You are on page 1of 7

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/257603467

Crack initiation mechanisms and fatigue lifetime of AISI 420 steel under
constant and variable amplitude loading

Conference Paper · August 2007

CITATIONS READS

0 1,221

1 author:

Bernhard Zettl
Fachhochschule Oberösterreich
36 PUBLICATIONS   708 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Thermochemical heat storage View project

All content following this page was uploaded by Bernhard Zettl on 05 June 2014.

The user has requested enhancement of the downloaded file.


VHCF-4 Fourth International Conference on Very High Cycle Fatigue
University of Michigan, Ann Arbor, Michigan USA
TMS (The Minerals, Metals & Materials Society), August 19-22, 2007

Crack initiation mechanisms and fatigue lifetime of AISI 420 steel under
constant and variable amplitude loading

B. Zettl, S. Stanzl-Tschegg, H. Mayer

University of Natural Resources and Applied Life Sciences, Vienna


Institute of Physics and Materials Science
Peter-Jordan Str. 82, A-1190 Vienna, Austria

Keywords: AISI 420, Martensitic Chromium Steel, Very High Cycle Fatigue,
Variable Amplitude Loading, Ultrasonic Resonance Testing

Abstract

Ultrasonic fatigue tests are performed with AISI 420 steel up to 1010 load cycles. The material is
tested in hardened and tempered condition and does not show a fatigue limit in S-N tests.
Different crack initiating mechanisms are found, depending on the numbers of cycles to failure.
At lifetimes below approximately 107 load cycles, crack initiation is preferentially at surface
inclusions. In the regime from about 107 to 109 cycles to failure, cracks initiate at interior
inclusions. When specimens fail at higher numbers of cycles (between 109 and 1010 cycles),
cracks initiate at the surface and no inclusions are found. Two-step variable amplitude tests are
performed in the VHCF regime. In contrast to the constant amplitude loading, crack initiation at
interior inclusions was found in all tests, and specimens failed at low damage sums.

Introduction

The fatigue behaviour of high strength steels has got increased attention due to possible failures
in the very high cycle fatigue regime. A stepwise or duplex S-N curve is found in several
investigations [1-5] where the discontinuity of the S-N curve is correlated to a fracture mode
change from surface crack initiation to crack initiation at interior inclusions. Stepwise S-N
curves were first found with rotating bending equipment and surface treated high carbon bearing
steel. The existence of a duplex S-N curve is less obvious in case of uniaxial loading with
polished or electro-polished steels [6-8].
The transition of fatigue crack initiation from surface to subsurface at a fatigue lifetimes between
106 and 108 cycles to failure (ctf) is most pronounced for specimen with high residual surface
stresses. Polished specimen often show mixed areas with both, surface and sub-surface crack
initiation. Some authors also show the evidence of a changing fracture behaviour at very high
numbers of cycles (above 108): Cracks can start from defect-free areas in austenitic chromium
steel [6], bearing steel [7] and spring steel [9].
High-frequency fatigue testing equipment with high accuracy [10] allows to investigate the
regime of very long lifetimes in a time saving way. In this work the crack initiating mechanisms
are investigated in martensitic chromium steel using ultrasonic equipment. In the first series of
experiments, the S-N curve and the associated crack initiating mechanisms are studied. It will be
shown, that three different mechanisms may be active, depending on the cyclic load and the
associated number of cycles to failure.
In a second series of experiments, fatigue behaviour is studied under varying loading conditions.
Mechanical structures are typically subjected to variable amplitude loading with the predominant
number of load cycles at low stress amplitudes. Several investigations show a significant
contribution of low load cycles to fatigue damage, leading to reduced fatigue lifetimes and
accelerated fatigue crack growth [11-13]. The investigated material shows different crack
initiating mechanisms at different cyclic stress amplitudes, and low load cycles of a random
collective may influence both, the fatigue lifetime and the crack initiating mechanism.

Experimental Setup

Endurance tests have been performed with martensitic chromium steel using ultrasonic fatigue
testing equipment working at cycling frequency of approximately 20 kHz. Specimens (shown in
Fig. 1) were machined and heat-treated in the following way: Grinding with abrasive paper up to
grade #500, heated (1 h, 980°C) in vacuum at, quenched in oil, tempered (2 h, 300°C) in vacuum
for maximum strength, air cooling and grinding to grade #1000. Strain gauges are used to
calibrate the experiments. Material properties are shown in Tab. 1.

Fig. 1: Specimen shape

Designation: Martensitic chromium stainless steel


DIN: X20Cr13 AISI: 420 JIS: SUS420J1
Chemical Composition:
C: 0.20% Si: 0.40% Mn: 0.40% Cr: 12.50%
Properties: (condition quenched and tempered at 300° for max. strength)
Yield: 1250 MPa Tensile: 1500 MPa Hardness HV: 460 Modulus: 215 GPa

Table 1: Material properties

S-N tests are performed at cyclic stress amplitudes between 620 MPa and 1000 MPa at load ratio
R=-1 in ambient air environment (20°C, 50% RH). The investigated range of fatigue lifetime is
from 5x104 to maximum 2x1010 cycles. In constant amplitude tests, the specimens are stressed in
sequences of pulses and pauses. During the pulse of each 70 ms, the specimen is stressed with
1400 load cycles at the cycling frequency of 20 kHz. Periodic cooling pauses between the pulses
are used to carry off the heat and to keep specimen's temperature below 25°C.
Ultrasonic variable amplitude tests are performed varying the amplitude of successive pulses
according to a pre-selected loading sequence [14]. In the present investigation, the fatigue
behaviour under two-step variable amplitude loading is investigated. Computer control serves to
determine the magnitude of successive pulses, and additionally, all load cycles during the
experiment are measured, classified and stored. Thus, the vibration amplitudes during increasing
the amplitude at the beginning of a pulse and decreasing afterwards can be considered in damage
accumulation calculations. The repeat sequence of the two-step tests is one pulse of 1400 load
cycles at the high load and pre-selected number of pulses of 1400 cycles each at the low load.
The number of pulses at the low amplitude is selected in order to derive comparable subtotals to
the Miner damage sum for low and high amplitude cycles.
Results

Constant-amplitude tests
Results of S-N tests are shown in Fig. 2a. The straight line in the double logarithmic plot shows
failure probability of 50%. The fracture surfaces of all failed specimens are analysed by scanning
electron microscopy, and the different symbols are used for failures starting at different crack
initiating sites.

(a) (b)
Fig. 2: Results of S-N tests; different symbols characterise the crack initiation site (a); number of
cycles to failure presented vs. the crack initiating stress intensity factor (b)

It is most interesting that crack initiation is strongly correlated to the fatigue lifetime: Specimens
that failed below 107 cycles show crack initiation at the surface, preferentially at surface
inclusions. Crack initiation in specimens, which failed between 107 and 109 cycles, is at internal
inclusions. All specimens that failed between 105 and 109 cycles show an inclusion at the
initiation place with mean diameters ranging from 10 to 25 µm. For longer lifetimes, above
109 ctf, the crack initiation site does not show an inclusion. Instead there is another type of
initiation, with brittle facets at the initiation site. Examples of all fracture modes are shown in
Figs. 3 and 4.

(a) (b)
Fig. 3: Fracture surface with crack initiation at the surface of a specimen, which was cycled at
967 MPa and failed after 8.4x104 cycles - two crack initiations sites are visible (a); crack
initiation at internal inclusion in a specimen, which was cycled at 677 MPa and failed after
8.5x108 cycles (b)
When the fatigue crack initiates at an (surface or interior) inclusion, its size has been measured.
The crack initiating stress intensity factor, KImax has been calculated according to [15] using the
cyclic stress amplitude and assuming that the inclusion is equivalent to a crack. The crack
initiating stress intensity is correlated to the numbers of cycles to failure in Fig. 2b. The lower
the crack initiating stress intensity factor, the slower the initial fatigue crack growth and
consequently the higher the number of cycles to failure.
Fig. 3a shows an example of a fracture surface obtained for lifetimes below 107 ctf. Crack
initiations at two different places are visible. The larger crack started from a surface inclusion
and the smaller from the defect-free surface. Typical fracture surface with a sub-surface failure is
shown in Fig. 3b. Fig. 4 shows in more detail the different crack initiation sites.

(a) (b)
Fig. 4: Crack initiation at surface inclusion (mean diameter 12 µm) in a specimen, which was
cycled at 860 MPa and failed after 2.24x105 cycles (a); crack initiation at interior inclusion
(mean diameter 25 µm) in a specimen, which was cycled at 753 MPa and failed after
1.7x108 cycles (b)

Above 109 ctf no inclusions have been found at crack initiation places. Instead, brittle facets
appear near the surface of the specimen. Considering the crack initiating stress intensity factor
(Fig. 3b), minimum KImax=2.1 MPa√m is necessary to initiate a fatigue crack at (internal)
inclusion leading to VHCF failure. When cyclic loads are lower, the mechanism leading to crack
initiation at the surface (which may be caused by surface roughening due to irreversible cyclic
slip [16]) becomes more damaging than inclusions.

(a) (b)
Fig. 5: Crack initiation in a specimen, which was cycled at 645 MPa and failed after
1.21x1010 cycles (a) and in a specimen, which was cycled at 624 MPa and failed after 2.25x109
cycles (b); no inclusion is found at the crack initiation place
Variable-amplitude tests
Four variable-amplitude tests using two different two-step variable loading sequences have been
performed. The low stress amplitude, σlow was 645 MPa in all tests while the high stress
amplitude, σhigh was 860 MPa and 753 MPa, respectively. S-N tests show mean lifetimes of
1.0x106 cycles (860 MPa), 2.0x107 cycles (753 MPa) and 1.3x109 cycles (645 MPa) at these
stress amplitudes.
The measured load amplitudes are used in Miner damage accumulation calculations. Fig. 6
shows the Miner sub-total of each stress-value and the accumulated value for all classes for one
specimen. The two peaks indicate the respective contribution of the low and high stress
amplitude. Both peaks are nearly equally high indicating comparable relative contribution of low
and high amplitude cycles to the calculated damage sum. In Fig. 7 the results of the two-step
variable amplitude tests are shown. The experiments show remarkably low accumulated damage
sums until failure (S=0.05-0.11) for the investigated loading sequences.

Fig. 6: Histogram of Miner subtotals (black line) of a variable amplitude test with σhigh=860 MPa
and σlow=645 MPa; the accumulated Miner sum (grey line) is shown at the right axis

Fig 7: Results of the two-step variable amplitude tests compared with the SN-curve. The results
are indicated with two markers for the high and low stress level. The specimens are numbered,
and corresponding Miner sums are shown.

Considering crack initiation, all four specimens tested under two-step variable amplitude
conditions failed after crack initiation at interior inclusions. In contrast, in constant amplitude
tests at 860 MPa, cracks started at surface inclusions. Surface or sub-surface initiation at
inclusions is found at 753 MPa, and for stress amplitudes of 645 MPa, the crack initiated either
at internal inclusions or at the surface without inclusion. During the two-step loading the
dominant crack initiation mechanism is always internal. Although different mechanisms may
cause failure in constant amplitude tests, internal inclusions are mainly responsible for failure in
the investigated variable amplitude sequences and failures in the VHCF regime.

Conclusion

Endurance tests of martensitic chromium steel AISI 420 in the range of 105 to 1010 cycles to
failure show different crack initiation places:
- below 107 cycles to failure, fatigue cracks initiate preferentially at surface inclusions
- in the regime from 107 and 109 cycles to failure, cracks initiate at interior inclusions
- failures in the regime from 109 to 1010 cycles started at the surface and no inclusion is found at
the crack initiation place.
Considering inclusions as initial cracks, the crack initiation stress intensity, KImax can be
correlated to the number of cycles to failure. Failure starting at inclusions was found for KImax
greater 2.1 MPa√m.
In two-step variable amplitude tests in the VHCF regime, crack initiation is at internal inclusions
and damage sums are low.

References

[1] K. Shiozawa, L. Lu, Very high-cycle fatigue behaviour of shot-peened high-carbon-chromium bearing steel,
Fatigue Fract. Engng. Mater. Struct. 25 (2002) 813-822.
[2] K. Tanaka, Y. Akiniwa, Fatigue crack propagation behaviour derived from S-N data in very high cycle regime,
Fatigue Fract. Engng. Mater. Struct. 25 (2002) 775-784.
[3] T. Sakai, Y. Sato, Y. Nagano, M. Takeda, N. Oguma, Effect of stress ratio on long life fatigue behavior of high
carbon chromium bearing steel under axial loading, Int. J. Fatigue 28 (2006) 1547-1554.
[4] Y. Murakami, N. Yokoyama, J. Nagata, Mechanism of fatigue failure in ultralong life regime, Fatigue Fract.
Eng. Mat. Struct. 25 (2002) 735-746.
[5] Y. Ochi, T. Matsumura, K. Masaki, S. Yoshida, High-cycle rotating bending fatigue property in very long life
regime of high-strength steel, Fatigue Fract. Eng. Mat. Struct. 25 (2002) 823-830.
[6] G. Chai, Fatigue behaviour of duplex stainless steels in the very high cycle regime, International Journal of
Fatigue 28 (2006) 1611-1617.
[7] K. Tanaka, Y. Akiniwa, N. Miyamoto, Notch effect on fatigue strength reduction in the very high cycle regime,
VHCF-3 (2004), Kusatsu, Japan, 2004, 56-67.
[8] T. Shiina, T. Nakamura, T. Noguchi, A fractographic comparison between fatigue crack propagation of surface-
originating fractures in vacuum and interior-originating fracture on high strenght steel, VHCF-3 (2004), Kusatsu,
Japan, 2004, 48-55.
[9] B Kaiser, C Berger, Fatigue behaviour of technical springs, Mat. Wiss. u. Werkstofftechn. 36 (2005) 685-696.
[10] H. Mayer, Ultrasonic torsion and tension–compression fatigue testing: Measuring principles and investigations
on 2024-T351 aluminium alloy, Int. J. Fatigue 28 (2006) 1446-1455.
[11] P. Heuler, T. Seeger, A criterion for omission of variable amplitude loading histories, Int. J. Fatigue 8 (1986)
225-230.
[12] R. Koterzawa, Y. Mudjijana, Y. Quinsheng, W. Tian-Jian, T. Nosho, Acceleration of fatigue crack growth
under intermitted overstressing with different mean levels, Fatigue Fract. Engng. Mater. Struct. 17 (1994) 1033-
1041.
[13] Y Murakami, K. Matsuda, Cause of unsuccessful results of Miner´s rule: Behavior of small fatigue crack
growth under repeated two-step loadings, in: K S Ravichandran, R O Ritchie, Y Murakami, Small fatigue cracks,
Elsevier Sci. Ltd., Amsterdam, 1999,
[14] S E Stanzl, E K Tschegg, H R Mayer, Lifetime Measurements for Random Loading in the Very High Cycle
Fatigue Range, Int. J. Fatigue 8 (1986) 195-200.
[15] Y. Murakami, M. Endo, Effects of defects, inclusions and inhomogeinities on fatigue strength, Fatigue 6 (1994)
163.
[16] H. Mughrabi, Specific features and mechanisms of fatigue in the ultrahigh-cycle regime, Int. J. Fatigue 28
(2006) 1501-1508.

View publication stats

You might also like