Professional Documents
Culture Documents
Determination of Chaboche Combined Hardening Parameters With Dual Backstress For Ratcheting Evaluation of AISI 52100 Bearing Steel
Determination of Chaboche Combined Hardening Parameters With Dual Backstress For Ratcheting Evaluation of AISI 52100 Bearing Steel
Keywords: This research proposes a method to obtain parameters of Chaboche combined hardening model for evaluating
Chaboche model ratcheting phenomenon in metallic materials when cyclic loadings are applied. Finite element (FE) model is used
Ratcheting evaluation for simulating ratcheting by changing Chaboche model parameters. Although Chaboche model with single term
Finite element analysis can describe the ratcheting characteristics, simulated stress - strain (σ-ε) relation deviates from that of metallic
Uniaxial ratcheting
materials when the applied alternative stress σa is smaller than the yield strength σo. Whereas, Chaboche model
Bearing steel
with dual backstress can simulate not only the ratcheting characteristics but also σ-ε relation even when σa is
smaller than σo. Finally, tensile and ratcheting experiments are conducted with bearing steel specimens (AISI
52100) to validate the proposed method for evaluation of ratcheting parameters and σ-ε relations.
1. Introduction model). These models became as most representative models for de-
scribing the cyclic behaviors of metals. Since, compared to O-W model,
Application of cyclic loadings under working condition significantly Chaboche model is highly accessible as it is implemented in Abaqus
affects the fatigue life of most engineering structures [1,2]. Ratcheting [36] and it shows a notable improvement in the ratcheting prediction,
phenomenon, which is the accumulation of plastic strain resulting from we intensively address the Chaboche model in this paper. Reviews on
cyclic loading due to cyclic mechanical or thermal stress, also shortens kinematic hardening models with cyclic plasticity are well documented
the fatigue life due to the plastic deformation under cyclic loading by Chaboche [37,38] and Kang [15].
[3,4]. Engineering structures under cyclic loading such as bearings, Several researchers investigated the ratcheting phenomenon with
rails, turbine blades, atomic reactors need to be designed with ratch- Chaboche model [16,22,39–41]. It is also important to predict the ac-
eting in consideration, since excessive plastic deformation by ratcheting curate parameters of Chaboche model for realistic simulation of
can be one of the primary reasons for structural failures [5–11]. Over ratcheting phenomenon in metallic materials. Lemaitre and Chaboche
the last decades, understanding the ratcheting phenomenon has been [40] proposed a method for obtaining the parameters of Chaboche
the particular interest of many researchers. As a result, quantitative and model by using stabilized hysteresis loops under strain controlled cyclic
qualitative studies have been conducted with ratcheting under cyclic stress test. Using a single backstress Chaboche model without isotropic
uniaxial and multiaxial loadings [12–20], cyclic indentation [21,22] hardening, which is the same model as A-F model [32], Xu and Yue
including thermal ratcheting [23–25] and time dependent ratcheting [22] and Goo [41] simulated the ratcheting phenomenon, however, it is
[26–29]. difficult to simulate entire ratcheting strain by using this model. Abaqus
Numerous kinematic hardening rules have been developed to de- User’s manual [36] also describes the method of Lemaitre and Chaboche
scribe the cyclic behavior including ratcheting phenomenon in metals [40] mainly for cyclic plasticity parameter prediction; but the method is
[30–35]. Prager [34] and Ziegler [35] proposed linear kinematic not suitable for precise ratcheting simulation. Alternatively, trial and
hardening models to account for the Bauschinger effect. Followed by, a error method can be used to determine the parameters by matching
multi-surface model of Mroz [31], and a nonlinear kinematic hardening simulation results with experimental ratcheting data. Bari and Hassan
model of Armstrong and Frederick [32] (A-F model), in which a re- [39] proposed a parameter acquisition method using a triple backstress
laxation term is added to the linear kinematic hardening model, were Chaboche model. Koo et al. [16] demonstrated that a dual backstress
also developed. Chaboche [30] and Ohno and Wang [33] improved the Chaboche model can simulate ratcheting, and proposed a parameter
A-F model with multiple backstresses (Chaboche model and O-W acquisition method. Bari and Hassan [39] and Koo et al. [16], albeit for
⁎
Corresponding author.
E-mail address: hylee@sogang.ac.kr (H. Lee).
https://doi.org/10.1016/j.ijfatigue.2019.01.009
Received 8 November 2018; Received in revised form 24 December 2018; Accepted 19 January 2019
Available online 22 January 2019
0142-1123/ © 2019 Elsevier Ltd. All rights reserved.
S. Koo et al. International Journal of Fatigue 122 (2019) 152–163
If the Eqs. (1) and (2) are used together, the size of yield surface
changes and the center of yield surface translates together in the de-
min
viatoric space as shown in Fig. 1; this is called as combined hardening.
In this paper, Chaboche model refers to Chaboche combined hardening
0 time
model; for the case without isotropic hardening, it will be additionally
specified. Fig. 2. Schematic representation of stress-controlled cyclic loading.
153
S. Koo et al. International Journal of Fatigue 122 (2019) 152–163
FE model is accounted by using nonlinear geometric change (NLGEOM) alternative stress σa satisfies σo < σa < σo + Qmax. The detailed de-
option in Abaqus/standard [36]. Uniformly distributed stress is im- scriptions of these dynamic phases with Chaboche model are covered in
posed on the top of the specimen. Material behavior for the FE model is the subsequent Section 3.3.
assigned based on the Chaboche combined hardening model. For
parametric studies, the reference material constants are chosen with a
3.3. Parametric study with single backstress
single backstress (A-F model) to analyze the relationship between ki-
nematic and isotropic hardening parameters in ratcheting simulation as
3.3.1. Conditions for occurrence of ratcheting strain
listed in Table 1; where E is the elastic modulus and ν is the Poisson’s
FE analyses confirm that the Chaboche model is essential for
ratio of the specimen.
ratcheting simulation; however, δε does not always occurs even with
Chaboche model. Therefore, a range of applied stress conditions is in-
3.2. Isotropic hardening vs. Chaboche model vestigated by changing only σm and σa while the reference material
parameters in Table 1 are kept constant. FE analyses are performed for
Most of the structural analyses use the isotropic hardening model three cases of (σm, σa) MPa = (150, 301), (200, 300), (250, 301)
considering only the size change of yield surface when the structure is without isotropic hardening (=w/o IH) to prevent the size change of
under monotonic loading. However, it is inappropriate to simulate the yield surface. As a result, δε occurs for (σm, σa) MPa = (150, 301), (250,
material behaviors under cyclic loading as it cannot explain several 301) as shown in Fig. 7; a common feature of both cases is that σo
cyclic plastic behaviors including the Bauschinger effect [35]. On the (=300 MPa) < σa (=301 MPa). In the case of (σm, σa) = (200, 300)
other hand, Chaboche model can demonstrate the cyclic plastic beha- MPa, the cyclic behavior is fully elastic under cyclic stress since the
viors and particularly simulate ratcheting in metals. Fig. 6 compares the applied alternative stress is smaller than the size of yield surface. This
numerically obtained σ-ε curves with isotropic hardening alone and the trend alone is not sufficient to reveal whether the occurrence of δε
Chaboche model under cyclic loadings. Material shows purely elastic depends on the size of the initial or subsequent yield surface. Therefore,
behavior for isotropic hardening after the first loading, even though it additional FE analyses including isotropic hardening are carried out. If
shows the same σ-ε relation up to the initial σmax. When the Chaboche isotropic hardening is included, the δε occurs in the case of σf < σa,
model is used, the strain accumulates in the form of open hysteresis and δε converges to zero as σf approaches the value of σa with in-
loop. It is also applicable for the shakedown phenomenon if the creasing the number of cycles N as shown in Fig. 8. Since the non-zero
steady
initial total
154
S. Koo et al. International Journal of Fatigue 122 (2019) 152–163
800
E, , o, C, from Table 1
5 600
400
(MPa)
MPa
200
1)
1)
0)
(150, 30
(250, 30
(200, 30
a)
0
m,
(
70 -200
0 1 2 3 4 5 6
(%)
Fig. 7. σ-ε curves for various values of (σm, σa) w/o IH.
14
12.5
3.5
(mm)
Fig. 5. FE model for ratcheting simulation; dimensions are selected according
to ASTM E606.
Table 1
Reference material constants for parametric study.
E (GPa) ν σo (MPa) C (GPa) γ Q max (MPa) b
155
S. Koo et al. International Journal of Fatigue 122 (2019) 152–163
0.05 321
pl
E, , o, C, , Qmax, b Qmax 1 exp b
0.04 f o eq
from Table 1 320
0.03 ( m , a ) MPa = (150, 320)
(MPa)
(%)
0.02 319
f
0.01
318
0.00
317
0 5 10 15 20 0 5 10 15 20
N N
Fig. 9. The change of δε (a) and σf (b) with N under (σm, σa) MPa = (150, 320).
3.0 3.0
E, , o from Table 1 = 40 E, , o from Table 1 C = 12 GPa
2.5 = 150, = 350 MPa C (GPa) 2.5 = 150, = 350 MPa
m a m a
2.0 2.0
10 50
(%)
(%)
1.5 12 1.5 40
1.0 1.0
0.5 14 0.5 30
0.0 0.0
0 2 4 6 8 10 0 2 4 6 8 10
N N
(a) (b)
Fig. 10. Effects of kinematic hardening parameters (a) C and (b) γ on δε -N data.
2.5 1.0
E, , o, C, , b from Table 1 Qmax (MPa) E, , C, , Qmax from Table 1
b o,
2.0 - 30 0.8 10
1.5 0.6 50
(%)
(%)
0.5 30 0.2
0.0 0.0
0 10 20 30 40 50 60 0 10 20 30 40 50
N N
(a) (b)
Fig. 11. Effects of isotropic hardening parameters (a) ± Qmax and (b) b on δε -N data.
Table 2 to increase and decrease the size of yield surface, respectively. When
Kinematic hardening parameters for dual backstress Chaboche model. ±Q max is used, conditions such as σf < σo + Qmax and σo + Qmax < σf
case C1 (GPa) C2 (GPa) γ γ can be established.
1 2
0 (single backstress) 12 – 40 – (i) For positive Qmax, if σf < σa < σo + Qmax, δε gradually decreases
1 6 6 40 40
with increasing N and converges to zero. Therefore, condition
2 11 1 40 40
3 6 6 70 10
σo + Qmax < σa should be satisfied to obtain a non-zero steady
4 6 6 80 0 state ratcheting strain (δε steady ≠ 0).
5 11 1 70 10 (ii) For negative Q max, as the condition (σf < σa) satisfies the condition
6 11 1 10 70 (σo + Q max < σa) definitely, δε never converges to zero. However,
156
S. Koo et al. International Journal of Fatigue 122 (2019) 152–163
Fig. 12. (a) σ-ε relations for various cases of Table 2 and (b) the enlargement.
3.0 20
E, , from Table 1 SUJ2
o
2.5 C1, C2, 1, from Table 2
2 15
2.0 m = 150, a = 350 MPa
case
P (kN)
5 10
(%)
1.5
0, 1, 2 specimen 1
1.0 5
specimen 2
0.5 specimen 3
3 0
6 0 2 4 6 8 10
0.0 4
0 20 40 60 80 100 120 (mm)
total = i
i (4)
157
S. Koo et al. International Journal of Fatigue 122 (2019) 152–163
800 1000
= 659 MPa necking point
TS = 777.0 MPa necking point
800 d
600
(MPa) 600
(MPa)
400
400
e
t
200
200
d = 0.1643
0 0
0.0 0.1 0.2 0.3 0.4 0.00 0.05 0.10 0.15 0.20 0.25 0.30
e t
(a) (b)
20 1400
SUJ2 SUJ2 Eqn. (6) with w = 0.6
1200
15 1000
(MPa)
800
P (kN)
10
600
f
5 400
exp. 200
FEA
0 0
0 2 4 6 8 10 0.0 0.2 0.4 0.6 0.8 1.0
pl
(mm) eq
(d) (c)
pl
Fig. 16. (a) σe-εe and (b) σt-εt data converted from P-δ data (c) σf-εeq data from the weighted- average method (c) P-δ data from an experiment and FEA with plastic
properties.
600 0.20
( (%), (MPa)) = 300 MPa (MPa) (%)
initial max
(3.02, 560) 0.18 m a steady
500
(2.55, 550) 0.16 260 0.019
400 (2.36, 540) 0.14 250 0.014
240 0.012
(MPa)
0.12
(%)
300 0.10
0.08 (MPa)
200 a
0.06 260
100 0.04 250
0.02
0 0.00 240
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0 20 40 60 80 100
(%) N
Fig. 17. Experimental results of (a) σ-ε data and (b) δε -N data for various values of σa.
performed for various set of C1, C2, γ 1 and γ 2, that satisfying the re- is satisfied, dσ/dε at the same strain increases as γ 1 – γ 2 increases. The
lation of Eq. (5) as listed in Table 2. σ-ε relation of case 6 is located above that of case 5, and the gap be-
tween them widens with increasing ε. From the fact that γ 1 and γ 2 of
cases 5 and 6 are inverted, it can be said that dσ/dε at the same strain
3.4.1. Stress-strain relation and ratcheting phenomenon with dual increases due to interaction of larger γ i with smaller Ci (here γ 2 in-
backstresses teracts with C2 in case 6).
The effects of dual backstress parameters on the σ-ε relation are Similarly, effects of dual backstress parameters on ratcheting phe-
investigated through FE analyses of monotonic tensile test with nomenon are investigated through cyclic loading FE analyses with the
σmax = 500 MPa as shown Fig. 12. If γ 1 = γ 2 = γ is satisfied, σ-ε curves stress condition σm = 150 and σa = 350 MPa as shown Fig. 13. Identical
for cases 0, 1 and 2 are completely identical regardless of C1 and C2 δε -N curve for the cases 0, 1 and 2 confirms that not only the σ-ε
values. The σ-ε curves for cases 1, 3 and 4, in which C1 = C2 and values relations but also the ratcheting phenomenon are the same regardless of
of γ 1 – γ 2 varies, are located in ascending order, and the gaps among C1 and C2 values if γ 1 = γ 2 = γ is satisfied. δε never changes with
them widen along with increasing ε. This indicates that, when C1 = C2
158
S. Koo et al. International Journal of Fatigue 122 (2019) 152–163
Table 3
Result summary of parametric study on Chaboche model.
Parametric study on Chaboche model Features
increasing N, and the δε values themselves are equal. The δε-N curve of strength σTS and elongation e of AISI 52100 steel are identified as
case 4 shows the δεsteady becomes zero if any γ i is equal to zero. In vice 659 MPa and 35%, respectively. To obtain the σt - εt data after diffusion
versa, δε decreases with increasing N and eventually converges to a non- necking point, the weighted-average method referred by Hyun et al.
zero δε steady for the cases 3, 5 and 6 in which γ 1 and γ 2 both are not (2014) [45] is used to calculate the true stress-strain data at the dif-
equal to zero and different. This is a typical ratcheting behavior. The δε fusion necking, where stress σd and strain εd are 777 MPa and 0.1643,
-N data of cases 5 and 6 show that δε steady decreases when larger γ i respectively. Based on these values, σt - εt data after the necking point is
interacts with smaller Ci (here γ 2 interacts with C2 in case 6). The obtained by varying the weight w from 0 to 1 as follows
plastic strain at the same stress level is inversely proportional to dσ/dε
d
(Fig. 12); as a result, except for case 4 (γ 2 = 0) where δε steady does not t = d w (1 + t d) + (1 w) t
d
occur, the δε steady in all cases tends to be inversely proportional to dσ/ d (6)
dε.
The obtained σt -εt data for w = 0.6, are then converted to σf- eqpl data
as shown in Fig. 16c to input as the plastic properties of AISI 52,100
4. Experiments steel in the FE analyses. Numerical P-δ curve up to fracture are in good
agreement with the experimental results as shown in Fig. 16d.
For both tensile test and ratchetting experiments, the most com-
monly used bearing steel AISI 52100 specimens were fabricated ac- 4.2. Ratcheting experiments
cording to the standards (i) ASTM E8 [43] for tensile tests and (ii)
ASTM E606 [44] for ratcheting tests. As round specimens are suitable With occurrence of Lüders band in AISI 52100 steel, a large de-
for cyclic loading tests especially with compressive loads inherent in a formation occurs at the same load as shown in Fig. 15. If the ratcheting
risk of buckling, the round specimens were selected and then tested in test is performed with load-control, it may cause fatal damages to the
Instron universal testing machine (3367). Customized jigs were manu- experimental instruments. Therefore, the specimen is stretched until the
factured to conduct both the tests in Instron 3367 as shown Fig. 14. The stress level increases above the Lüders band under the displacement-
load data was obtained with Instron load cell, which can measure the control, and then ratcheting experiments are carried out by changing to
load up to ± 30 kN. The deformation within the specimen’s gauge the load-control. Experimental results for σa = 240, 250, 260 MPa with
length was measured with the extensometer (base distance = 25 mm, constant σm = 300 MPa are shown in Fig. 17. As the initial σ-ε curves up
maximum displacement = ±5 mm). For tensile test, the round speci- to σmax are the same for all three cases, only the σ-ε data of σ
mens with gauge length of 24 mm and the displacement rate of 2 mm/ max = 560 MPa is drawn, and the (εinitial, σmax) for three cases are in-
min were considered. The calculated strain rate is about 1 × 10−3 s−1. dicated by gray circles (Fig. 17a). The δεsteady is displayed on the δε-N
Whereas for ratcheting experiments, the round specimens with gauge data (Fig. 17b).
length of 25 mm and the loading rate of 5 kN/min were considered. The
calculated stress rate is about 2.12 MPa/s. It is beneficial to understand 5. Results
that this low loading rate level is more related to thermal cyclic loading
rather than actual rotating rate of bearing steel. 5.1. Parametric study results
159
S. Koo et al. International Journal of Fatigue 122 (2019) 152–163
is a < o ?
Yes No
are
= initial| Exp.,
initial| FEA No
steady | FEA = steady | Exp.
satisfied?
Yes
Fig. 18. Procedure for obtaining the parameters of the Chaboche model.
160
S. Koo et al. International Journal of Fatigue 122 (2019) 152–163
250 0.05
Exp. by Portier et al. (2000)
200 0.04
FE results by
Exp. by Portier et al. (2000) Portier et al. (2000)
(MPa) 150 0.03
Koo et al. (2017)
(%)
FE results by present work
100 0.02
Portier et al. (2000)
Koo et al. (2017)
50 0.01
present work
0 0.00
0.0 0.2 0.4 0.6 0.8 1.0 1.2 0 20 40 60 80 100
(%) N
(a) (b)
Fig. 19. Comparison of simulation results of Portier et al. [46], Koo et al. [16] and present work.
600 0.35
steady|FEA = steady|Exp. = 0.014 %
500 Exp. 0.30
FEA 0.25
400
(MPa)
600 0.30
steady|FEA = 0.011 %
500 Exp. 0.25
steady|Exp. = 0.012 %
FEA
400 0.20 error = 8.3 %
(MPa)
(%)
5.2. Proposed method of parameter acquisition δε-N data, δε-N data are extracted from εtotal-N data. The σ-ε curve up to
σmax = 240 MPa and the δεN data up to N = 100 are compared between
Based on the results of parametric analyses, a simple but effective all the classes as shown in Fig. 19. The simulation results of Portier et al.
method to obtain the parameters of the Chaboche model for ratcheting [46] are very far from the experimental results in both the σ-ε relation
simulation is suggested as follows (Fig. 18). and the ratcheting characteristics; whereas, the simulation results of
Portier et al. [46] conducted ratcheting tests with 316 austenitic Koo et al. [16] show that the σ-ε relation is largely different, but the
steel (AISI 316L) and approximated the results with the Chaboche ratcheting characteristics are very consistent with the experimental
model. Koo et al. [16] also approximated experimental results of Portier result. Contrary to these previous researches, FE analyses based on the
et al. [46] with the Chaboche model. Therefore, the validity of the proposed acquisition method give the σ-ε relation and the ratcheting
acquisition method presented in this paper can be verified by com- characteristics are very close to the experimental results. Hence, the
paring with literature. Since the work of Portier et al. [46] contains no parameter acquisition method proposed in this research enables more
161
S. Koo et al. International Journal of Fatigue 122 (2019) 152–163
600 0.35
steady|FEA = 0.017 %
500 Exp. 0.30
= 0.019 %
FEA 0.25
steady|Exp.
400
(MPa) error = 10.5 %
0.20
(%)
300 E = 200 GPa C1 = 62.1 GPa FEA
= 0.3 C2 = 8.1 GPa 0.15
200
o = 400 MPa 1 = 398.68 0.10
Qmax = - 200 MPa
100 2 = 1.32 0.05 Exp.
b = 120
0 0.00
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0 20 40 60 80 100
(%) N
(a) (b)
Fig. 22. Experimental and FE results of (a) σ-ε data and (b) δε-N data for σm = 300, σa = 260 MPa using the same parameters as Fig. 20.
precise ratcheting simulation with the combined hardening Chaboche method is highly applicable as it can handle the nearby alternative
model. stresses. Therefore, it is simple but effective method for suggesting
acceptable approximations of ratcheting behavior.
5.3. Experimental verification of the proposed method
Acknowledgment
By using the experimental results for the case of σm = 300 and
σa = 250 MPa, the Chaboche model parameters for AISI 52100 steel are This research was supported by the Basic Scien ce Research Program
obtained by the proposed method. The simulated σ-ε relation and δε-N through the National Research Foundation of Korea (NRF-
data are then compared with experimental results as shown in Fig. 20. 2017R1A2B3009706).
Similarly, σ-ε relation and δε-N data for σa = 240, 260 MPa are com-
pared with the experimental results (Fig. 21 and Fig. 22) by using the References
same Chaboche model parameters. Although a wavy curve appears in
[1] Wu H. An empirical non-proportional cyclic plasticity approach under multiaxial
the σ-ε relations when σf converges to σo + Q max, σ-ε curves after 1%
low-cycle fatigue loading. Int J Mech Sci 2018;142–143:66–73. https://doi.org/10.
strain are well approximated and the error of δεsteady is about only 10% 1016/J.IJMECSCI.2018.04.031.
or less for all three cases. It should be emphasized that, in cyclic plas- [2] Zhu Y, Kang G, Yu C. A finite cyclic elasto-plastic constitutive model to improve the
description of cyclic stress-strain hysteresis loops. Int J Plast 2017;95:191–215.
ticity, the negative Qmax does not mean the cyclic softening in nature
https://doi.org/10.1016/J.IJPLAS.2017.04.009.
[47]. In this work, Qmax is mainly used to improve the prediction of σ-ε [3] Agius D, Kajtaz M, Kourousis KI, Wallbrink C, Wang CH, Hu W, et al. Sensitivity and
curves. The comparison results confirm that the Chaboche model optimisation of the Chaboche plasticity model parameters in strain-life fatigue
parameters obtained from the proposed method using dual backstress predictions. Mater Des 2017;118:107–21. https://doi.org/10.1016/J.MATDES.
2017.01.027.
can even simulate the cases of σa values slightly different from the σa [4] Holopainen S, Barriere T, Cheng G, Kouhia R. Continuum approach for modeling
values used in the parameter acquisition. fatigue in amorphous glassy polymers. applications to the investigation of damage-
Since the ratcheting strain of initial cycles is greater than the steady ratcheting interaction in polycarbonate. Int J Plast 2017;91:109–33. https://doi.
org/10.1016/J.IJPLAS.2016.12.001.
state value, it can be an important factor in early structural integrity. In [5] Donzella G, Petrogalli C. A failure assessment diagram for components subjected to
this study, the initial ratcheting strain are overestimated as it focuses on rolling contact loading. Int J Fatigue 2010;32:256–68. https://doi.org/10.1016/j.
the ratcheting strain in steady state. Therefore, the prediction of overall ijfatigue.2009.06.016.
[6] Hassan T, Taleb L, Krishna S. Influence of non-proportional loading on ratcheting
ratcheting strain will be discussed in the future investigation. responses and simulations by two recent cyclic plasticity models. Int J Plast
2008;24:1863–89. https://doi.org/10.1016/J.IJPLAS.2008.04.008.
[7] Kang G, Liu Y, Ding J. Multiaxial ratchetting–fatigue interactions of annealed and
6. Conclusions tempered 42CrMo steels: Experimental observations. Int J Fatigue
2008;30:2104–18. https://doi.org/10.1016/J.IJFATIGUE.2008.06.002.
In this work, we investigated the effect of combined hardening [8] Kim HS, Kim KS. Multiaxial ratcheting behavior of Inconel 718 at elevated tem-
perature. Proc KSME 2008:344–9.
Chaboche model parameters on stress-strain relation and ratcheting [9] Lee HY, Kim JB, Lee JH. Test and analysis of thermal ratcheting deformation for
phenomenon in metals. Tensile test and ratcheting experiments were 316L stainless steel cylindrical structure. Trans Korean Soc Mech Eng A
conducted with the bearing steel material AISI 52100. An effective 2002;26:479–86.
[10] Olver AV. The mechanism of rolling contact fatigue: an update. Proc Inst Mech Eng
parameter acquisition method was then proposed and verified through Part J J Eng Tribol 2005;219:313–30. https://doi.org/10.1243/135065005X9808.
the comparison with the work from literature. Based on the proposed [11] Zhu Y, Kang G, Kan Q, Bruhns OT, Liu Y. Thermo-mechanically coupled cyclic
method, the obtained parameters were Qmax = −200 MPa, b = 120, elasto-viscoplastic constitutive model of metals: theory and application. Int J Plast
2016;79:111–52. https://doi.org/10.1016/J.IJPLAS.2015.12.005.
(C1, C2) GPa = (62.1, 8.1) and (γ 1,γ 2,) = (398.68, 1.32). With these
[12] Abdel-Karim M. An evaluation for several kinematic hardening rules on prediction
values, the σ-ε curve after 1% strain was well approximated and of multiaxial stress-controlled ratchetting. Int J Plast 2010;26:711–30. https://doi.
ratcheting strain in steady state (δε steady) of the FE result was equal to org/10.1016/J.IJPLAS.2009.10.002.
[13] Dafalias YF, Feigenbaum HP. Biaxial ratchetting with novel variations of kinematic
the experimental result in case of (σm, σa) MPa = (300, 250). Similarly,
hardening. Int J Plast 2011;27:479–91. https://doi.org/10.1016/J.IJPLAS.2010.06.
under the other two conditions of (σm, σa) MPa = (300, 240) and (300, 002.
260), the differences of δε steady between FE and experimental results [14] Hamidinejad SM, Varvani-Farahani A. Ratcheting assessment of steel samples under
were 8.3% and 10.5%, respectively. Therefore, these results sufficiently various non-proportional loading paths by means of kinematic hardening rules.
Mater Des 2015;85:367–76. https://doi.org/10.1016/J.MATDES.2015.06.153.
demonstrate the validity of the proposed method for effectively pre- [15] Kang G. Ratchetting: recent progresses in phenomenon observation, constitutive
dicting both σ-ε relation and ratcheting characteristics of metals by modeling and application. Int J Fatigue 2008;30:1448–72. https://doi.org/10.
using only two backstress. Although it is not possible to deal with the 1016/J.IJFATIGUE.2007.10.002.
[16] Koo S, Marimuthu KP, Lee H. A parametric study on Chaboche model to simulate
entire alternative stress range with one parametric case, the proposed
162
S. Koo et al. International Journal of Fatigue 122 (2019) 152–163
163