You are on page 1of 9

Journal of Alloys and Compounds 579 (2013) 540–548

Contents lists available at ScienceDirect

Journal of Alloys and Compounds


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

Low-cycle fatigue behaviors of hot-rolled AZ91 magnesium alloy


under asymmetrical stress-controlled cyclic loadings
Xiao-Min Chen a,b, Y.C. Lin a,b,⇑, Jian Chen c
a
School of Mechanical and Electrical Engineering, Central South University, Changsha 410083, China
b
State Key Laboratory of High Performance Complex Manufacturing, Changsha 410083, China
c
School of Energy and Power Engineering, Key Laboratory of Efficient and Clean Energy Utilization, Changsha University of Science and Technology, Changsha 410114, China

a r t i c l e i n f o a b s t r a c t

Article history: The low-cycle fatigue behaviors of hot-rolled AZ91 magnesium alloy are investigated by the uniaxial
Received 4 May 2013 asymmetric stress-controlled cyclic experiments at room temperature. The fatigue failure mechanisms
Received in revised form 8 July 2013 are analyzed by the fracture morphology characteristics. The effects of stress ratio and peak stress on
Accepted 8 July 2013
the low-cycle fatigue life are discussed. Considering the coupled effects of ratcheting damage and fatigue
Available online 16 July 2013
damage on the material failure, a phenomenological fatigue life model is developed to evaluate the low-
cycle fatigue life of the studied magnesium alloy. Results show that: (1) The fatigue cracks initiate from
Keywords:
the surface of specimens. The cleavage-like facet features appear in the crack propagation region, while
Alloy
Low-cycle fatigue
the residual ductile dimples and tearing edges appear in the fracture region. (2) The fatigue life increases
Fatigue life with the increase of stress ratio and decrease of peak stress, and the stress intensity factor range and
Asymmetrical stress-controlled cyclic twinning deformation mechanism are sensitive to the stress ratio and peak stress. (3) The amount of
loading residual twins on the fracture surface of specimens are greatly affected by the stress ratio and peak stress.
Microstructure Increasing the stress ratio and decreasing the peak stress can obviously decrease the amount of residual
twins. The residual twins kink with each other, which deteriorate the grain boundary. (4) Comparisons
between the measured and predicted fatigue life confirm that the developed phenomenological fatigue
life model can give an accurate estimate of the low-cycle fatigue life of hot-rolled AZ91 magnesium alloy.
Ó 2013 Elsevier B.V. All rights reserved.

1. Introduction lished an accurate model to describe the relationship between


the linear density of twins and the fatigue parameter. In addition,
Due to the low density, high specific strength and fatigue Lin et al. [6] developed a new stress-based fatigue life prediction
strength, and fine cutting performance, etc. [1,2], magnesium al- model for AZ31B magnesium alloy under single-step and multi-
loys are widely used in a variety of technology related applications, step asymmetric stress-controlled cyclic loadings. Considering
especially in automotive and aerospace products [3,4]. During the the effects of mean stress on the fatigue strength coefficient and fa-
actual service process, magnesium alloys and their structural com- tigue strength exponent, a modified Basquin model was proposed
ponents are inevitably subject to the cyclic loadings, which lead to to evaluate the fatigue life of hot-rolled AZ91 magnesium alloy un-
the catastrophic fatigue failures after a service period [5–7]. It is der asymmetrical stress-controlled cyclic loadings [7]. Srivatsan
significant to investigate the fatigue behavior of magnesium alloy et al. [8] studied the cyclic strain resistance, fatigue life and final
for life assessment and reliability design of components. fracture behaviors of magnesium alloy. Albinmousa et al. [9] and
During the last decades, considerable studies have been carried Lin et al. [10] investigated the cyclic behaviors of extruded
out on the ratcheting and fatigue behaviors of magnesium alloys AZ31B magnesium alloy. Shiozawa et al. [11] investigated the
and some other materials [4–31]. Yu et al. [4] studied the cyclic high-cycle fatigue failure mechanisms of extruded AZ80 magne-
deformation and fatigue behaviors of extruded ZK60 magnesium sium alloy, and found that the fatigue crack initiation can be con-
alloy. Lin et al. [5] investigated the uniaxial ratcheting and fatigue trolled by the irreversibility of twinning under the high-stress
failure behaviors of hot-rolled AZ31B magnesium alloy under amplitude. Zhang et al. [12] investigated the effects of percent
asymmetrical cyclic stress-controlled loadings, and they estab- reduction and specimen orientation on the ratcheting behaviors
of hot-rolled AZ31B magnesium alloy. Dutta et al. [13] investigated
⇑ Corresponding author at: School of Mechanical and Electrical Engineering, the influence of asymmetric cyclic loadings on the substructure
Central South University, Changsha 410083, China. Tel.: +86 013469071208. formation and ratcheting fatigue behaviors of AISI 304LN stainless
E-mail addresses: yclin@csu.edu.cn, linyongcheng@163.com (Y.C. Lin). steel. Dutta and Ray [14] studied the ratcheting phenomenon and

0925-8388/$ - see front matter Ó 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jallcom.2013.07.049
X.-M. Chen et al. / Journal of Alloys and Compounds 579 (2013) 540–548 541

post-ratcheting tensile behaviors of an aluminum alloy. Paul et al. 2. Materials and experiments
investigated the cyclic plastic deformation behaviors and damage
The commercial hot-rolled AZ91 magnesium alloy plate was used in this study.
evolution of 304LN stainless steel [15,16], SA333 Gr. 6 C–Mn steel
The chemical compositions of the studied alloy are shown in Table 1. The specimens
[17] and IFHS steel [18]. Kang et al. studied the ratcheting and fa- were prepared according to the ASTM-E-466-07 standard [40]. Dog-bone flat spec-
tigue failure behaviors of 316L stainless steel [19], and SS304 stain- imens with a rectangular section of 3 mm  2 mm and a gauge length of 20 mm
less steel [20]. Wang et al. [21] investigated the effects of pre-strain were machined along the transverse direction (TD). A series of uniaxial low-cycle
fatigue tests were conducted at room temperature. The sine wave form stress-con-
on uniaxial ratcheting and fatigue failure of Z2CN18.10 austenitic
trolled mode (the initial phase of sine waveform is 0°) was used. The constant stress
stainless steel. Chen et al. [22] proposed a new modified Ohno– rate (r_ ) is 50 MPa/s. The stress ratios (R) are 0.4, 0.2, 0, 0.2 and 0.4. The peak
Wang hardening rule for better adaptability under diverse situa- stresses (rp) are 140, 160, 180, 200 and 220 MPa. The low-cycle fatigue life (Nf)
tions by multiplying a factor to the dynamic recovery term. Also, can be automatically recorded by the testing system.
Chen and Jiao [23] proposed a modified kinematic hardening rule Usually, observing the fracture morphology of the specimens is a useful method
to understand the fatigue failure mechanisms of engineering materials or struc-
for multiaxial ratcheting prediction. Lin et al. [24] and Ma et al.
tures. In this study, a scanning electron microscope (SEM) Quanta 200 was used
[25] studied the uniaxial ratcheting behaviors of anisotropic con- to observe the fracture morphologies of specimens after fatigue tests. In addition,
ductive adhesive film, and found that the material has a strong the microstructures near the fracture surface were observed by optical microscope
memory of loading history. Chen et al. [26,27] investigated the rat- (OM). The etching solution used for optical observation was acetic picral solution
(4.2 g picric acid, 10 ml distilled water, 10 ml acetic acid and 90 ml ethanol), and
cheting and fatigue properties of the sandwiched power electron-
an etching time of 8–10 s was used. Fig. 1 shows the optical microstructures of
ics assembly and X13CrMnMoN18-14-3 steel. Drozdov and his the as-received material. It can be found that the initial microstructures of the stud-
cooperators [28–31] investigated the cyclic viscoelasticity and ied alloy consist of the fine and equiaxed grains with an average size of 25.85 lm,
viscoplasticity of semicrystalline polymers and polypropylene/ and no twins can be observed.
nanoclay hybrids.
It is well known that magnesium and its alloys have a hexag-
3. Results and discussions
onal close packed (HCP) crystal structure with the c/a p ratio
ffiffiffi of
1.624, which is less than the ideal hard-sphere value of 3. The
3.1. Fatigue failure mechanisms under asymmetrical stress-controlled
basal slip is not sufficient for the cyclic plastic deformation at
cyclic loadings
room temperature. Therefore, twinning is an important mecha-
nism to accommodate the plastic deformation. Usually,
 0i is the basal slip system. f1 0 1
 2gh1 0 1
 1i extension Fig. 2 shows the fracture morphologies of the specimen under
f0 0 0 1gh1 1 2
  the peak stress of 160 MPa and stress ratio of 0.2. The fatigue fail-
(or tension) twin and f1 0 1 1gh1 0 1 2i compression (contraction)
ure life are 717 cycles. It is observed that the fatigue fracture sur-
twin are the common deformation modes for magnesium alloys
face (Fig. 2a) can be divided into the crack initiation, crack
at room temperature [32,33]. It is found that the twinning and
propagation and final rupture zones, which are marked with ‘A0 ,
detwinning occur alternately under the cyclic loadings, and the
‘B0 , ‘C0 , respectively. It can be seen that the cracks initiates from
compressive loading can promote the twinning process, while
the specimen surface (marked with arrows), and there are two
the tensile loading favors the detwinning process [34,35]. There
crack initiation locations. The cracks propagate in two directions,
are some residual twins on the fatigue fracture surface because
and two group of cracks merge along the weakest path when the
the detwinning process cannot be fully reversible. Luo et al. [36]
 2g twins exist in the fatigue final fracture occurs [41]. Fig. 2b is the magnified image of the
observed that the residual f1 0 1
crack initiation zone ‘A0 in Fig. 2a. It reveals that the surface is rel-
fracture surface of the extruded AM30 alloy, and the coarse and
atively flat. Also, a number of parallel lamella structures appear at
long twins become more and more obvious with the increase of
the crack initiation position. Generally, the lamellar structures
total strain amplitude. Park et al. [37] confirmed that the alterna-
indicate the features of twin. It is known that the slip and twinning
tion of the twinning and detwinning appears within each loading
are two main deformation mechanisms during the cyclic plastic
cycle, and its interaction with the residual twins makes the cyclic
deformation of magnesium alloys at room temperature. However,
deformation complicated by affecting the cyclic hardening charac-
the dominant basal slip mode has only limited slip systems which
teristic. Yu et al. [38] found that the twinning-detwinning process
cannot meet the requirements of the five independent slip systems
plays a dominant role in the crack initiation under high strain
according to von Misses criterion. The basal plane slip cannot
amplitudes, and dislocation slip or interactions between the
accommodate the strain along the c-axis. Besides, the dislocation
dislocation slip and residual twin may result in the crack initiation
slip with hc + ai Burger vector is hard to initiate. Therefore, twin-
under low strain amplitudes. However, these results cannot be di-
ning is activated to provide the strain along c-axis. However, the
rectly used to describe the fatigue behaviors of magnesium alloys
newly-formed twins are in a plastically hard orientation with their
under the asymmetric stress-controlled cyclic loadings, due to the
basal poles parallel to the compressive strain direction, and can in-
significant influence of loading parameters on the cyclic deforma-
duce the dislocation pile-up at the intersection of slip and twin
tion mechanisms. Besides, the previous studies on the low-cycle
planes, as well as the twin bands [42–44]. So, the stress concentra-
fatigue behaviors of magnesium alloys mainly focused on the
tion will hinder the further plastic deformation, which results in
strain-controlled cyclic loadings. Only a few studies [5–7,39] were
the microcrack nucleation at the intersection of slip and twin
involved in the uniaxial asymmetric stress-controlled cyclic load-
planes or bands.
ings for magnesium alloys. Therefore, the mechanisms of fatigue
The arrow in Fig. 2b indicates the propagating direction of fati-
failure, as well as the effects of asymmetric cyclic loading (such
gue cracks. Usually, the primary cracks propagate in the intergran-
as the stress ratio and peak stress) on the fatigue life, need to
ular mode along a specific direction, which is presumably related
be comprehensively investigated.
to the twin boundaries [45]. But the secondary cracks may propa-
In this study, the low-cycle fatigue failure behaviors of the hot-
gate in the transgranular mode. In this study, it is hard to observe
rolled AZ91 magnesium alloy are investigated by the uniaxial
asymmetric stress-controlled cyclic experiments at room tempera-
ture. The characteristics of fracture morphology of the specimens Table 1
are analyzed in detail. Especially, the effects of stress ratio and Chemical compositions of the hot-rolled AZ91 magnesium alloy (wt.%).
peak stress on the low-cycle fatigue failure behaviors are dis- Al Zn Mn Si Cu Ni Fe Mg
cussed. Meanwhile, a stress-based fatigue life prediction model is
8.5–9.5 0.45–0.90 0.17–0.4 60.05 60.025 60.001 60.004 Bal.
proposed and validated.
542 X.-M. Chen et al. / Journal of Alloys and Compounds 579 (2013) 540–548

the high magnification SEM image of the rupture region (zone ‘C0 in
Fig. 2a). It is observed that the fracture morphology is mainly com-
posed of the ductile dimples and tearing edges, which indicate the
ductile fracture features.
It can be easily found that there are great differences in the frac-
ture morphologies among three stages of the fatigue crack devel-
opment. This attributes to the different fatigue crack propagation
rates in different fatigue stages. Usually, fatigue cracks initiate in
only two to five grains, and its propagation rate is relatively low
at the initial stage. Because the alteration of the slip and twinning
occurs during the middle stage of fatigue, the fatigue crack propa-
gation maintains a relatively stable state. However, due to the con-
tinuous increase of the crack size and plastic zone sizes at the crack
tip, the fatigue crack propagation accelerates before the material
failure [34]. Once the twinning terminates and no further slip is
possible, the rapid fracture occurs, and then the fracture surface
with the ductile dimples and tearing edges appears.
Fig. 1. Microstructure of the as-received AZ91 magnesium alloy. The microstructures near the fracture surface are observed by
optical microscope (OM). Fig. 3 shows the microstructures near
the tiny secondary cracks. Fig. 2c shows the high magnification the fracture surface under the peak stress of 160 MPa and stress ra-
SEM image of the fatigue crack propagation characteristics (zone tio of 0.4. From Fig. 3, it can be found that some residual twins
‘B0 in Fig. 2a). It reveals some convex and cleavage-like facet fea- appear near the fracture surface (marked with a rectangular
tures. Generally, the fatigue striations are the most important fea- frame). Due to the high activation energy, the twin favors to initi-
tures of fatigue crack propagation region. For the studied ate in the large grains. While the twins often cannot be observed
magnesium alloy, there is no classical striations at the fatigue within the very small grains, because the critical resolved sheer
propagation zone. This may be related to the abundance of twins stress (CRSS) for twinning is not achieved (shown in Fig. 3b). Evi-
due to their sharp texture characters in wrought magnesium alloy. dently, the plastic deformation in these small grains is delivered
According to the fatigue crack propagation mechanism, the inner by the slipping. It is confirmed that the distribution of the residual
fatigue damage of the studied alloy gradually accumulates under twins is a consequence of the competition between the twinning
the cyclic loadings. Once the slip and twinning cannot accommo- and detwinning [46,47]. In addition, from Fig. 3b, it can be found
date the further plastic deformation, the fatigue cracks will rapidly that the twins are straight and almost parallel to each other
propagate, leading to the fast failure of the material. Fig. 2d shows (marked with arrows).

Fig. 2. The fracture morphologies of the specimen under the peak stress of 160 MPa and stress ratio of 0.2. (a) Overall view at low magnification (80); (b) magnified view
near the crack initiation area A (1000); (c) magnified view of the crack propagation area B (5000) and (d) magnified view of the fracture area C (5000).
X.-M. Chen et al. / Journal of Alloys and Compounds 579 (2013) 540–548 543

Fig. 3. The microstructures near the fracture surface under the peak stress of 160 MPa and stress ratio of 0.4. (a) Low magnification OM image; (b) high magnification OM
image of the rectangle zone.

The residual twins are mainly composed of f1 0 1  2gh1 0 1


 1i loadings. For this reason, the residual twins can be regarded as a
extension twins. This can be explained by twinning initiation serious damage to the fatigue properties of the material.
mechanism. The critical resolved shear stress of extension twin-
ning f1 0 1 2gh1 0 1 1i is relatively low (2–2.8 MPa). When the com-
3.2. Effects of asymmetric stress-controlled cyclic loadings on low-
pression stress is perpendicular (tensile stress is parallel to c-axial cycle fatigue behaviors
of grains), the extension twin f1 0 1  2gh1 0 1
 1i easily appears [48].
Furthermore, the tested specimens are subject to the asymmetric 3.2.1. Effects of stress ratio on low-cycle fatigue behaviors
stress-controlled cyclic tension–compression loading. Due to the Fig. 4 shows the microstructures near the fracture surfaces un-
limited slip systems for the studied material, twinning is a signifi- der the peak stress of 200 MPa and different stress ratios (R). The
cant deformation mode. Specifically, although the tensile loading is fatigue failure lives are 717, 1913 and 9989 cycles for the cases
perpendicular to the c-axis of the studied alloy during the tensile with stress ratios of 0.4, 0 and 0.4, respectively. Obviously, the ef-
half cycle, the compression twinning is hard to initiate and the ba- fects of stress ratio on the fatigue failure behavior of the studied
sal slip is the main plastic deformation mechanism. Then, in order magnesium alloy are significant. The fatigue life increases with
to accommodate the further plastic deformation, the extension the increase of stress ratio, and the specimen under the stress ratio
twins f1 0 1  2gh1 01 1i generate during the tensile half cycle, caus-
of 0.4 shows a superior fatigue resistance.
ing the extension along the c-axial of the grains. So, the residual From Fig. 4, it can be seen that some residual twins appear near
twins near the fracture surface are mainly composed of the fracture surface under three different stress ratios. Figs. 4a and
 2gh1 0 1
f1 0 1  1i extension twins. However, after the saturation of
b show the microstructures near the fracture surfaces under the
the extension twinning system, the extension twinning is not ac- peak stress of 200 MPa and stress ratio of 0.4. It reveals that
tive. Therefore, the compression twins f1 0 1  1gh1 0 1
 2i, which is
the twin bands appear in some large grains. The twins with lentic-
difficult to be activated at room temperature [49], can be activated. ular shape are often recognized as extension twins f1 0 1  2gh1 01
 1i
The twinning deformation plays a significant role under cyclic (zones I, II and III), and some tiny and narrow twin bands are re-
loadings because it changes the orientations of the neighboring garded as compression twins f1 0 1  1gh1 0 1
 2i (zone IV). Mean-
grains and induces the initiation of new slip deformation. How- while, the kinked twin bands appear at local positions near the
ever, it also greatly influences the fatigue failure behaviors of mag- fracture surface (inside the zones II and III), as shown in Fig. 4b.
nesium alloys. This is attributed to its significant effects on the This is because the growing speeds of twin are different, and the
fatigue crack initiation and propagation stages. On the one hand, growing process may be hindered when the growing twin meets
the fatigue cracks easily initiate at the twin boundaries. This is be- with the pre-existed one. In addition, there are intense interactions
cause that the twin boundaries are the plane imperfections with among the twins. Generally, the interactions are a twin going into
the higher energy, and the sharp edge of the existed twin lamellas another one, or two twins kinking together or crossing certain
can cause the local stress concentration. In addition, there are in- stumbling twins, and generating a new twin with the same shear
tense interactions between the twin and matrix grain boundaries. variable of the growing twin [52]. It is recognized that the defor-
The remained twin-matrix interfaces can act as the preferential mation twins cannot grow from one grain to another, due to the
sites for the crack nucleation [50,51]. On the other hand, the ex- deformation impediment of grain boundaries. Specifically, the
isted twins can hinder the crack propagation, namely, when the grain boundaries, as a kind of deformation impediment, can termi-
crack propagation direction is not in accordance with the twinning nate the growth of deformation twins and slip bands, thus the
orientation, fatigue crack will round or cross the twins. In addition, grain boundaries can terminate the twin boundary cracks. How-
the twins can release the local stress and inactivate the crack tip, as ever, the kinked twin bands may destroy the original grain bound-
well as reduce the nucleation activity of crack during the low-cycle aries, leading to the decreased viscous force between the grains.
fatigue failure. Therefore, the twinning is beneficial to the fatigue Then, the plastic resistance of crystal grain will decrease corre-
propagation to some extent. However, just like the cyclic slip, spondingly, which leads to the decreased fatigue resistance of
twinning is also a form of plastic deformation, which can lead to the studied material. Fig. 4c and d shows the microstructures near
a certain degree of the irreversible fatigue damage. This is relative fracture surface under the peak stress of 200 MPa and stress ratio
to the fact that the existed twins presumably cause the distortion of 0. It is observed that there is abundance of twin bands near
of lattices, and act as barriers to dislocation slip and pile-up. This the fracture surface. Most of them are compression twins
is often considered as the main reason for the cyclic hardening  1gh1 0 1
f1 0 1  2i. The twin bands are characterized as straight and
[42]. Therefore, the bearing capacity of the grain boundary will narrow shapes, and parallel to each other in the same grains. Some
be deteriorated under the asymmetrical stress-controlled cyclic twin bands intersect with the adjacent twins, as shown in Fig. 4d
544 X.-M. Chen et al. / Journal of Alloys and Compounds 579 (2013) 540–548

Fig. 4. The microstructures near the fatigue fracture surfaces under the peak stress of 200 MPa and stress ratios of: (a) 0.4; (c) 0; (e) 0.4. ((b), (d) and (f) are the high
magnification OM images of the rectangle zones of (a), (c) and (e), respectively).

(zone I). Besides, there are also some kinked twin bands with parallel to the length and width direction of the metallographic
hedge-like shapes (zone II of Fig. 4d). As discussed above, the microstructure graph, respectively. N k is the number of the twins
kinked twin bands can deteriorate the grain boundaries, and thus through the lines and can be counted by the image analysis soft-
can be regarded as a kind of potential fatigue damage to the stud- ware, PHOTOGRAPHER.
ied material. The microstructures near fracture surface under the According to Eq. (1), the line density of twins for Fig. 4b, d and f
peak stress of 200 MPa and stress ratio of 0 are shown in Fig. 4e can be estimated as 7.49%/lm, 3.50%/lm and 1.81%/lm, respec-
and f. It can be seen that there are only a few residual twins with tively. Obviously, the amount of residual twins decreases approxi-
short bar shape (zones I and II) near the fracture surface. The spec- mately with the increase of stress ratio. This is because that the
imen undergoes the cyclic loadings with no compressive stress, primary fatigue crack nucleation is at grain boundaries when the
and the slip deformation mechanism is not sufficient for the cyclic stress ratio is relatively low (negative). Usually, the dominant
plastic deformation. So, the compression twinning is activated. nucleation mechanism is switching to cracking at twin boundaries
However, the features of compression twins in Fig. 4e and f are for the case with low stress ratios. The main plastic deformation
greatly different from those in Fig. 4c and d. The reason is that mechanisms are slipping, twinning and detwinning, as well as their
the short bar twin may be other kind of compression twins expect interactions under the cyclic loadings. The fatigue life under the low
for the compression twins f1 0 1  1gh1 0 1
 2i. stress ratio (especially negative stress ratio) is mainly affected by
The amount of twins can be estimated by line density of twins the serious fatigue damage resulting from the twinning and det-
(k) [5], winning in the low-cycle fatigue failure process. The amount of
residual twins, as the result of the competition of twinning and det-
Nk winning, has a significant influence to the fatigue life. It can be con-
k¼ ð1Þ
mpþnq cluded that the fatigue life decreases with the increase of residual
where p and q are the real length and width of the metallographic twins. Therefore, the hot-rolled AZ91 magnesium alloy shows a
microstructure graph, respectively. m and n are the number of lines superior low-cycle fatigue resistance under high stress ratios.
X.-M. Chen et al. / Journal of Alloys and Compounds 579 (2013) 540–548 545

3.2.2. Effects of peak stress on low-cycle fatigue behaviors alloy under single-step and multi-step asymmetric stress-con-
In this section, the effects of peak stress (rp) on the microstruc- trolled cyclic loadings. Here, the fatigue parameter can be ex-
tures near fracture surfaces of the hot-rolled AZ91 magnesium al- pressed as,
loy are discussed. Figs. 3b and 4f show the high magnification OM
images under the applied stress ratio of 0.4 with peak stresses of ðra  rf1 Þ þ kðrp  rf1 Þ
FP ¼ ð4Þ
160 MPa and 200 MPa, respectively. According to Eq. (1), the line E
densities of twins can be evaluated as 1.93%/lm and 7.49%/lm
where E is the elastic modulus. rp is the peak stress. ra is the stress
for the cases with peak stresses of 160 MPa and 200 MPa, respec-
amplitude. rf1 is the fatigue limit. It is assumed that there is no
tively. The measured fatigue failure lives are 1209 and 717 cycles
damage if the applied stress is lower than the fatigue limit (rf1 ).
under the peak stresses of 160 and 200 MPa, respectively. It can
k is the weight factor of the ratcheting damage, which can be deter-
be found that the specimen under the low peak stress shows a high
mined by the trial and error method to maximize the correlation
fatigue resistance. This is because that the twinning can lead to fa-
coefficient between the fatigue parameter (FP) and measured fati-
tigue damage due to its irreversible plastic deformation, as well as
gue life (Nf). For the studied magnesium alloy, the values of E and
its distortion of lattices. Thus, the increase of residual twins re-
rb are determined as 10,600 MPa and 248.8 MPa by uniaxial tests,
duces the fatigue resistance of the studied magnesium alloy.
respectively. rf1 can be estimated as 49.3 MPa [55], and the value
Usually, the fatigue cracks open and close alternatively under
of k is 4.
the cyclic loadings, and gradually propagate to the critical failure
From Eq. (4), it can be seen that the total damage can be re-
size. The plastic zone size ahead of crack tip simultaneously
flected by the linear superposition of the fatigue damage and rat-
changes with the propagation of the fatigue cracks. Fatigue crack
cheting damage. Therefore, it is thought that the fatigue
propagation rate (da/dN) is dominated by the stress intensity fac-
parameter (FP) can be used as a suitable variable to describe the
tor range (DK), and the relationship between DK and da/dN can
damage induced by the asymmetrical stress-controlled cyclic load-
be described as [52],
ings for the hot-rolled AZ91 magnesium. Liu et al. [54] described
da=dN ¼ CðDKÞb ð2Þ the relationship between the low-cycle fatigue life (Nf) and the fa-
tigue life parameter (FP) by the following exponential function,
where a and N are the initial crack length and cycle numbers,
respectively. b and C are material constants, which are influenced
Nf ¼ a  ½expðb  FPÞ ð5Þ
by the microstructure of the tested material, loading waves, loading where a and b are material constants, which can be determined by
frequency, stress ratio, as well as the ambient temperature. Usually, regressing the experimental results.
the value of b is between 2 and 4 for ductile materials. For long Based on the experimental conditions and results, the calcu-
cracks, the stress intensity factor range (DK) can be expressed as lated fatigue parameters (FP) for the studied magnesium alloy
[53], are summarized in Table 2. By the nonlinear fitting method, the
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi relationship between low-cycle fatigue life (Nf) and fatigue param-
DP pa pa
DK ¼ sec ð3Þ eter (FP) can be obtained (shown in Fig. 5). The material constants
B 2W 2
a and b in Eq. (5) can be estimated as 4.381  108 and
where DP is the stress range, which is equal to rp(1R). W and B are 2.538  102, respectively. It is observed that the experimental
the width and thickness of specimens, respectively. a is an interme- data points extremely disperse beside the fitting line. The main
diate variable, which can be calculated as (2a/W). Generally, the va- reason is that the low-cycle fatigue life of the hot-rolled AZ91 mag-
lue of a is less than 0.95. nesium alloy is significantly influenced by the stress ratio, but the
From Eqs. (2) and (3), it can be easily concluded that the stress effect of the stress ratio on low-cycle fatigue behavior is not
intensity factor range (DK) increases markedly with the increase of
peak stress or decrease of stress ratio, and then the fatigue propa-
Table 2
gation rate (da/dN) increases correspondingly, which results in the
Values of fatigue parameter (FP) for the hot-rolled AZ91 magnesium alloy.
short fatigue life. Furthermore, the fatigue crack propagation rate
(da/dN) sharply increases with the slightly increase of stress inten- Specimen ID Peak stress (rp, MPa) Stress ratio (R) FP

sity factor range (DK). Therefore, the specimen under the high peak 1 140 0.4 0.0388
stress (such as 200 MPa) or low stress ratio (0.4) for the studied 2 140 0.2 0.0375
3 140 0 0.0362
magnesium alloy has a short fatigue life due to the high crack prop-
4 140 0.2 0.0349
agation rate. 5 140 0.4 0.0335
6 160 0.4 0.0477
7 160 0.2 0.0462
3.3. Fatigue life evaluation models
8 160 0 0.0447
9 160 0.2 0.0432
3.3.1. The fatigue parameter method 10 160 0.4 0.0417
It is recognized that both the fatigue damage and ratcheting 11 180 0.4 0.0566
damage occur when the material are subject to the asymmetrical 12 180 0.2 0.0549
13 180 0 0.0532
stress-controlled cyclic loadings, which results in the final failure
14 180 0.2 0.0515
of material [6,19–31,39]. So, an accurate model to evaluate the 15 180 0.4 0.0498
low-cycle fatigue life under the asymmetrical stress-controlled 16 200 0.4 0.0654
cyclic loadings should not only consider the fatigue damage, but 17 200 0.2 0.0635
also the ratcheting damage. Liu et al. [54] proposed a fatigue 18 200 0 0.0617
19 200 0.2 0.0598
parameter (FP) to collectively consider the effects of fatigue dam- 20 200 0.4 0.0579
age and ratcheting damage on the low-cycle fatigue life. Based 21 220 0.4 0.0743
on the fatigue parameter method, Lin et al. [5,6] established an 22 220 0.2 0.0722
accurate model to describe the relationship between the linear 23 220 0 0.0701
24 220 0.2 0.0681
density of twins and the fatigue parameter, and developed a new
25 220 0.4 0.0660
stress-based fatigue life prediction model for AZ31B magnesium
546 X.-M. Chen et al. / Journal of Alloys and Compounds 579 (2013) 540–548

250000 Table 3
Values of the modified fatigue parameter (FP0 ) for the hot-rolled AZ91 magnesium
alloy.

Specimen ID Peak stress (rp, MPa) Stress ratio (R) FP0


200000
1 140 0.4 0.0432
2 140 0.2 0.0364
Fatigue life Nf (cycle)

Nf =4.381 108 exp(-2.538 102 FP) 3 140 0 0.0295


150000 4 140 0.2 0.0227
5 140 0.4 0.0159
6 160 0.4 0.0538
7 160 0.2 0.0454
100000 8 160 0 0.0371
9 160 0.2 0.0287
10 160 0.4 0.0204
11 180 0.4 0.0644
50000 12 180 0.2 0.0545
13 180 0 0.0446
14 180 0.2 0.0348
15 180 0.4 0.0249
0
16 200 0.4 0.0749
0.03 0.04 0.05 0.06 0.07 0.08 17 200 0.2 0.0636
18 200 0 0.0522
FP 19 200 0.2 0.0408
20 200 0.4 0.0294
Fig. 5. Relationships between the low-cycle fatigue life (Nf) and fatigue parameter 21 220 0.4 0.0855
(FP). 22 220 0.2 0.0726
23 220 0 0.0597
24 220 0.2 0.0468
25 220 0.4 0.0340
adequately considered. In Eq. (4), only the peak stress is chosen to
consider the ratcheting damage in the fatigue parameter FP, while
the effect of stress ratio on the low-cycle fatigue life is neglected. In 160000
order to accurately characterize the total damage of the hot-rolled
AZ91 magnesium under asymmetrical stress-controlled cyclic
loadings, the fatigue parameter (FP) method still needs to be im-
proved by considering the synthetical effects of peak stress and 120000
Fatigue life N f (cycle)

stress ratio on the fatigue behavior in the studied magnesium alloy.


Nf =4.859 106 exp(-1.754 10 2FP')

3.3.2. The modified fatigue parameter method


80000
In order to consider the synthetical effects of peak stress and
stress ratio on the fatigue behavior of the studied magnesium alloy,
the following modified fatigue parameter (FP0 ) is proposed by
importing the term (1R) into the fatigue parameter FP. 40000
f f
ðra  r 1 Þ þ kð1  RÞðrp  r1 Þ
FP0 ¼ ð6Þ
E
where FP0 is the modified fatigue parameter, and R is the applied 0
stress ratio. E, rp, ra, and rf1 are the elastic modulus, peak stress,
0.02 0.04 0.06 0.08 0.1 0.12
stress amplitude and the fatigue limit, respectively. k is the weight
Modified fatigue parameter FP'
factor of the ratcheting damage, which can be determined by the
trial and error method. The product of k, (1R) and ðrp  rf1 Þ rep- Fig. 6. Relationships between the low-cycle fatigue life (Nf) and modified fatigue
resents the ratcheting damage induced by the asymmetric stress- parameter (FP0 ).
controlled cyclic loading. For the studied magnesium alloy, the val-
ues of E, rb, rf1 and k are same to Section 3.3.1. Table 3 gives the
calculated FP0 for the studied magnesium alloy. Then, according to Nf ¼ a1  expðb1  FP0 Þ þ a2  expðb2  FP0 Þ ð7Þ
the exponential function in Eq. (5), the variations of Nf with FP0
for the studied magnesium alloy can be easily obtained, as shown
in Fig. 6. The value of a and b can be evaluated as 4.859  106 and where a1, b1, a2 and b2 are the material constants, which can reflect
1.754  102, respectively. the sensitivity of the fatigue life to the fatigue parameter FP0 . By the
The comparisons between the measured and predicted fatigue nonlinear fitting method, the values of a1, b1, a2 and b2 for the stud-
life are shown in Fig. 7. It can be seen that the established fatigue ied alloy can be estimated as 3.123  104, 4.255  101,
life model can well predict the fatigue life of the studied material 1.079  107, 2.188  102, respectively. The second term a2 -
only when the fatigue life is higher than 10,000 cycles. However,  exp(b2  FP0 ) can be seen as an additional part, which can ensure
there are large errors between the measured and predicted results a good prediction capability of the proposed model. Fig. 9 shows the
when the fatigue life is smaller than 10,000 cycles. This indicates comparisons between the measured and predicted fatigue life for
the exponential function (Eq. (5)) may not be accurate enough to the studied magnesium alloy. It is shown that all the predicted re-
describe the relationship between fatigue life Nf and the modified sults are located within the twice error band, which indicates that
fatigue parameter FP0 . Therefore, by the nonlinear fitting method the modified fatigue parameter (FP0 ) method can well predict the
(shown in Fig. 8), the relationships between Nf and FP0 are im- low-cycle fatigue life of the hot-rolled AZ91magnesium alloy under
proved as [6], asymmetrical stress-controlled cyclic loadings.
X.-M. Chen et al. / Journal of Alloys and Compounds 579 (2013) 540–548 547

1000000 1000000
Twice error band Twice error band
100000
Predicted life Nfp (cycle)

100000

Predicted life Nfp (cycle)


10000

1000
10000
100

10
1000

0.1
100
0.1 1 10 100 1000 10000 100000 1000000
100 1000 10000 100000 1000000
Measured life Nf (cycle)
Measured life Nf (cycle)
Fig. 7. Comparisons between the predicted and measured fatigue life of hot-rolled
AZ91 magnesium alloy. Fig. 9. Comparisons between the predicted and measured low-cycle fatigue life of
the hot-rolled AZ91 magnesium alloy.

low-cycle fatigue life for the studied magnesium alloy under the
200000
asymmetrical stress-controlled cyclic loading.

160000 Acknowledgements

This work was financially supported by Key Laboratory of Effi-


Fatigue life Nf (cycle)

Nf =3.123 104 exp(-4.255 101FP')


120000 cient & Clean Energy Utilization, College of Hunan Province (No.
+1.079 107 exp(-2.188 10 2FP')
2012NGQ001), Sheng-hua Yu-ying Program of Central South Uni-
versity, Program for New Century Excellent Talents in University
80000 (No. NCET-10-0838), and the Fundamental Research Funds for
the Central Universities of Central South University (No.
2013zzts031), China.
40000
References

[1] J. Deng, Y.C. Lin, J. Chen, S.S. Li, Y. Ding, Mater. Des. 49 (2013) 209–219.
0 [2] Y.J. Qin, Q.L. Pan, Y.B. He, W.B. Li, X.Y. Liu, X. Fan, Mater. Manuf. Process. 25
(2010) 539–545.
0 0.02 0.04 0.06 0.08 0.1 0.12 [3] M.C.L. de Oliveira, V.S.M. Pereira, O.V. Correa, N.B. de Lima, R.A. Antunes,
Modified fatigue parameter FP' Corros. Sci. 69 (2013) 311–321.
[4] Q. Yu, J.X. Zhang, Y.Y. Jiang, Q.Z. Li, Int. J. Fatigue 36 (2012) 47–58.
Fig. 8. Relationships between low-cycle fatigue life (Nf) and the modified fatigue [5] Y.C. Lin, Z.H. Liu, X.M. Chen, J. Chen, Mater. Sci. Eng. A 573 (2013) 234–244.
[6] Y.C. Lin, Z.H. Liu, X.M. Chen, J. Chen, Comput. Mater. Sci. 73 (2013) 128–138.
parameter (FP0 ).
[7] Y.C. Lin, X.M. Chen, Z.H. Liu, Int. J. Fatigue 48 (2013) 122–132.
[8] T.S. Srivatsan, L. Wei, C.F. Chang, Eng. Fract. Mech. 56 (1997) 735–747.
[9] J. Albinmousa, H. Jahed, S. Lambert, Int. J. Fatigue 33 (2011) 1403–1416.
4. Conclusions [10] X.Z. Lin, D.L. Chen, Mater. Sci. Eng. A 496 (2008) 106–113.
[11] K. Shiozawa, T. Kashiwagi, T. Murai, T. Takahashi, Proc. Eng. 2 (2010) 183–191.
[12] H. Zhang, D.X. Dong, S.J. Ma, C.F. Gu, S. Chen, X.P. Zhang, Mater. Sci. Eng. A 575
Typical low-cycle fatigue behaviors of the hot-rolled AZ91 mag- (2013) 223–230.
nesium alloy is investigated by the uniaxial asymmetric stress- [13] K. Dutta, S. Sivaprasad, S. Tarafder, K.K. Ray, Mater. Sci. Eng. A 527 (2010)
7571–7579.
controlled cyclic tests at room temperature. The effects of stress ra- [14] K. Dutta, K.K. Ray, Mater. Sci. Eng. A 540 (2012) 30–37.
tio and peak stress on residual twins, as well as the fatigue failure [15] S.K. Paul, S. Sivaprasad, S. Dhar, S. Tarafder, Mech. Mater. 43 (2011) 705–720.
life, are discussed. A modified fatigue parameter (FP0 ) is proposed [16] S.K. Paul, S. Sivaprasad, S. Dhar, S. Tarafder, Mater. Sci. Eng. A 528 (2011) 4873–
4882.
to fully consider the synthetical effects of peak stress and stress ra- [17] S.K. Paul, S. Sivaprasad, S. Dhar, S. Tarafder, Mater. Sci. Eng. A 528 (2011) 7341–
tio on the fatigue behaviors of the material. It can be concluded 7349.
that: (1) Fatigue cracks initiate from the specimen surface and [18] S.K. Paul, Mater. Sci. Eng. A 538 (2012) 349–355.
[19] G.Z. Kang, O.T. Bruhns, K. Sai, Comput. Mater. Sci. 50 (2011) 1399–1405.
propagate along twin boundaries. (2) The residual twins indicate
[20] G.Z. Kang, Y.J. Liu, Z. Li, Mater. Sci. Eng. A 435 (2006) 396–404.
a serious potential damage to fatigue resistance. (3) The effects [21] Y. Wang, D. Yu, G. Chen, X. Chen, Int. J. Fatigue 52 (2013) 106–113.
of peak stress and stress ratio on the fatigue crack propagation rate [22] X. Chen, R. Jiao, K.S. Kim, Int. J. Plast. 2 (2005) 161–184.
and the features of residual twins are significant. With the decrease [23] X. Chen, R. Jiao, Int. J. Plast. 20 (2004) 871–898.
[24] Y.C. Lin, X.M. Chen, J. Zhang, Polym. Test. 30 (2011) 8–15.
of stress ratio or peak stress, the amount of residual twins and the [25] J. Ma, H. Gao, L. Gao, X. Chen, Polym. Test. 30 (2011) 571–577.
fatigue crack propagation rate both increase. The material under [26] G. Chen, Z.S. Zhang, Y.H. Mei, X. Li, G.Q. Lu, X. Chen, Microelectron. Reliab. 53
the high peak stress or low stress ratio shows a short fatigue (2013) 645–651.
[27] G. Chen, S.C. Shan, X. Chen, H. Yuan, Comput. Mater. Sci. 46 (2009) 572–578.
life. (4) Comparisons of the measured and predicted results indi- [28] A.D. Drozdov, R. Klitkou, J.deC. Christiansen, Mech. Mater. 56 (2013) 53–64.
cate that the developed fatigue life model can well predict the [29] A.D. Drozdov, Mater. Sci. Eng. A 528 (2011) 8781–8789.
548 X.-M. Chen et al. / Journal of Alloys and Compounds 579 (2013) 540–548

[30] A.D. Drozdov, R. Klitkou, J.deC. Christiansen, Mech. Res. Commun. 48 (2013) [42] S. Begum, D.L. Chen, S. Xu, A.A. Luo, Mater. Sci. Eng. A 517 (2009) 334–343.
70–75. [43] L. Wu, S.R. Agnew, D.W. Brown, G.M. Stoica, B. Clausen, A. Jain, D.E. Fielden,
[31] A.D. Drozdov, J.deC. Christiansen, Comput. Mater. Sci. 53 (2012) 396–408. P.K. Liaw, Acta Mater. 56 (2008) 3699–3707.
[32] M.H. Yoo, J.K. Lee, Philos. Mag. A 63 (1991) 987–1000. [44] D.W. Brown, A. Jain, S.R. Agnew, B. Clausen, Mater. Sci. Forum 539–543 (2007)
[33] M.R. Barnett, Mater. Sci. Eng. A 464 (2007) 8–16. 3407–3413.
[34] S.M. Yin, F. Yang, X.M. Yang, S.D. Wu, S.X. Li, G.Y. Li, Mater. Sci. Eng. A 494 [45] L. Wu, A. Jain, D.W. Brown, G.M. Stoica, S.R. Agnew, B. Clausen, D.E. Fielden,
(2008) 397–400. P.K. Liaw, Acta Mater. 56 (2008) 688–695.
[35] W. Wu, S.Y. Lee, A.M. Paradowska, Y. Gao, P.K. Liaw, Mater. Sci. Eng. A 556 [46] M. Huppmann, M. Lentz, S. Chedid, W. Reimers, J. Mater. Sci. 46 (2011) 938–
(2012) 278–286. 950.
[36] T.J. Luo, Y.S. Yang, W.H. Tong, Q.Q. Duan, X.G. Dong, Mater. Des. 31 (2010) [47] C.L. Fan, D.L. Chen, A.A. Luo, Mater. Sci. Eng. A 519 (2009) 38–45.
1617–1621. [48] Y.N. Wang, J.C. Huang, Acta Mater. 55 (2007) 897–905.
[37] S.H. Park, S.G. Hong, B.H. Lee, W. Bang, C.S. Lee, Int. J. Fatigue 32 (2010) 1835– [49] L. Wu, S.R. Agnew, Y. Ren, D.W. Brown, B. Clausen, G.M. Stoica, H.R. Wenk, P.K.
1842. Liaw, Mater. Sci. Eng. A 527 (2010) 7057–7067.
[38] Q. Yu, J.X. Zhang, Y.Y. Jiang, Q.Z. Li, Int. J. Fatigue 33 (2011) 437–447. [50] H.R. Mayer, H.J. Lipowsky, M. Papakyriacou, R. Rösch, A. Stich, S. Stanzl-
[39] Y.C. Lin, X.M. Chen, G. Chen, J. Alloy. Comp. 509 (2011) 6838–6843. Tschegg, Fatigue Fract. Eng. Mater. Struct. 22 (1999) 591–600.
[40] ASTM-E-466-07, Standard Practice for Conducting Force Controlled Constant [51] X.L. Tan, H.C. Gu, Int. J. Fatigue 18 (1996) 329–333.
Amplitude Axial Fatigue Tests of Metallic Materials, ASTM International, [52] P.C. Paris, M.P. Gomez, W.E. Anderson, The Trend Eng. 13 (1961) 9–14.
Philadelphia, 2007. [53] GB/T6398-2000, Standards Press of China, Beijing, China, 2000 (in Chinese).
[41] H.A. Patel, D.L. Chen, S.D. Bhole, K. Sadayappan, Mater. Sci. Eng. A 528 (2010) [54] Y.J. Liu, G.Z. Kang, Q. Gao, Int. J. Fatigue 32 (2010) 678–684.
208–219. [55] Y.J. Qiao, Q. Sun, Mech. Eng. 29 (2007) 47–50 (in Chinese).

You might also like