iron under ultrasonic loading.
J. Mater. Sci., 39, 687689.
110 Zhang, J., Song, Q., Zhang, N., Lu, L., Zhang, M. and Cui, G.
(2015) Very high cycle fatigue property of high-strength
austempered ductile iron at conventional and ultrasonic frequency loading. Int. J. Fatigue, 70, 235240.
111 Laird, C. and Charsley, P (1982) Strain rate sensitivity effects
in cyclic deformation and fatigue fracture. In: Ultrasonic
Fatigue (Proc 1st Int Conf on Fatigue and Corrosion Fatigue up
to Ultrasonic Frequencies) (Edited by J. M Wells, O. Buck,
Roth LD, Tien JK), The Metall. Soc. of AIME, Philadelphia,
187205.
112 Murakami, Y. (2012) Material defects as the basis of fatigue
design. Int. J. Fatigue, 41, 210.
113 Mayer, H., Papakyriacou, M., Pippan, R. and Stanzl-Tschegg,
S. (2001) Inuence of loading frequency on the high cycle
fatigue properties of alznmgcu1.5 aluminium alloy. Mater. Sci.
Eng. A, 314, 5157.
114 Morrissey, R. J. and Nicholas, T. (2006) Staircase testing of a
titanium alloy in the gigacycle regime. Int. J. Fatigue, 28,
15771582.

2015 Wiley Publishing Ltd. Fatigue Fract Engng Mater Struct, 2016, 39, 329

29

115 Takeuchi, E., Furuya, Y., Nagashima, N. and Matsuoka, S.


(2008) The effect of frequency on the giga-cycle fatigue properties of a Ti-6Al-4V alloy. Fatigue Fract. Eng. Mater. Struct.,
31, 599605.
116 Mayer, H., Holper, B., Zettl, B. and Stanzl-Tschegg, S. E.
(2003) Slow fatigue crack growth in 2024-T3 and Ti-6Al4V at low and ultrasonic frequency. Z. Metallkde., 94,
539546.
117 Morrissey, R. J. and Golden, P. J. (2007) Fatigue strength of a
single crystal in the gigacycle regime. Int. J. Fatigue, 29,
20792084.
118 Stcker, C., Zimmermann, M. and Christ, H. -J. (2011) Localized cyclic deformation and corresponding dislocation arrangements of polycrystalline Ni-base superalloys and pure nickel in
the VHCF regime. Int. J. Fatigue, 33, 29.
119 Stcker, C. (2013) Einuss des Versetzungsgleitverhaltens und
der Vorgeschichteabhngigkeit auf das Ermdungsverhalten
von Nickelbasis-Superlegierungen und Nickel im VHCFBereich. Lehrstuhl fr Materialkunde und Werkstoffprfung.
Universitt Siegen, Siegener Werkstoffkundliche Berichte,
Band 8/2013.
120 Zimmermann, M., Stcker, C. and Christ, H. J. (2011) On
the effects of particle strengthening and temperature on
the VHCF behavior at high frequency. Int. J. Fatigue, 33,
4248.

También podría gustarte