You are on page 1of 12

Journal of Constructional Steel Research 190 (2022) 107148

Contents lists available at ScienceDirect

Journal of Constructional Steel Research


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

The hysteresis behavior analysis model of BLY160 low-yield-point steel


Bin Zeng a, Tao Shuai a, Ze-Shen Li b, Ke-Shi Zhang a, *
a
Key Lab of Disaster Prevention and Structural Safety of the Ministry of Education, Guangxi Key Lab Disaster Prevention and Engineering Safety, College of Civil
Engineering and Architecture, Guangxi University, 530004 Nanning, China
b
College of Civil Engineering, Shaoxing University, 312000 Shaoxing, China

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

Keywords: Cyclic tests of BLY160 low-yield-point steel were conducted in different strain ranges. Stable hysteresis loops of
Cyclic plasticity constant strain amplitude and multistage strain amplitude were obtained from the tests. The results show that the
Non-Masing behavior material exhibits obvious non-Masing characteristics; the elastic range of the hysteresis loop varies with strain
Hysteresis loops
amplitude, with clearly history-dependent loading. To describe the hysteresis behavior with non-Masing char­
Variable amplitude cycles
Constitutive model
acteristics, a constitutive model was proposed to reduce the tensor model to a one-dimensional model. The model
introduces the historical maximum strain amplitude εmax and the current strain amplitude εcur into the
strengthening parameters as internal variables of the Chaboche constitutive equation to describe the non-Masing
behavior. The steady-state hysteresis loops of strain obtained in simulations using this model were in good
agreement with the experimental results, demonstrating the rationality and validity of the model.

1. Introduction closely related to their cyclic hysteresis behaviors, studying the


description and analysis methods of non-Masing behavior is important
BLY160 low-yield-point steel is used mostly for energy dissipation for anti-fatigue design and the improvement of plasticity and fatigue
and shock absorbance owing to its toughness, high elongation, good theory [15–18].
welding and processing properties, and excellent low-cycle fatigue Based on the Armstrong–Frederick model [19], researchers have
properties [1,2]. Dampers made of low-yield-point steel dissipate studied the cyclic plastic constitutive relation and developed corre­
seismic energy mainly by reciprocating the elastic–plastic deformation sponding cyclic plastic constitutive models [20–22]. Investigations have
of steel in an earthquake. Understanding the hysteresis behavior of focused on cyclic hardening/softening behavior [23–25], dynamic
BLY160 steel is key to evaluating the seismic capacity of damper strain aging behavior [26,27], the Bauschinger effect [28,29], and
members [3–5]. ratchet deformation [30–32]. However, the non-Masing behavior of
Upon transferring the compressive tips of all stable hysteresis loops materials under cyclic loading has not often been studied. One method
from different strain amplitudes to a common origin, if the upper halves for describing the non-Masing phenomena of materials introduces a
of the hysteresis curves at least approximately coincide, the material is strain memory surface to reflect the expansion and contraction evolu­
considered to have Masing characteristics (and is referred to as a Masing tion of the material yield surface caused by cyclic hardening/softening
material) [6], as shown in Fig. 1(a). Whereas if the upper halves of the according to the memory [33]. Jiang et al. [34] used this method to
hysteresis curves cannot coincide, the material has non Masing prop­ simulate the non-proportional hardening of 304 stainless steel. Another
erties, as shown in Fig. 1(b) and (c). Compared with the hysteresis method assumes that the size of the yield surface is constant and con­
behavior of Masing material independent of strain amplitude (for strain nects the back stress with the memory equivalent plastic strain and
cycles), the hysteresis behavior of non Masing material is related to equivalent stress according to the description of follow-up nonlinear
strain amplitude of cycles. hardening to simulate the non-Masing behavior of polycrystalline metals
Metals such as 304 stainless steel, copper, 16MnR steel, and AL6XN [35]. There are also scholars who combine the two methods to construct
stainless steel exhibit non-Masing characteristics to some extent [7–9]. models. For example, literature [24,36] correlates the yield surface
Studies have shown that Masing/non-Masing properties are related to change and back stress evolution with the memory equivalent plastic
several factors [10–14]. As the anti-fatigue properties of materials are strain, realizes the simulation of cyclic hardening/softening hysteretic

* Corresponding author.
E-mail address: zhangks@gxu.edu.cn (K.-S. Zhang).

https://doi.org/10.1016/j.jcsr.2022.107148
Received 5 September 2021; Received in revised form 13 December 2021; Accepted 13 January 2022
Available online 29 January 2022
0143-974X/© 2022 Elsevier Ltd. All rights reserved.
B. Zeng et al. Journal of Constructional Steel Research 190 (2022) 107148

is presented. A one-dimensional model and the corresponding algorithm


for material cyclic plasticity analysis via experiments in the framework
of the Chaboche constitutive relation are proposed by introducing in­
ternal variable parameters in memory with nonlinear iteration. A series
axial strain cyclic test with increasing/decreasing BLY160 steel ampli­
tude is used to verify the rationality of the description and the validity of
the model in terms of the non-Masing effect.

2. Materials, samples, and tests

The chemical composition of BLY160 steel is shown in Table 1. The


mechanical parameters in Table 2 were obtained from a basic mechan­
ical properties test (the elongation was measured according to the
original length of 30 mm). The metallographic structure is shown in
Fig. 2. The specimen was a smooth round bar; the specimen and its di­
mensions are shown in Figs. 3 and 4. The surface roughness of the
specimen was Ra 0.4.
Symmetrical cyclic tests with a constant strain amplitude and
multistage strain amplitude (increase/decrease strain amplitude) were
conducted using an MTS809 tension torsion electro-hydraulic servo fa­
tigue testing machine at room temperature. In test, the bending (plastic
buckling) of the sample should be avoided. The constant strain ampli­
tude test was conducted first; the strain amplitudes were 0.2, 0.4, 0.6,
0.8 and 1.0%; each specimen was tested at only one amplitude. In the
multistage loading test, ascending loading was applied using one spec­
imen with strain amplitudes of 0.2, 0.4, 0.6, 0.8 and 1.0%, stage by
stage; descending loading was applied stage by stage from 1.0% to 0.2%.
In the tests, an extensometer with a gauge length of 25 mm was used
to control the total strain range Δεt (2εa); the cyclic wave was triangular,
the strain ratio R = − 1, and the loading frequency was 0.05 Hz.

3. Experimental analysis and discussion

3.1. Cyclic response characteristics

The relationship between the peak stress and the number of cycles in
the cyclic tests with different strain amplitudes is shown in Fig. 5 (150
loading cycles for each test). The peak cyclic stress increased rapidly at
the beginning and gradually saturated regardless of the strain ampli­
tude. The growth rate of the peak stress increased with an increase in
strain range at the beginning of the cycle. The stress–strain hysteresis
loops with a strain amplitude εa = 0.4% are shown in Fig. 6.
Because the hysteresis loop reached a stable state after 50 cycles,
each stage with constant strain amplitude is set to 50 cycles in the
multistage test of strain amplitude loading (whether the strain ampli­
Fig. 1. Cyclic stress–strain hysteresis loops for different materials: (a) Masing tude increases or decreases). Fig. 7 shows the stress–strain hysteresis
materials; (b) and (c) non-Masing materials. loops for increasing and decreasing strain amplitude loading, stage by
stage.
behavior of materials related to corresponding amplitude, and verifies With increasing strain amplitude stage by stage (Fig. 7(a)), there is a
the effectiveness of the model from material level to structure level. new cyclic hardening at the beginning of each loading stage that soon
Materials generally undergo cyclic hardening or softening and gradually reaches saturation. However, in Fig. 7(b), with decreasing strain
saturate under constant-amplitude cyclic loading. For a single sample of amplitude, the peak stress decreases stage by stage, from softening to
a non-Masing material, the hysteresis loop is significantly different saturation.
under increasing and decreasing amplitude loading conditions. The Fig. 8 shows the stable hysteresis loops corresponding to each strain
description of these phenomena has not been studied. amplitude cycle in cyclic tests with increasing amplitude, decreasing
Considering the change in the elastic range of the material with amplitude, and constant amplitude. In the increasing amplitude cycle,
different strain amplitudes, we study the relationship between the hys­ the stable hysteresis loop at each loading level was essentially the same
teresis loops and the strain amplitude range and loading history by as that of the multisample constant strain amplitude cycle test. In the
combining the test and constitutive model analyses. A constitutive decreasing amplitude cycle, the stress peak value of the hysteresis loop
description of the non-Masing cyclic hysteresis behavior of the materials in the steady state was greater than the stress peak value of the hysteresis

Table 1
Chemical composition of BLY160.
Element C Si Mn P S Alt Nb Ti B N
Percent (%) 0.02 0.06 0.36 0.009 0.005 0.034 0.002 0.045 0.004 0.024

2
B. Zeng et al. Journal of Constructional Steel Research 190 (2022) 107148

loop with increasing amplitude and constant amplitude loading cycles. hardening (movement) of a material can be described by the Chaboche
This result indicates that during a variable amplitude loading cycle, the model, which can simulate the cyclic behavior of a material at a speci­
loading history (the historically experienced strain amplitude) has a fied strain amplitude [37,38]. The model master equation is expressed
strong influence on the hysteresis loop, which is significant for accu­ as.
rately assessing the hysteresis behavior and energy dissipation capacity
of materials. 1. von Mises yield criterion
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
3
3.2. Non-Masing material properties f = (s-α′ ) : (s-α′ )-Ks -R (1)
2
The cyclic stable hysteresis loops of BLY160 steel obtained from the where f is the yield surface function; S is the deviatoric tensor of the
cyclic test with constant strain amplitude are illustrated in Fig. 9(a); the Cauchy stress tensor σ ; α’ is the back stress deviatoric tensor, and (Ks +
compressive tips of all loops are transferred to a common origin. The R) is the radius of the yield surface, where Ks and R are the initial value
hysteresis behavior of BLY160 steel reveals obvious non-Masing and its variation in the radius of the yield surface.
characteristics.
In Fig. 9(b), stable hysteresis loops are translated along the elastic 2. Nonlinear isotropic hardening rule involving non- Masing behavior
section such that the yield points coincide. The upper left halves of the
hardening curves nearly overlap. The translation difference of each The variable R in Eq. (1) describes the expansion or contraction of
hysteresis loop in the figure reflects the change in the elastic range of the the yield surface.
hysteresis curve, which can be used to determine the isotropic
strengthening parameters related to the strain amplitude and to R = Q[1-exp(-bp) ] (2)
construct an improved cyclic plastic constitutive relationship. Thus, the
non-Masing characteristics of BLY160 steel can be described by the where Q is a strengthening parameter related to the loading strain
elastic range variation dependent on the strain amplitude. amplitude, which is the saturation value of R; b is a material parameter,
The material yield is defined using the translational strain method, as and p is the cumulative equivalent plastic strain. When R tends to be
shown in Fig. 10. The elastic range D of the stable hysteresis curve was saturated, the radius of the yield surface tends to be constant (Ks + Q).
determined by the target translational offset strain. With offsets of 50 με, The cumulative equivalent plastic strain can be expressed as
200 με, and 500 με, the corresponding elastic range of the stable hys­ √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
∑ 2 p
teresis loop is identified, as shown in Fig. 11. p= Δp, Δp = Δε : Δεp (3)
Fig. 11 also shows that the elastic range of the cyclic stability hys­ 3
teresis loop is approximately linearly related to the strain amplitude and
where εp is the plastic part of the strain tensorε. Q is a constant only if the
increases monotonically. The variation in the elastic range of the hys­
hysteresis behavior of the material is independent of the strain
teresis loop can be characterized by a linear equation with the strain
amplitude.
amplitude as a variable. By fitting the data points with different offset
strains, the obtained curve slopes are similar. A larger offset strain
3. Nonlinear kinematic hardening rule
produces a greater elastic range.
Nonlinear kinematic hardening reflects the movement of the center
4. Improved Chaboche model reflecting non-Masing of the yield surface through the evolution of the back stress. To approach
characteristics the stress–strain curve obtained from the cyclic test, the back stress and
its rate are expressed as the superposition of multiple components:
4.1. Constitutive model of non-Masing material
˙∑M
(σ − α)
Isotropic hardening (expansion and contraction) and kinematic α̇(k) = C(k) ṗ − γ(k) α(k) p, α = α(k) (4)
(Ks + R) k=1

Table 2 where α(k) is the kth component of the back stress α; α̇(k) is its rate; C(k)
Mechanical properties of BLY160.
and γ (k) are the material parameters representing the initial hardening
E/MPa G/MPa v σb/MPa δ /% ψ /% modulus and its variation, respectively; M is the number of back stress
169,640 65,880 0.29 265 53.5 82.6 components, and M = 2.
Considering the non-Masing property of the materials, the stress–­
strain relationship is related to the range of strain amplitude and loading
history. Thus, the historical maximum strain amplitude εmax (in mem­
ory) and the current strain amplitude εcur (in the current half cycle) are
considered as the internal variables of the isotropic strengthening

Fig. 3. Dimensions of round bar specimen (unit: mm).

Fig. 2. Metallographic diagram of BLY160. Fig. 4. Specimen photo.

3
B. Zeng et al. Journal of Constructional Steel Research 190 (2022) 107148

Fig. 5. Cyclic stress peak vs. number of cycles at different amplitudes.

Fig. 6. Stress–strain hysteresis loops for Δεt /2 = 0.4%.

parameter, defined as.


√̅̅̅̅̅̅̅̅̅̅̅̅
2 ( )
εeq = ε : ε, εmax = max εeq k , (k = 1, 2, …, h) εcur = εeq h , (5)
3

where εeq is the equivalent strain; εeq_k is the extreme value (peak or
valley) of the equivalent strain in the cyclic loading process at the kth
reversal; one cycle equals two reversals, and h represents the current
reversal number.
Introducing εmax and εcur, the isotropic strengthening parameter can
be expressed as

Q = a1 εmax + d1 -(a2 f + d2 )H(f ) (6)

where f = εmax - εcur, and a1, a2, d1, and d2 are the material parameters. In
Eq. (6), H(x) is the Heaviside step function; its operation is defined as
{
0, x ≤ 0
H(x) = (7)
1, x > 0

Fig. 12 shows a diagram of the use of Eq. (6), with the evolution of
parameter Qwith variable amplitude cyclic loading (ε1 is greater than
Fig. 7. Characteristic response loops of BLY160 steel in multistage strain
the initial yield strain).
amplitude loading test: (a) hysteresis loops of increasing amplitude; (b) hys­
teresis loops of decreasing amplitude; (c) variation in cyclic stress peak value
with number of cycles.

4
B. Zeng et al. Journal of Constructional Steel Research 190 (2022) 107148

Fig. 9. Non-Masing characteristics of hysteresis loops of BLY160 steel: (a)


coincidence of lowest points of stable hysteresis loops with different strain
Fig. 8. Comparison of cyclic stability hysteresis loops with constant strain
amplitudes; (b) coincidence of increasing segments of stable hysteresis loops
amplitude and multistage strain amplitude: (a) constant strain amplitude and
with different strain amplitudes.
increasing amplitude; (b) constant strain amplitude and decreasing amplitude.

where Δεp denotes the axial plastic strain increment.


4.2. Integral algorithm of one-dimensional model
Nonlinear kinematic hardening can be expressed in the following
form:
For axial loading (such as the cyclic axial tension compression test),
( )
the constitutive model can be simplified to a one-dimensional model. 2
The stress state at a point in the bar can be determined by the normal 3
σ − α ˙∑M

stress on its section. Based on Eq. (1), the yield equation of the non- α̇(k) = C(k) ṗ − γ (k) α(k) p, α = α(k) (10)
(σ 0 + R) k=1
Masing material constitutive model can be written as
⃒ ⃒ The elastic and plastic strains are recorded as εe and εp, respectively.
⃒ 3 ⃒
f = ⃒⃒σ- α⃒⃒-Ks -R (8) For small strains, the total and incremental strains can be decomposed
2 as.

where f is the yield surface function; σ and α are the axial components of ε = εe + εp , Δε = Δεe + Δεp (11)
the normal stress and the back stress, respectively; Ks is the radius of the
Thus, for step n + 1,
initial yield surface, which can be expressed as Ks = σ0, where σ0 is the
initial yield strength. ε e
n+1 = εen + Δεen+1 , εpn+1 = εpn + Δεpn+1 (12)
The flow law is presented in one-dimensional form as
According to Hooke’s Law,
|Δεp | = Δp (9) ( )
σn+1 = E εn+1 -εpn -Δεpn+1 (13)

5
B. Zeng et al. Journal of Constructional Steel Research 190 (2022) 107148

(a)

Fig. 10. Diagram of yield definition method.

(b)

Fig. 11. Elastic range of material hysteresis loop with different


strain amplitudes.

From step n to n + 1, the initial plastic strain increment is unknown.


Assuming it is 0, the trial stress of step n + 1 can be obtained as
( )
σtrial e p
n+1 = E εn+1 = E εn+1 -εn = σn + EΔεn+1 (14)

In the following calculation, the yield criterion is used to determine


whether the material yields. If the result is elastic, the next incremental
calculation is performed. Otherwise, the plastic strain increment must be
calculated. If an accurate plastic strain increment is obtained, the
required strain and stress can be calculated using Eqs. (11)–(13).
⃒ ⃒
⃒ 3 ⃒⃒
trial
fn+1 = ⃒⃒σtrial
n+1 - αn ⃒-σ 0 -Rn (15) (c )
2
The backward differential evolution method is applied to the back Fig. 12. Evolution of isotropic strengthening parameter Qwith cyclic loading
stress evolution equation, and the following equations can be obtained. path: (a) cyclic loading path; (b) diagram of hysteresis loops of cyclic loading
( ) path; (c) trajectories of Qwith change in εcur.
Δpn+1 2
α(k) (k)
n+1 -αn = C
(k)
σ n+1 -αn+1
(σ 0 + Rn+1 ) 3 (16)
(k) 3(σ 0 + Rn+1 )W + 2Δpn+1 σn+1 U
-γ(k) αn+1 Δpn+1 αn+1 =
3(σ 0 + Rn+1 + Δpn+1 U)
The back stress consists of many terms. ∑
M ∑
M
U= θ(k) C(k) , W = θ(k) αn (k) , (18)

M
αn+1 = α (k)
n+1 (17) k=1 k=1

k=1 1
θ(k) =
Accordingly, it can be derived that 1 + γ(k) Δpn+1

6
B. Zeng et al. Journal of Constructional Steel Research 190 (2022) 107148

Fig. 14. Cyclic hysteresis behavior of multistage strain amplitude simulated by


Chaboche model without considering effect of strain amplitude dependence and
effects of variation in strain amplitude.

Table 3
Material parameters of non-Masing model for BLY160 steel calibrated under the
cycles of strain amplitude 0.5%.
E/GPa v σ0/ Q/MPa b c(1)/ γ(1) c(2)/ γ(2)
MPa MPa MPa

169.64 0.29 80 26 10 95,000 2000 17,000 400

Fig. 13. Hysteresis behavior of multistage strain amplitude simulated in the


proposed model: (a) increasing amplitude; (b) decreasing amplitude.

According to Eqs. (13) and (14),

σn+1 = σtrial p
n+1 -EΔεn+1 (19)

According to Eqs. (18) and (19) with Eqs. (9), (20) can be obtained.
Fig. 15. Comparison of simulated and measured constant amplitude cyclic
⃒ ⃒
⃒ ⃒ stability hysteresis loops with strain amplitude of 0.5%.
⃒σ n+1 -3αn+1 ⃒
⃒ 2 ⃒
[ ( )] (20)
(σ0 + Rn+1 ) -2EΔpn+1 ± 2σ trial
n+1 -3W ( )
= -2(σ 0 + Rn+1 )-2EΔpn+1 -2Δpn+1 U ± 2σtrial
n+1 -3W = 0 (22)
2(σ0 + Rn+1 + Δpn+1 U)

where “±” corresponds to tension and compression, respectively. The where Rn+1 is calculated using Eq. (23):
historical maximum strain amplitude εmax and current strain amplitude Rn+1 = Qm+1 {1-exp[-b(pn + Δpn+1 ) ] } (23)
εcur are updated when the total strain reaches the extremum. The
updated formula is expressed as. Eq. (22) is a one-dimensional nonlinear equation with only one un­
known, Δpn+1. It can be solved using the standard Newton–Raphson
(εcur )m+1 = |ε|, (εmax )m+1 = (εmax )m + gH(g) (21) iterative method, by substituting the value of Δpn+1 into Eqs. (11) and
(13); the state variables are updated simultaneously, and the stress–­
where the initial value (εmax)0 is the initial yield strain, and (εcur)0 = 0;g strain relationship can be solved.
= (εcur)m+1 - (εmax)m. From the analysis, the one-dimensional nonlinear iterative method
When the material meets the yield condition in Eq. (8), according to can be used to simulate the cyclic hysteresis loop for the axial strain
Eqs. (20), (22) is obtained.

7
B. Zeng et al. Journal of Constructional Steel Research 190 (2022) 107148

5. Parameter determination and model validation

The constitutive model in this study was established by improving


the Chaboche model based on the concept of internal variables. For
comparison, the one-dimensional model integration algorithm was used
to simulate the cyclic hysteresis behavior of multistage strain amplitude
with the Chaboche model and the modified constitutive model; the re­
sults are shown in Figs. 13 and 14.
From Figs. 13 and 14, compared with the Chaboche model, the
proposed model reflects the non-Masing characteristics of the materials,
and reflects the influence of the historical loading process on the cyclic
hysteresis curve.

Fig. 16. Comparison of stable hysteresis loops in numerical simulation and


experimental cycles with different strain amplitudes.

Table 4
Isotropic strengthening parameter Q with different strain amplitudes (increasing
amplitude).
εa 0.002 0.004 0.006 0.008 0.010
Q/MPa − 1 18 34 49 67

Table 5
Isotropic strengthening parameter Q with different strain amplitudes
(decreasing amplitude).
εa 0.010 0.008 0.006 0.004 0.002
Q/MPa 67 65 60 51 43 Fig. 18. Calibrated value of isometric strengthening parameter Q varying with
strain amplitude in multistage strain amplitude cycle.

cyclic test. Using this method, the corresponding calculation program Table 6
was compiled to determine the parameters of the non-Masing constitu­ Non-Masing model material parameters for BLY160 steel.
tive model and analyze the constitutive behavior in the constant axial
a1/MPa d1/MPa a2/MPa d2/MPa
strain amplitude and multistage strain amplitude cyclic tests.
8350 − 16.7 3100 − 2.2

Fig. 17. Comparison of cyclic stability hysteresis loops in simulation (Q varies with strain amplitude) and experiment (multistage strain amplitude): (a) increasing
amplitude; (b) decreasing amplitude.

8
B. Zeng et al. Journal of Constructional Steel Research 190 (2022) 107148

Fig. 19. Comparison of cyclic stability hysteresis loops in model simulation and
experiment under constant strain loading.

5.1. Parameter determination

Based on the measured cyclic stable stress–strain hysteresis curves in


Section 2, and referring to the measured cyclic stable hysteresis curves
with constant amplitude loading at each stage in the multistage strain
amplitude test, the model parameters of BLY160 steel were calibrated
using the proposed constitutive model calculation method and trial-and-
error method. This was accomplished in two steps.
First, determining the elastic parameters and kinematic hardening
parameters of the model (neglecting the strain amplitude dependence of
Q). The elastic parameters (E and v) come from the mechanical prop­
erties test. Other parameters are related to the curve characteristics of
the measured hysteresis loops, but cannot be measured directly by the
existing test device. These parameters can be calibrated by trial-and-
error method through successive adjustments. First, the initial values
of the parameters σ0, Q and b are given according to the measured stress-
strain response curve of the constant strain amplitude cycle test, and
then the correction values and the kinematic hardening parameter
values (c(1), γ (1), c(2) and γ (2)) are determined by iterations one by one
according to the measured hysteresis loops. The specific calibration
process can be carried out in the order of σ 0, Q, b, c(1), γ (1), c(2) and γ (2).
When the constant strain amplitude is taken as εa = 0.5%, the model
material parameters can be determined as shown in Table 3, and the
simulation results are shown in Fig. 15.
Cyclic stable hysteresis loops with different strain amplitudes were
calculated using the parameters in Table 3; the results are shown in
Fig. 16. The simulated stable hysteresis loops are in good agreement
with the measured loops only in a small range near the strain amplitude
εa = 0.5%. The calculation error is large for other values, indicating that
the simulation cannot reasonably describe the hysteresis behavior of
non-Masing materials without considering the strain amplitude
dependence.
Second, the isotropic strengthening parameter Q was calibrated for
stable hysteresis loops in the multistage strain amplitude cyclic tests
(Table 3, other material parameters were unchanged). The strength­
ening parameter Q corresponding to different strain amplitude cycles Fig. 20. Comparison of hysteresis behavior in model simulation and experi­
was obtained, as shown in Tables 4 and 5. The simulations were per­ ment under multistage strain loading (increase first and then decrease): (a)
formed using these parameters; the obtained results and the measured variation of cyclic stress peak value with the number of cycles; (b) cyclic sta­
values are shown in Fig. 17. bility hysteresis loops with increasing strain amplitude; (c) cyclic stability
The relationship between Q and strain amplitude in Tables 4 and 5 is hysteresis loops with decreasing strain amplitude.
plotted in Fig. 18. The linear equation expression of Q and strain
amplitude εa can be obtained through linear fitting of the data points.
With Eq. (6), the values of a1, d1, a2, and d2 can be calibrated, as shown
in Table 6.

9
B. Zeng et al. Journal of Constructional Steel Research 190 (2022) 107148

Fig. 21. Comparison of hysteresis behavior in the model simulation and experiment under mixed strain loading (alternate increase and decrease): (a) variation of
cyclic stress peak value with the number of cycles; (b) cyclic stable hysteresis loops.

(including 8 in Tables 3, and 4 in Table 6), with fewer undetermined


Table 7 parameters.
Relative errors between simulated and measured results under different cyclic
loading.
5.2. Validation
Type Constant strain loading (%) Multistage Mixed
of test strain strain
εa = εa εa = εa = εa =
loading loading The proposed model was used to simulate the hysteresis loops of the
0.2 0.6 0.8 0.1
=
cyclic test with constant strain amplitude and variable strain amplitude
0.4
using the model parameters in Tables 3 and 6 respectively. Under con­
R (%) 1.706 0. 0.615 0.530 2.685 0.7035 2.315 stant strain amplitude loading, five groups of tests were conducted, and
929
the strain amplitudes were 0.2, 0.4, 0.6, 0.8, and 1.0% respectively. The
hysteresis loops of the 50th cycle of each test is compared with the
simulation results as shown in Fig. 19. For the case of variable strain
amplitude loading, two groups of tests are included: multistage strain
loading (the strain amplitude increases first and then decreases) and
mixed strain loading. The comparison between the results of the two
groups of tests and the simulation results is shown in Figs. 20 and 21
respectively, including the comparison of the variations in cyclic stress
peaks with the number of cycles and the comparison of the hysteresis
loop at the 50th cycle of the corresponding stage. It can be seen that the
results obtained using the proposed model agree with the experimental
observations. It should be noted that the model neglects the cyclic
hardening/softening process in each loading stage proceeded rapidly at
the beginning and gradually saturates.
In order to verify the accuracy of the simulation results, the relative
errors (R) of the above seven groups of tests as shown in Table 7
comparing respectively with simulation are evaluated by Eq. (24), and
the results are also shown in Table 7.

1∑ M ∑Nm
|σn *-σn |
R= ,
k m=1 n=1 Δσ m (24)

Fig. 22. Comparison of cyclic stability hysteresis loops in model simulation and Δσ m = σmax m -σmin m , (m = 1, 2, …, M)
experiment under constant strain loading.
where, M is the number of hysteresis loops; Nm is the segment number of
It should be noted that the isotropic strengthening parameters in the mth hysteresis loop partitioned by strain equidistance; k is the sum of
Table 6 need be determined under variable strain amplitude loading. the segment number of all hysteresis loops (k = N1 + N2 + … + Nm + …
Theoretically, only one variable strain amplitude cyclic test is required. + NM); σn and σ n* are the measured value and simulated value of the
The variable strain amplitude loading process includes at least one strain stress at the nth point respectively; Δσm is the difference between the
amplitude increasing and one strain amplitude decreasing process, and maximum stress σmax_m and the minimum stress σmin_m in the mth hys­
the initial strain amplitude is less than the maximum strain amplitude. teresis loop.
In addition, compared with the model proposed in [24], which is also The maximum strain amplitude of this test is 1.0% as mentioned in
used to describe the hysteretic behavior of materials related to strain Section 2. In order to verify presented model suitable for simulating
amplitude, the calibration parameters for this model are only 12 cyclic hysteresis behavior with larger strain amplitude range, the large
strain amplitude cyclic test of the same material, where the strain

10
B. Zeng et al. Journal of Constructional Steel Research 190 (2022) 107148

amplitude is 0.5, 1.0, 1.5, 2.0 and 2.5% respectively, in reference [24] [5] A. Ghadami, G. Pourmoosavi, A. Gh Amari, Seismic design of elements outside of
the short low-yield-point steel shear links, J. Constr. Steel Res. 178 (2021).
are simulated and analyzed. In this calculation the model parameters are
[6] G. Masing, Eigenspannungen und verfestigung beim messing, 2nd International
determined according to the test hysteresis loops of literature [24]. Congresso of Applied Mechanics, Orell Fussliverlag, Zurich (1926) 332–335.
Compared with test, the relative errors (amplitude from small to large) [7] W.G. Guo, Plastic flow stresses and constitutive model of four newer naval vessel
of hysteresis loop simulation as showed in Fig. 22 evaluated by Eq. (24) steel, Acta Metall. Sin. 42 (2006) 463–468.
[8] S. Kalnaus, F. Fan, A.K. Vasudevan, Y. Jiang, An experimental investigation on
are 1.838, 1.459, 1.056, 1.325 and 1.616% respectively for each strain fatigue crack growth of AL6XN stainless steel, Eng. Fract. Mech. 75 (8) (2008)
amplitude cycles. It is proved that the presented model can well describe 2002–2019.
the cyclic hysteresis behavior of BLY160 steel with large strain [9] Y. Jiang, J. Zhang, Benchmark experiments and characteristic cyclic plasticity
deformation, Int. J. Plast. 24 (9) (2008) 1481–1515.
amplitude. [10] I. Nikulin, T. Sawaguchi, A. Kushibe, Y. Inoue, H. Otsuka, K. Tsuzaki, Effect of
strain amplitude on the low-cycle fatigue behavior of a new
6. Conclusion Fe–15Mn–10Cr–8Ni–4Si seismic damping alloy, Int. J. Fatigue 88 (7) (2016)
132–141.
[11] P. Arora, S.K. Gupta, V. Bhasin, R.K. Singh, S. Sivaprasad, S. Tarafder, Testing and
In the presented paper, a phenomenological model describing the assessment of fatigue life prediction models for Indian PHWRs piping material
change of elastic range associated with the process of amplitude- under multi-axial load cycling, Int. J. Fatigue 85 (4) (2016) 98–113.
[12] S. Goyal, S. Mandal, P. Parameswaran, R. Sandhya, C.N. Athreya, K. Laha,
increasing and -decreasing cycles is proposed on the basis of Chaboche A comparative assessment of fatigue deformation behavior of 316 LN SS at ambient
model, and its one-dimensional model degenerated from the tensor and high temperature, Mater. Sci. Eng. A 696 (2017) 407–415.
model, suitable for axial tension compression test, is established. By [13] Y. Zhang, C.L. Hu, Z. Zhao, A.P. Li, X.L. Xu, W.B. Shi, Low cycle fatigue behaviour
of a Cr–Mo–V matrix-type high-speed steel used for cold forging, Mater. Des. 44 (2)
introducing the historical maximum strain amplitude and current strain
(2013) 612–621.
amplitude as internal variable parameters, the model can describe the [14] W. Song, X. Liu, F. Berto, S. Razavi, Low-cycle fatigue behavior of 10CrNi3MoV
strain amplitude dependence (the non-Masing effect) of the material high strength steel and its undermatched welds, Materials 11 (5) (2018).
[15] S.C. Roy, S. Goyal, R. Sandhya, S.K. Ray, Low cycle fatigue life prediction of 316 L
hysteresis behavior. And it has the advantages of less parameter deter­
(N) stainless steel based on cyclic elasto-plastic response, Nuclear Eng. Des. 253
mination and simple calibration. Through a series of axial strain cycle (12) (2012) 219–225.
tests and simulation of BLY160 steel, the following conclusions are [16] M. Wang, W. Yang, Y. Shi, J. Xu, Seismic behaviors of steel plate shear wall
drawn: structures with construction details and materials, J. Constr. Steel Res. 107 (4)
(2015) 194–210.
[17] S. Sivaprasad, S.K. Paul, A. Das, N. Narasaiah, S. Tarafder, Cyclic plastic behaviour
1. The cyclic stress development law of BLY160 steel presents obvious of primary heat transport piping materials: influence of loading schemes on
non-Masing characteristics; the elastic range of the stable hysteresis hysteresis loop, Mater. Sci. Eng. A 527 (26) (2010) 6858–6869.
[18] G.S. Hi, Y. Gao, X. Wang, Y. Cui, Energy-based low cycle fatigue analysis of low
loop is linearly related to the strain amplitude. yield point steels, J. Constr. Steel Res. 150 (11) (2018) 346–353.
2. During the stage-by-stage multistage increasing to decreasing strain [19] C.O. Frederick, P.J. Armstrong, A mathematical representation of the multiaxial
amplitude cyclic tests, hysteresis loops between increasing and Bauschinger effect, High Temp. Technol. 24 (1) (2007) 1–26.
[20] M.P. Anderson, G.S. Grest, R.D. Doherty, K. Li, D.J. Srolovitz, Inhibition of grain
decreasing amplitudes with the same strain amplitude exhibited growth by second phase particles: three dimensional Monte Carlo computer
obvious differences and an obvious historical load correlation. simulations - ScienceDirect, Scr. Metall. 23 (5) (1989) 753–758.
3. Applying the presented model, the hysteresis behavior of increasing, [21] P.R. Rios, M.E. Glicksman, Topological theory of abnormal grain growth, Acta
Mater. 54 (19) (2006) 5313–5321.
decreasing and variable amplitude cycles of BLY160 steel observed
[22] Dierk Raabe, Cellular automata in materials science with particular reference to
in tests can be represented well. recrystallization simulation, Annu. Rev. Mater. Res. 32 (1) (2002) 53.
[23] J. Zhou, Z. Sun, P. Kanouté, D. Retraint, Experimental analysis and constitutive
modelling of cyclic behaviour of 316L steels including hardening/softening and
It is necessary to point out that the numerical simulation was per­
strain range memory effect in LCF regime, Int. J. Plast. 107 (2018) 54–78.
formed using the new model in the multistage strain amplitude cyclic [24] L. Xu, X. Nie, J. Fan, M. Tao, R. Ding, Cyclic hardening and softening behavior of
test of BLY160 steel, the hysteresis behavior of the model was in good the low yield point steel BLY160: experimental response and constitutive
agreement with the experimental results. But the model has some lim­ modeling, Int. J. Plast. 78 (2016) 44–63.
[25] M. Dehghani, R. Tremblay, An analytical model for estimating restrainer design
itations in the case of continuous variation of strain amplitude and small forces in bolted buckling-restrained braces, J. Constr. Steel Res. 138 (2017)
strain amplitude of different types of steel (carbon steel with yield 608–620.
platform). [26] M. Madivala, A. Schwedt, U. Prahl, W. Bleck, Anisotropy and strain rate effects on
the failure behavior of TWIP steel: a multiscale experimental study, Int. J. Plast.
115 (2019) 178–199.
Declaration of Competing Interest [27] D. Yu, C. Xu, W. Yu, C. Gang, Thermo-viscoplastic modeling incorporating dynamic
strain aging effect on the uniaxial behavior of Z2CND18.12N stainless steel, Int. J.
Plast. 37 (2012) 119–139.
We declare that the authors have no known competing financial [28] J. Hu, B. Chen, D.J. Smith, P. Flewitt, A. Cocks, On the evaluation of the
interests or personal relationships that could have appeared to influence Bauschinger effect in an austenitic stainless steel—the role of multi-scale residual
the work reported in this paper. stresses, Int. J. Plast. 84 (2016) 203–223.
[29] F. Yoshida, H. Hamasaki, T. Uemori, Modeling of anisotropic hardening of sheet
metals including description of the Bauschinger effect, Int. J. Plast. 75 (2015)
Acknowledgements 170–188.
[30] C.H. Lee, V. Do, K.H. Chang, Analysis of uniaxial ratcheting behavior and cyclic
mean stress relaxation of a duplex stainless steel, Int. J. Plast. 62 (2014) 17–33.
This work was supported by the National Natural Science Foundation [31] Y. Xiong, Q. Yu, Y. Jiang, An experimental study of cyclic plastic deformation of
of China (Nos.11632007 and 11472085) and the Systematic Project of extruded ZK60 magnesium alloy under uniaxial loading at room temperature, Int.
Guangxi Key Laboratory of Disaster Prevention and Structural Safety J. Plast. 53 (2014) 107–124.
[32] G. Kang, Q. Kan, J. Zhang, Y. Sun, Time-dependent ratchetting experiments of
(2019ZDX018). SS304 stainless steel, Int. J. Plast. 22 (5) (2006) 858–894.
[33] E. Tanaka, A nonproportionality parameter and a cyclic viscoplastic constitutive
References model taking into account amplitude dependences and memory effects of isotropic
hardening, Eur. J. Mech. A-solids 13 (2) (1994) 155–173.
[34] Y. Jiang, P. Kurath, Nonproportional cyclic deformation: critical experiments and
[1] T. Yamaguchi, T. Takeuchi, T. Nagao, T. Suzuki, A. Minami, Seismic control
analytical modeling, Int. J. Plast. 13 (8–9) (1997) 743–763.
devices using low-yield-point steel, Nippon Steel Tech. Report 77 (1998) 65–72.
[35] J. Zhang, Y. Jiang, Constitutive modeling of cyclic plasticity deformation of a pure
[2] L.Y. Xu, X. Nie, J.S. Fan, Cyclic behaviour of low-yield-point steel shear panel
polycrystalline copper, Int. J. Plast. 24 (10) (2008) 1890–1915.
dampers, Eng. Struct. 126 (1) (2016) 391–404.
[3] Y. Qi, J. Teng, Q. Shan, J. Ding, W. Yi, Seismic performance of a novel
prefabricated beam-to-column steel joint considering buckling behaviour of
dampers, Eng. Struct. 229 (2021), 111591.
[4] Q. He, C.Y. Yi, W. Wang, H. Tian, Hysteretic features of low yield point steel and its
influence on shear plate damper behavior, Key Eng. Mater. 763 (2018) 718–725.

11
B. Zeng et al. Journal of Constructional Steel Research 190 (2022) 107148

[36] C. Wang, J.S. Fan, L.Y. Xu, X. Nie, Cyclic hardening and softening behavior of the [37] L.T. Hai, G.Q. Li, Y.B. Wang, Y.Z. Wang, A fast calibration approach of modified
low yield point steel: implementation and validation, Eng. Struct. 210 (5) (2020) Chaboche hardening rule for low yield point steel, mild steel and high strength
110220–110221. steels, J. Build. Eng. 38 (2) (2021), 102168.
[38] A. Hy, Z.A. Wen, A. Xz, Z. Zhen, Calibration of Chaboche combined hardening
model for large strain Rangen- ScienceDirect, Procedia Manuf. 47 (2020) 867–872.

12

You might also like