Elastic-Plastic Behavior of Steel Sheets Under In-Plane Cyclic Tension-Compression at Large Strain

International Journal of Plasticity 18 (2002) 633–659
www.elsevier.com/locate/ijplas
Elastic–plastic behavior of steel sheets

under in-plane cyclic tension–compression
at large strain
Fusahito Yoshida*, Takeshi Uemori, Kenji Fujiwara
Department of Mechanical Systems Engineering, Hiroshima University 1-4-1, Kagamiyama,
Higashi-Hiroshima, 739-8527, Japan
Received in final revised form 1 May 2001
Abstract
Elastic–plastic behavior of two types of steel sheets for press-forming (an aluminum-killed
mild steel and a dual-phase high strength steel of 590 MPa ultimate tensile strength) under in-
plane cyclic tension–compression at large strain (up to 25% strain for mild steel and 13% for
high strength steel) have been investigated. From the experiments, it was found that the cyclic
hardening is strongly influenced by cyclic strain range and mean strain. Transient softening
and workhardening stagnation due to the Bauschinger effect, as well as the decrease in
Young’s moduli with increasing prestrain, were also observed during stress reversals. Some
important points in constitutive modeling for such large-strain cyclic elasto-plasticity are dis-
cussed by comparing the stress–strain responses calculated by typical constitutive models of
mixed isotropic–kinematic hardening with the corresponding experimental observations.
# 2002 Elsevier Science Ltd. All rights reserved.
Keywords: B. Cyclic loading; B. Constitutive behaviour; B. Finite strain; B. Mechanical testing; Bau-
schinger effect
1. Introduction
For numerical simulation of sheet metal forming processes, an appropriate con-

stitutive model that properly describes material behavior at large strain is needed.
Especially for the analysis of springback an accurate description of stress–strain
responses during a stress (or strain) reversal is of vital importance. For example, in a
* Corresponding author. Tel.: +81-824-24-7541; fax: +81-824-22-7193.

E-mail address: [email protected] (F. Yoshida).
0749-6419/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S0749-6419(01)00049-3
634 F. Yoshida et al. / International Journal of Plasticity 18 (2002) 633–659
draw-bend process, as schematically shown in Fig. 1, a sheet metal is subjected to

stretch bending and the subsequent unbending when drawn over a die-corner, it is a
process of large-strain deformation. After such a large plastic deformation, during
the unloading process springback takes place almost elastically, but in some cases re-
yielding would occur due to the Bauschinger effect of metals. Another typical
example of cyclic bending of sheet metals is tension leveling or roller leveling process
(e.g. Yoshida and Urabe, 1999). Therefore, in constitutive modeling, we have to pay
much attention to an accurate description of cyclic behavior of metals for both the
ranges of large strain and the subsequent small-scale yielding.
Recently, some researchers pointed out that the numerical results of springback
are strongly influenced by modeling of the Bauschinger effect. The Bauschinger
effect is characterized by two distinct phenomena of stress–strain responses. One is
the so-called ‘transient softening0 which is the smooth transient stress–strain
response at the early stage of stress reversal (hereafter called ‘transient Baushinger
Fig. 1. Schematic illustration of stress–strain path during draw-bend and the subsequent springback.
F. Yoshida et al. / International Journal of Plasticity 18 (2002) 633–659 635
effect’) and the other is the so-called ‘permanent softening’ appearing after the
transient period, as schematically illustrated in Fig. 2. The present authors and their
co-worker (Uemori et al., 1998, 2000) have reported that the transient Bauschinger effect
should be taken into account for an accurate prediction of springback, presenting
the numerical simulation results of hat-bending on a mild steel sheet, as well as V-
bending on two-ply (aluminum/stainless steel) sheet metal laminates. Wagoner et al.
(2000) have stressed that the permanent softening mainly affects springback, show-
ing the results of numerical simulations and experimental data of draw-bending on
aluminum alloy strips. Besdo (2000) discussed the effect of early re-yielding on
springback. Furthermore, Yamaguchi et al. (1998) reported interesting experimental
observations that Young’s modulus drastically decreases with prestrain, which in
turn, would strongly influence springback.
Although numerous experimental data of cyclic plasticity for various metallic
materials, as well as propositions of constitutive models, have been reported, quite
limited numbers of papers on cyclic deformation of sheet metals have been pub-
lished, especially at large strain range. One of the reasons for this situation is that
the conventional tension–compression experiments are almost impracticable for thin
sheet metals because of the buckling of sheets. Instead of tension–compression tests,
Fig. 2. Schematic illustration of stress–strain response during forward and reverse deformation.
one of the present authors and his co-workers (Yoshida et al., 1998) proposed an
inverse analysis to identify material parameters in a constitutive model using cyclic
bending tests of thin sheet metals. However, this approach would also encounter a
difficulty in determining stress–strain responses at large strain, because the strain
level that can be given to thin sheets by cyclic bending tests is rather small. Hu
(1994) has investigated large shear stress–strain responses by performing cyclic tests
of planar simple shear of a mild steel sheet, and based on the experimental obser-
vations, he has proposed a uniaxial constitutive model of cyclic plasticity. However,
we are not confident whether the cyclic shear stress–strain responses are identical
with those under in-plane tension–compression, because during simple shear the
principal orientations of stress rates changes continuously with respect to the material,
and it may affect especially for large simple shear deformation. Kuwabara et al.
(1995) performed experiments on mild steel sheets and an aluminum alloy sheet
under in-plane reverse deformations using a special device for preventing the buckling
of the sheets. Although, their experimental data were at large strain, it seems that
their interests were mostly related to the permanent softening during a reverse defor-
mation, but not either the transient softening or cyclic hardening characteristics.
The present research work is aimed at (i) revealing the deformation characteristics of
large-strain cyclic elasto-plasticity on steel sheets under in-plane tension–compression;
and (ii) evaluating the performance of typical existing constitutive models in
describing such large-strain cyclic plasticity.
In the first part of this paper, we present new experimental data of elastic–plastic
stress–strain responses of two types of steel sheets, an aluminum-killed mild steel
and a dual-phase high strength steel for press-forming operation. Tests were per-
formed under in-plane cyclic tension–compression at large strain (up to 25% strain
for mild steel and 13% for high strength steel). In-plane tension–compression tests
became possible by employing adhesively bonded sheet laminate specimens, as well
as a special device for preventing the buckling of specimens. On the basis of the
obtained experimental results, we summarize the deformation characteristics of the
steel sheets, i.e. the Bauschinger effect, cyclic hardening and its cyclic strain-range
and mean-strain dependencies, as well as the effect of prestrain on Young’s modulus.
In the second part of this paper, we indicate some important points in constitutive
modeling for such large-strain cyclic elasto-plasticity by comparing the results of
numerical stress–strain responses calculated by typical constitutive models of a
mixed isotropic–kinematic hardening plastic potential with the corresponding
experimental observations.
2. Experimental observations
2.1. Experimental procedures
Two types of steel sheets, i.e. an aluminum-killed mild steel sheet (SPCC, 1 mm
thick) and a dual-phase high strength steel sheet (SPFC, 1 mm thick, 590 MPa ultimate
tensile strength) for press-forming, were tested in the as-received state. All specimens
were cut in the rolled direction. In order to prevent the buckling, five pieces of the
sheets were adhesively bonded together before machining so that the thickness of the
specimen was 5.0 mm. The adhesive employed in the present work was an acrylic
adhesive (M-372-20 Hardrock, Denki Kagaku Kogyo, Co. Ltd). Since this adhesive
has high ductility (fracture shear strain is 200%) as well as high strength (maximum
shear stress is 14 MPa), the debonding did not occur during the tests. The thickness
of each adhesive layer was as thin as 0.01 mm, therefore the effect of the adhesive
layer on the stress–strain responses was negligibly small. The shape of the specimen
is illustrated in Fig. 3(a). Cyclic tension–compression tests were conducted by con-
trolling the crosshead speed (1.0 mm s1). A special device (specimen holder) for
preventing buckling was attached to the specimen by coil-springs, as schematically
shown in Fig. 4. To reduce the friction between the anti-buckling device and the
specimen, teflon sheets coated with petroleum jelly (vaseline) were put in their
interfaces. The pressure on the surfaces of the specimen due to the anti-buckling
device was small, less than 0.1 MPa, so that its affect on the determination of stress–
strain responses was negligible. We found that it was possible to apply large strain
reversals of up to 10% to the specimen only when the anti-buckling device was
used. The displacement in the gauge length of 14 mm was measured by a clip-on
Fig. 3. Shapes of specimens for: (a) cyclic tension–compression test; (b) uniaxial tension test.
638
F. Yoshida et al. / International Journal of Plasticity 18 (2002) 633–659
Fig. 4. Schematic illustrations of in-plane cyclic tension–compression tests of sheet metals.
extensometer (see Fig. 4). By comparing two stress–strain curves obtained from a
uniaxial tension test using the laminate specimen [Fig. 3(a)] and the test using a
specimen of 90 mm long with 15 mm wide [see Fig. 3(b)], we confirmed that the
influence of shoulders of the specimen on the measured stress–strain responses is
negligibly small. For cyclic experiments at large strain of over 10% prestrain, the
laminate specimens were prepared with pieces of sheets cut from uniaxially pre-
strained sheets, as schematically shown in Fig. 5, because the extensometer had a
limit in measurable displacement, and furthermore, for such prepared specimens
higher uniformity of strain in the gauge length was guaranteed. The maximum pre-
strains given before strain reversals were 25 and 13% for the mild steel and high
strength steel, respectively. When the strains exceeded these values in uniaxial ten-
sion, the localized necking appeared at strains of about 30–35% in the mild sheet
and 20–22% in the high strength sheet.
For the accurate measurement of Young’s modulus for virgin and prestrained
specimens, strain gauges were also used. Consequently, we found that the unloading
stress–strain responses measured by strain gauges bonded on the uniaxial tension
specimens agreed well with those registered by the clip-on extensometer mounted on
the laminate specimens.
Fig. 5. Preparation of the laminated specimen with pieces of sheets cut from a unaxially prestrained sheet.
2.2. Experimental results and discussions
2.2.1. Cyclic hardening characteristics and the Bauschinger effect

Stress–strain responses during cyclic straining of the strain ranges of 4 and 10%
for the mild steel and the high strength steel are depicted in Figs. 6(a) and (b),
respectively. For the mild steel, the stress amplitude increases with the number of
strain cycles and becomes progressively saturated, while for high strength steel,
Fig. 6. Stress–strain responses during cyclic deformations of strain ranges of 0.04 and 0.10: (a) mild steel
sheet (SPCC); (b) high strength steel sheet (SPFC).
almost saturated stress–strain curves appear after one strain cycle. For both steels,
cyclic hardening is strongly dependent on the cyclic strain range, namely, the larger
the cyclic strain range the larger the stress amplitude. Stress–strain responses during
cyclic straining with several mean strains are illustrated in Fig. 7(a) and (b) for the
mild steel sheet, and in Fig. 8(a) and (b) for the high strength steel sheet. It should be
noted that almost no cyclic hardening is observed in the small cyclic strain range of
2% when having a large prestrain of over 5% [see Figs. 7(a) and 8(a)]. In contrast,
slight cyclic hardening appears for the mild steel in the large strain ranges (5 and
10%), even when having mean strains [see Fig. 7(b)].
The effective stress vs. strain ðj jvs:j"jÞ curves under reverse deformation for the
mild steel and the high strength steel are depicted in Fig. 9(a) and (b). As can be seen
in these figures, the reverse deformation has two distinct stages: i.e. (i) the transient
Fig. 7. Stress–strain responses of a mild steel sheet (SPCC) during cyclic deformations: (a) strain range of 0.02
with various mean strains; (b) strain range of 0.10 with three different mean strains of 0.05, 0.075 and 0.16.
Fig. 8. Stress–strain responses of high strength steel sheet (SPFC) during cyclic deformations: (a) strain
range of 0.02 with various mean strains; (b) three different experiments of (i) cyclic straining of strain
range of 0.10 with mean strain of 0.05; (ii) cyclic straining of strain range of 0.02 with mean strain of 0.04
followed by cyclic straining of strain range of 0.08 with mean strain of 0.04; (iii) strain reversal after pre-
strain of 0.13.
Bauschinger deformation characterized by early re-yielding and the rapid change of

workhardening rate ds/deP and (ii) the permanent softening characterized by the
flow stress offset observed in a region after the transient period, where reverse
workhardening rate is lower than (or almost equal to) that during forward defor-
mation. It is found from Fig. 9(b) for the high strength steel that the stress–strain
curves under reverse deformations, after a transient softening period, become almost
parallel to the forward curve. In contrast, for mild steel [in Fig. 9(a)], workhardening
stagnation is clearly seen in the reverse curves, where the reverse workhardening rate
stays below the forward one. Consequently, the stress–strain curves under reverse
deformations are not parallel to the forward curve. Such abnormal stress–strain
Fig. 9. Effective stress vs. strain curves under reverse deformations: (a) mild steel sheet (SPCC); (b) high
strength steel sheet (SPFC).
curves of large-strain reverse deformations have been already reported for some
other metals, e.g. for a bulk aluminum under compression after tension by Hase-
gawa and Yakou (1975); a bulk aluminum, copper and iron under compression after
tension by Yakou et al. (1985); a bulk copper under compression (or tension) after
tension (or compression) by Christodoulou et al. (1986); bulk steels under cyclic
tension–compression by Wilson and Bate (1986); pipes of various fcc metals (copper,
brass and aluminum alloys) under cyclic torsion by Stout and Rollett (1990); an
aluminum pipe under cyclic torsion by Takahashi and Shiono (1991); a mild steel
sheet under planar simple shear by Hu et al. (1992) as well as cyclic shear by Hu
(1994); and mild steel sheets under in-plane compression (or tension) after tension (or
compression) by Kuwabara et al. (1995). From the physical point of view, the tran-
sient Bauschinger effect is attributed to the motion of less stable dislocations, such as
piled-up dislocations. On the other hand, the permanent softening associated with the
workhardening stagnation is mainly caused by the dissolution of dislocation cell walls
preformed during forward deformation and the formation of new dislocation micro–
structures during reverse deformation (e.g. Hasegawa and Yakou, 1975; Christo-
doulou et al., 1986).
As a measure of the Bauschinger effect, here we examine the stress offset, B ,
defined as the difference in the flow stress between forward and reverse deforma-
tions, as schematically illustrated in Fig. 2. The stress offset B can be divided into
two parts, namely the transient offset, BðtÞ , (so-called ‘transient softening’) and the
permanent offset, BðpÞ , (so-called ‘permanent softening’) which appears after the
transient period, as
B ¼ BðtÞ þ BðpÞ : ð1Þ
Here, the transient stress offset, BðtÞ is defined as the difference between the reverse
stress–strain curve (c–d in Fig. 2) and the extrapolated curve of the region of per-
manent softening (d–g in Fig. 2). Figs. 10(a) and (b) show thus defined transient
stress offset, BðtÞ as a function of reverse plastic strain "^ p . The determination of Point
d inludes a certain ambiguity, however, it would not affect BðtÞ vs. "^p relationship so
much, since the workhardening rate at Point d is so small compared to the transient
softening region. From these figures, it is found that for both steels the initial value
of transient softening when reverse re-yielding just starts (at "^ p ¼ 0 which is indi-
ðt Þ
cated as Bo ¼ 2a in Fig. 2, becomes larger with the plastic prestrain Moreover, the
transient stress offset BðtÞ decreases more slowly with reverse plastic strain when the
prestrain is large, and the BðtÞ vs. "^ p curve approach a saturated curve when pre-
strains become large enough. The permanent softening at the beginning of reverse
ðp Þ
deformation (at "^p ¼ 0) which is indicated as Bo and in Fig. 2, for both steels are
depicted in Fig. 11.
2.2.2. Effect of prestrain on Young’s modulus

If we observe carefully the unloading stress–strain responses, we will find that the
unloading stress–strain paths are not linear, but slightly curved (see Fig. 12). Here,
we define the instantaneous Young’s modulus, E, as schematically illustrated in
Fig. 13, as the stress-strain slope at a certain stress point on the unloading curve. In
Fig. 14(a) and (b), measured instantaneous Young’s moduli of the mild steel and the
high strength steel for various prestrains "po are plotted against the normalized mea-
suring stress point, =o . The data after full strain reversal (" ¼ 0) are also shown.
From these results, it is found that Young’s moduli drastically decrease and
approach their saturated values when prestrains become large enough. This fact
suggests that the plastic flow starts at the very early stage of unloading. From the
results, it can be concluded that the effect of prestrain on Young’s modulus is
expressed as a simple function of accumulated plastic strain, and the influence of
prestrain path seems to be minor. As for the effect of a biaxial prestrain path,
Fig. 10. Transient stress offset vs. reverse plastic strain: (a) mild steel sheet (SPCC); (b) high strength steel
sheet (SPFC).
ðp Þ
Fig. 11. The permanent softening at the begining of reverse deformation B0 as a function of plastic
p
prestrain eo.
Yamaguchi et al. (1998) measured Young’s moduli on several sheet metals after
biaxial stretching prestraining, and found that, for a low carbon steel and type 430
and 304 stainless steels, the decrease of Young’s moduli is expressed also as a unique
function of the effective plastic prestrain independently of prestrain paths.
Such a prestrain dependency of Young’s modulus would much influence the
springback behavior especially when sheets are subjected to a large deformation
during press forming. From the viewpoint of constitutive modeling, we need a certain
size of a yield surface, and within the yield surface the value of Young’s modulus
should not be changed. Therefore, for practical modeling, instead of instantaneous
Young’s modulus, here we introduce a measure termed ‘average Young’s modulus’,
Eav defined as an average slope of the unloading stress–strain curve in a certain
stress range (see Fig. 12). As shown in Fig. 12, we have chosen the following
four stress ranges: (1) 0:75o 4 4 0:95o , (2) 0:50o 4 4 0:95o , (3)
0:25o 4 4 0:95o and (A) 0 4 4 0:95o . Since the stress–strain response just
after the beginning of unloading was highly nonlinear, which would be due to the
viscosity of the steels (refer to Yoshida, 2000), we have chosen 0.95o as the starting
point of average Young’s modulus determination, but not just the stress reversal
point, o . Thus determined average Young’s moduli for the mild steel and the high
strength steel are depicted in Fig. 15(a) and (b), respectively. It is clear that average
Young’s moduli Eav decrease rapidly with increasing prestrain "po and they gradually
approach their asymptotic values. Such a prestrain dependency of average Young’s

modulus is well expressed by the following equation:

Eav ¼ Eo ðEo Ea Þ 1 exp "po ; ð2Þ
where Eo and Ea stand for Young’s modulus for virgin and infinitely large pre-
strained materials, respectively, and is a material constant. The values of these
material parameters are listed in Table 1.
3. Evaluation of constitutive models
3.1. Constitutive models
In this section, we will evaluate the performance of existing typical constitutive

models of cyclic plasticity to describe the stress–strain responses under large-strain
cyclic deformations. From these results, we will summarize the important issues in
constitutive modeling of large-strain cyclic plasticity. Among numerous existing
constitutive models, for the present discussion, we have chosen the framework of
modeling proposed by Chaboche and Rousselier (1983), since it is very popular and
Fig. 12. An example of unloading stress–strain response for the high strength steel sheet. The definition of
average Young’s modulus Ea is defined as an average slope of the unloading stress–strain curve in a
certain stress range, as shown by broken lines.
Fig. 13. Definition of instanteneous Young’s modulus.
even recommended to use for cyclic plasticity analyses in some widely distributed
commercial FE-codes (e.g. Anon., 1997; PAM-STAMPTM, 2000).
The yield function with mixed isotropic–kinematic hardening and the associated
flow rule are given by the equations:
3 : @f :
f ¼ ðs Þ : ðs Þ ðY þ RÞ2 ¼ 0; "p ¼ l; ð3Þ
2 @s
where s and denote the Cauchy stress deviator and the backstress deviator,
respectively, and Y and R are the initial yield stress and the isotropic hardening
:
stress. The plastic strain rate is denoted by "p ,
The evolution equation for the isotropic hardening stress R is
: :
R ¼ mðRsat RÞp ð4Þ
. : pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
: :
where p denotes the effective plastic strain rate defined as p ¼ ð2=3Þ"p : "p , Rsat is
the saturated value of R at infinitely large plastic strain, and m is a material para-
meter that controls the rate of isotropic hardening. This evolution equation yields
Fig. 14. Instantaneous Young’s moduli at various measuring stress points: (a) mild steel sheet (SPCC); (b)
high strength steel sheet (SPFC).
Fig. 15. Average Young’s moduli vs. reverse plastic strain: (a) mild steel sheet (SPCC); (b) high strength
steel sheet (SPFC).
Table 1
Material constants E0, Ea and in the empirical
equation
to express
the effect of pre strain "po on the
average Young’s modulus: Eav ¼ E0 ðE0 Ea Þ 1 exp "p0
E0 ðGPaÞ Ea(GPa)
Case 1 Case 2 Case 3 Case 4
SPCC 206.0 154 152 147 140 30.8

SPFC 206.0 163 159 154 145 61.0
ð
:
R ¼ Rsat ð1 emp Þ; p ¼ pdt: ð5Þ
For simplicity, hereafter we will call the above isotropic hardening rule ‘IH-
model’.
Chaboche and Rosselier (1983) has given the backstress evolution equation as a
summation of M components of the backstress of the Armstrong and Frederick
(1966) type, i ði ¼ 1; 2; . . . ; MÞ as:
X
M
¼ i ð6aÞ
i¼1

: 2 : :
i ¼ C i ai " p i p ; ð6bÞ
3
where ai and Ci are material parameters. Hereafter, we call the above nonlinear
kinematic hardening rule ‘NLK model’. Since a smaller number of material para-
meters in a constitutive model is preferable for practical application of the model,
some commercial FE-codes (e.g. Anon., 1997; PAM-STAMPSTM, 2000) have
recommended to use only one NLK component:

: 2 : :
¼ C " p p : ð7Þ
3
It should be noted that the above kinematic hardening rule can be also written in the
following form

: a :
¼C ð s Þ p : ð8Þ
YþR
As Chaboche and Rousselier (1983) already mentioned, this model is identical with
the two-surface model that assumes the NLK motion of the yield surface within the
bounding surface with the radius of ðY þ R þ aÞ.
Using the above backstress evolution equation, the transient stress offset BðtÞ in
reverse deformation after large prestrain is expressed as a function of reverse plastic
strain "^p , as
p
BðtÞ ¼ a þ 2aeCe^ : ð9Þ
Furthermore, as will be discussed later, one NLK term is not sufficient for the
description of large-strain stress–strain responses, therefore an additional term is
needed. The simplest choice is just adding a linear kinematic hardening component
(‘LK model’), as
¼ 1 þ 2 ; ð10aÞ

: 2 : :
1 ¼ C a"p 1 p ð10bÞ
3
: 2 :
2 ¼ H01 "p : ð10cÞ
3
This type of kinematic hardening rule, the NLK+LK model, was sometimes used
for modeling of small-strain cyclic plasticity (e.g. Yoshida, 1995). This idea is iden-
tical with the two-surface model that assumes LK of the bounding surface. Sets of
material parameters in the IH+NLK model
and the IH+NLK+LK model are [Y,
Rsat, m, a, C] and Y; Rsat ; m; a; C; H01 , respectively.
The Chaboche–Rousselier model has been developed as an infinitesimal theory of
plasticity. Therefore, in order to extend this model into a large strain range, some
modifications of the expressions are necessary. In the finite plasticity model, the plastic
:
strain rate "p is defined as the plastic part of the rate of deformation, Dp and then the
: pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
effective plastic strain rate should be expressed by p ¼ ð2=3ÞDp : Dp . Furthermore, in

the evolution equations for the backstress, the objective rate such
2 as the Jaumann
rate, ,
: : :
should be used instead of e.g. Eq. (7) is written as ¼ C 3 aDp p . However, we
would not argue further the difference between the infinitesimal and finite plasticity
models, since, in the present paper, we only deal with uniaxial stress states.
3.2. Numerical results and discussions
Numerical simulations of stress–strain responses under cyclic deformation were

conducted using the above-mentioned constitutive models. In the calculation, the
effect of prestrain on Young’s moduli, which had been given by Eq. (2), was
involved. First let us examine the stress–strain responses under uniaxial tension fol-
lowed by compression (forward and the subsequent reverse deformation). As an
example, Fig. 16(a) and (b) illustrate the stress–strain responses, for the mild steel
and the high strength steel, calculated by the above mentioned IH+NLK model,
together with the experimental results. As for material parameter identification, first a
and C in the NLK rule [Eq. (7)] had been determined using the experimental data of
the transient stress offset vs. the reverse plastic strain [BðtÞ vs. "^ p in Fig. 10(a) and (b)]
from Eq. (9), and then, parameters Y, Rsat and m were identified such that the cal-
culated stress–strain curves of uniaxial tension give the best-fit to the corresponding
experimental results. For the parameter identification, an optimization technique
based on the iterative multipoint approximation concept (Toropov et al., 1993;
Yoshida et al., 1998) was employed, where a set of material parameters were determined
simultaneously by minimizing the difference between the experimental stress–strain
responses and the corresponding results of numerical simulation. Thus determined
material parameters are listed in the caption of Fig. 16(a) and (b).
Fig. 16. Comparison of effective stress vs. strain curves under reverse deformations calculated by the
IH+NLK model with the experimental results: (a) mild steel sheet (SPCC, Y=120 MPa; C=390; a=43
MPa; Rsat,=166 MPa and m=8.4); (b) high strength steel sheet (SPFC, Y=410 MPa; C=70; a=144
MPa; Rsat= 157 MPa and m=16.0).
It should be noted that the calculated reverse flow stresses, for both steels, are
always larger (in its absolute value) than the experimental results, because in the
IH+NLK model the reverse flow curves always approach their tensile flow curves when
reverse plastic strain becomes large. As a result, the model cannot describe the permanent
softening behavior. In addition, the predicted shapes of reverse stress–strain curves of
the high strength steel in the transient softening period look slightly different from
the experimental results. This can be attributed to the assumption that material
parameters a and C are constant, although in reality they vary with the prestrain, as
mentioned in Section 2.2. In order to overcome the above problems, we may add the LK
component to the IH+NLK model. Furthermore, to describe the prestrain dependent
BðtÞ vs. "^ p curves, the following proposition (Chaboche, 1991) for the IH dependency
of parameters C and a (under the condition of Ca=constant) was introduced.
Cinit
C¼ ; a ¼ ainit ð1 þ R=YÞ; ð11Þ
1 þ R=Y
where Cinit and ainit denote the initial values of C and a. Using these equations, the
decrease of C and the increase of a with the IH evolution is expressed.
The stress–strain responses under forward and the subsequent reverse deforma-
tions calculated by the IH+NLK+LK model, together with Eq. (11), for the mild
steel and the high strength steel, are illustrated in Fig. 17(a) and (b), respectively.
These predictions are improved compared to the results given by the IH+NL model, at
least for the high strength steel, for description of the permanent softening. However,
the prediction of the reverse flow stress for the mild steel is not satisfactory, as
shown in Fig. 16(a). This is because the model assumes the continuous evolution of
Fig. 17. Comparison of effective stress vs. strain curves under reverse deformations calculated by the
IH+NLK+LK model with the experimental results: (a) mild steel sheet (SPCC, Y=100 MPa,
Cinit.=710, ainit=70 MPa, Rsat=135 MPa, m=6.5 and H01 = 80 MPa); (b) high strength steel sheet
(SPFC, Y=408 MPa; Cinit=100; ainit=130 MPa; Rsat= 163 MPa; m=14.5 and H01 =254 MPa).
the isotropic hardening, but not the workhardening stagnation under reverse defor-
mation.
A serious problem is that neither the IH+NLK or IH+NLK+LK model can
simultaneously describe the phenomena of cyclic strain-range and also mean-strain
dependencies of stress amplitude. Fig. 18(a) and (b) show examples of cyclic stress–
strain responses for the mild steel calculated by the IH+NLK+LK model. The
calculated results exhibit apparent overestimations for the stress amplitudes
Fig. 18. Comparison of cyclic stress–strain responses of mild steel sheet (SPCC) under cyclic deforma-
tions calculated by the IH+NLK+LK model with the experimental results: (a) cyclic straining of strain
range of 0.02 with various mean strains; (b) various types of cyclic straining including a large strain range
of 0.10.
especially for cases of small cyclic strain range. This is because in this model the
stress amplitude always reaches the unique value of ðY þ Rsat þ aÞ after a certain
number of strain cycles, independently of cyclic strain range. The numerical simu-
lation for cyclic deformation of high strength steel sheet is shown in Fig. 19. When
we compare the cases of the mild steel and the high strength steel sheets, the corre-
lation between the calculated and experimental results for the high strength steel
looks better than for the mild steel. It may come from the fact that the isotropic
hardening in the high strength steel is less strong compared to the mild steel, and as
a result, the calculated results for the high strength steel is not so strongly affected by
the modeling of the isotropic hardening and its stagnation.
One practical way of modifying such models to describe the cyclic strain-range
dependent workhardening has been proposed by Chaboche (1991). The saturated
value of IH stress, Rsat , is assumed to be a function of cyclic plastic strain range "p as
p
Rsat ¼ R2 þ ðR1 R2 Þe
" ; ð12Þ
where R1, R2, and
are material constants. This model may be efficient for describing
small-strain cyclic plasticity with a constant strain range, however, if we intend to
use it for large-strain cyclic plasticity problems under randomly changing strain-
range conditions, the prediction would not be so accurate.
For a description of workhardening stagnation, an idea of non-hardening plastic
strain surface was proposed by Chaboche et al. (1979) and generalized by Ohno,
(1982). They have introduced this idea in their constitutive models, and conse-
quently, succeeded in describing the cyclic strain-range dependency of stress ampli-
tudes. However, it is hardly possible to use this model for large-strain cyclic
plasticity problems, since, theoretically, we cannot define the Cartesian strain space
when a large shear deformation occurs (see Yoshida and Uemori, 2001). It is worth
Fig. 19. Comparison of cyclic stress–strain responses of high strength steel sheet (SPFC) under various
cyclic deformations calculated by the IH+NLK+LK model with the experimental results.
mentioning that Hu et al. (1992; Hu, 1994) have proposed a uniaxial model that
describes the workhardening stagnation, as well as the cyclic strain-range depen-
dency of stress amplitude. Later, Teodosiu et al. (1997) generalized the model for
multiaxial plasticity. To the authors’ best knowledge, out of these works, none of
models predicts all of the behaviors of large-strain cyclic plasticity described in this
paper. Recently, the present authors (Yoshida and Uemori, 2001) have proposed an
original two-surface model which can well describe the above-mentioned deforma-
tion behaviors including cyclic hardening characteristics, the Bauschinger effect, as
well as the workhardening stagnation.
4. Concluding remarks
Elastic–plastic behavior of two types of steel sheets for press-forming (an aluminum-
killed mild steel and a dual-phase high strength steel) under in-plane cyclic tension–
compression at large strain, have been investigated. It should be emphasized that,
for thin sheet metals, such a variety of experimental data of in-plane cyclic tension–
compression at large strain have never been published before. The present experi-
mental technique that uses adhesively bonded specimens with an anti-buckling
device allowed us to perform such experiments. Specific results of the experimental
observations are summarized as follows:
. During reverse deformation, the transient Bauschinger deformation, which is

characterized by the early re-yielding and the smooth elastic–plastic transition
with rapid change of workhardening rate, is followed by the plastic deforma-
tion with apparent permanent softening.
. For the mild steel sheet, abnormal shapes of reverse stress–strain curves appear
due to the workhardening stagnation. In contrast, for the high strength steel,
they are not visible.
. Cyclic stress amplitudes strongly depend on cyclic strain ranges, as well as the
mean strains. The larger the strain ranges, the larger the saturated stress
amplitudes.
. For both steels, unloading Young’s moduli decrease rapidly with increasing
prestrain, and they approach their asymptotic values when the prestrain
becomes large. This behavior is well expressed by Eq. (2).
. Cyclic behavior of the mild steel sheet under in-plane tension–compression is
quite similar to that of planar shear experiments reported by Hu (1994).
The performance of the IH+NLK and IH+NLK+LK models in describing such

deformation behavior of large-strain cyclic plasticity was evaluated. The results are
summarized as follows:
. One of the biggest problems in the IH+NLK model is that it cannot predict
the permanent softening behavior appearing during reverse deformation, while
it can simulate the transient Bauschinger effect. The IH+NLK+LK model
describes qualitatively the permanent softening, but certain discrepancies are

still found between the calculated reverse stress–strain curves and the corre-
sponding experimental observations. This is mainly due to the lack of model-
ing for the workhardening stagnation.
. Neither the IH+NLK model or the IH+NLK+LK model describes both the
phenomena of cyclic strain-range and mean-strain dependencies of stress
amplitudes. In order to overcome this difficulty, it is necessary to introduce
models for the workhardening stagnation.
References
Anon., 1997. ABAQUS Standards User’s Manual, Version 5.7. Models for Metals Subjected to Cyclic
Loading. Hibbit, Karlson & Sorensen, Inc., pp. 11.2.2.1–11.2.2.5.
Armstrong, P.J., Frederick, C.O., 1966. A mathematical representation of the multiaxial Bauschinger
effect. GEGB report RD/B/N731. Berkeley Nuclear Laboratories.
Besdo, D., 2000. On numerical problems with the simulation of the spring back phase of sheet metal
forming process. In: Khan, A.S., Zhan, H., Yuen, Y. (Eds.) Proc. 8th Int. Symp. on Plasticity and Its
Current Applications. NEAT Press, pp. 17–19.
Chaboche, J.L., Dang-Van, K., Cordier, G., 1979. Modelization of the Strain Memory Effect on the
Cyclic Hardening of 316 Stainless Steels, SMiRT-5, Div. L, Paper No. L. 11/3.
Chaboche, J.L., Rousselier, G., 1983. On the plastic and viscoplastic constitutive equations, part I and II.
ASME Journal of Pressure Vessel Technology, 105, 153–164.
Chaboche, J.L., 1991. On some modifications of kinematic hardening to improve the description of
ratchetting effects. Int. J. Plasticity 7, 661–678.
Christodoulou, N., Woo, O.T., MacEwen, S.R., 1986. Effect of stress reversals on the work hardening
behaviour of polycrystalline copper. Acta Metall 34, 1553–1562.
Hasegawa, T., Yakou, T., 1975. Deformation behaviour and dislocation structures upon stress reversal in
polycrystalline aluminium. Mater. Sci. Eng. 20, 267–276.
Hu, Z., Rauch, E.F., Teodosiu, C., 1992. Work-hardening behavior of mild steel under stress reversal at
finite strains. Int. J. Plasticity 8, 839–856.
Hu, Z., 1994. Work-hardening behavior of mild steel under cyclic deformation at finite strains, Acta
Metall. Mater. 42, 3481–3491.
Kuwabara, T., Morita, Y., Miyashita, Y., Takahashi, S., 1995. Elastic-plastic behavior of sheet metal
subjected to in-plane reverse loading. In: Tanimura, S., Khan, A.S. (Eds). Proc. 5th Int. Symp. on
Plasticity and Its Current Applications, Gordon and Breach Publishers, pp. 841–844.
Ohno, N., 1982. A constitutive model of cyclic plasticity with a non-hardening strain range. ASME J.
Applied Mechanics 49, 721–727.
PAM-STAMPTM, 2000. Release Notes, Vol. 9. Version 2000. Mixed Kinematic-isotropic Hardening
Model, 9, ESI/PSI.
Stout, M.G., Rollett, A.D., 1990. Large-strain Bauschinger effects in fcc metals and alloys. Metal. Trans.
A, 21A, 3201–3213.
Takahashi, H., Shiono, I., 1991. Backlash model for large deformation behavior of aluminium under
torsional cyclic loading. Int. J. Plasticity 7, 199–217.
Teodosiu, C. Duval, J.-L., Haddadi, H., 1997. Modelling the microstructural evolution during large
plastic deformations. In: Bruhns, O.T., Stein, E. (Eds.) Proceedings of the IUTAM Symposium on
Micro- and Macrostructural Aspects of Thermoplasticity, Kluwer Academic, pp. 479–488.
Uemori, T., Okada, T., Yoshida, F., 1998. Simulation of springback in V-bending process by elastoplastic
finite element method with consideration of Bauschinger effect. Metals and Materials 4 (3), 311–314.
Uemori, T., Okada, T., Yoshida, F., 2000. FE analysis of springback in hat-bending with consideration of
initial anisotropy and the Bauschinger effect. Key Engineering Materials 177–180, 497–502.
Toropov, V.V., Filatov, A.A., Polynkin, A.A., 1993. Multiparameter structural optimization using FEM
and multipoint explicit approximates, Structural Optimization, 6, 7–14.
Wagoner, R.H., Geng, L., Balakrishnan, V., 2000. Role of hardening law in springback. In: Khan, A.S.,
Zhan, H., Yuen Y. (Eds.) Proc. 8th Symp. on Plasticity and Its Current Applications., NEAT Press, pp.
609–611.
Wilson, D.V., Bate, P.S., 1986. Reversibility in the workhardening of spheroidised steels. Acta Metall 34,
1107–1120.
Yakou, T., Hasegawa, T., Karashima, S., 1985. Stagnation of strain hardening during reversed straining
of prestrained aluminium, copper and iron. Trans JIM 26, 88–93.
Yamaguchi, K., Adachi, H., Takakura, N., 1998. Effects of plastic strain path on Young’s modulus of
sheet metals. Metals and Materials 4, 420–425.
Yoshida, F., 1995. Ratchetting behaviour of 304 stainless steel at 650
C under multiaxially strain-con-
trolled and uniaxially/multiaxially stress-controlled conditions. European Journal of Mechanics A/
Solids 14, 97–117.
Yoshida, F., Urabe, M., Toropov, V.V., 1998. Identification of material parameters in constitutive model
for sheet metals from cyclic bending tests. Int. J. Mech. Sci. 40, 237–249.
Yoshida, F., Urabe, M., 1999. Computer-aided process design for the tension levelling of metallic strips.
J. Materials Processing Technology 89/90, 218–223.
Yoshida, F., 2000. A constitutive model of cyclic plasticity. Int. J. Plasticity 16, 359–380.
Yoshida, F., Uemori, T., 2002. A model of large-strain cyclic plasticity describing the Bauschinger effect
and workhardening stagnation, Int. J. Plasticity, 18, 661–686.

Elastic-Plastic Behavior of Steel Sheets Under In-Plane Cyclic Tension-Compression at Large Strain

Uploaded by

Copyright:

Available Formats

Elastic-Plastic Behavior of Steel Sheets Under In-Plane Cyclic Tension-Compression at Large Strain

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Elastic-Plastic Behavior of Steel Sheets Under In-Plane Cyclic Tension-Compression at Large Strain

Uploaded by

Copyright:

Available Formats

International Journal of Plasticity 18 (2002) 633–659

Elastic–plastic behavior of steel sheets

Received in ﬁnal revised form 1 May 2001

For numerical simulation of sheet metal forming processes, an appropriate con-

* Corresponding author. Tel.: +81-824-24-7541; fax: +81-824-22-7193.

draw-bend process, as schematically shown in Fig. 1, a sheet metal is subjected to

2.1. Experimental procedures

2.2. Experimental results and discussions

2.2.1. Cyclic hardening characteristics and the Bauschinger eﬀect

Bauschinger deformation characterized by early re-yielding and the rapid change of

B ¼ BðtÞ þ BðpÞ : ð1Þ

2.2.2. Eﬀect of prestrain on Young’s modulus

approach their asymptotic values. Such a prestrain dependency of average Young’s

3. Evaluation of constitutive models

3.1. Constitutive models

In this section, we will evaluate the performance of existing typical constitutive

Fig. 13. Deﬁnition of instanteneous Young’s modulus.

Case 1 Case 2 Case 3 Case 4

SPCC 206.0 154 152 147 140 30.8

3.2. Numerical results and discussions

Numerical simulations of stress–strain responses under cyclic deformation were

. During reverse deformation, the transient Bauschinger deformation, which is

The performance of the IH+NLK and IH+NLK+LK models in describing such

describes qualitatively the permanent softening, but certain discrepancies are

You might also like