You are on page 1of 30

International Journal of Plasticity 19 (2003) 1215–1244

www.elsevier.com/locate/ijplas

Plastic flow for non-monotonic loading


conditions of an aluminum alloy sheet sample
F. Barlata,b,*, J.M. Ferreira Duartec, J.J. Graciob,d,
A.B. Lopese, E.F. Rauchf
a
Materials Science Division, Alcoa Technical Center, 100 Technical Drive, Alcoa Center, PA 15069-0001, USA
b
Centro de Tecnologia Mecânica e Automação, Universidade de Aveiro, 3810 Aveiro, Portugal
c
Departamento de Engenharia Mecânica, Universidade do Porto, 4099 Porto Codex, Portugal
d
Departamento de Engenharia Mecânica, Universidade do Aveiro, 3810 Aveiro, Portugal
e
Departamento de Engenharia Cerâmica e do Vidro, Universidade de Aveiro, 3810 Aveiro, Portugal
f
Génie Physique et Mécanique des Matériaux, ENSPG-INPG (ESA CNRS 5010), BP 46, 38402
Saint Martin d’Hères Cedex, France

Received in final revised form 28 March 2002

Abstract
Non-linear deformation paths obtained using uniaxial tension followed by simple shear
tests were performed for a 1050-O aluminum alloy sheet sample in different specimen orien-
tations with respect to the material symmetry axes. In order to eliminate the time influence,
the time interval between the first and second loading steps was kept constant for all the tests.
Monotonic uniaxial tension tests interrupted during loading were used to assess the recovery
that takes place during this time. In order to eliminate the influence of the initial plastic ani-
sotropy and to compare the results as if the material hardening was isotropic, the flow stress
was represented as a function of the plastic work. The behavior of the material after reloading
was analyzed in terms of dislocation microstructure and crystallographic texture evolutions.
For more quantitative assessment, the full constraints [Int. J. Plasticity 13 (1997) 75] and
visco-plastic self-consistent [Acta Metall. Mater. 41 (1993) 2611] polycrystal models were used
to simulate the material behavior in the non-linear deformation paths. Based on experimental
and simulation results, the relative contributions of the crystallographic texture and disloca-
tion microstructure evolution to the anisotropic hardening behavior of the material were
discussed.
# 2002 Elsevier Science Ltd. All rights reserved.
Keywords: A. Dislocations; A. Microstructure; B. Polycrystalline material; C. Electron microscopy;
C. Mechanical testing

* Corresponding author. Tel.: +1-724-337-3920; fax: +1-724-337-2044.


E-mail address: frederic.barlat@alcoa.com (F. Barlat).

0749-6419/03/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S0749-6419(02)00020-7
1216 F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244

1. Introduction

In order to optimize forming processes using numerical simulations, it is impor-


tant to provide an accurate description of the material stress–strain behavior. Since
forming processes involve non-linear deformation paths under single- or multiple-
step operations, it is important to assess the stress–strain behavior during the
deformation along such complex paths. For instance, the prediction of springback in
sheet when forming loads are removed depends on the plastic behavior of the
material during reverse loading (Geng and Wagoner, in press). At a microscopic
level, strain hardening results from the accumulation of dislocations, and their dis-
tribution depends on the nature of the material and on process conditions (Ronde-
Oustau and Baudelet, 1977; Hatherly, 1983; Juul Jensen and Hansen, 1990, 1991;
Hansen, 1992; Kuhlmann-Wildsdorf, 1992; Hughes, 1992, 1993; Bay et al., 1992). It
is important to understand how strain path changes affect the dislocation accumu-
lation process and what their consequences are on strain hardening.

1.1. Strain path change characterization

The influence that strain path changes produce on the dislocation substructure
and the stress–strain curve depend on the amplitude of the applied change. To
characterize this amplitude, Schmitt et al. (1985) used a parameter 
d"p :d"
¼ 
d"p kd"k ð1Þ

where d" and d"p represent, in vector form, the plastic strain tensors just prior to
and after the strain path change.  ¼ 1,  ¼ 0 and  ¼ 1 correspond to monotonic,
cross and reverse deformations, respectively. The Bauschinger effect, for instance,
corresponds to a strain path change characterized by  ¼ 1.
Usually, a change in strain path during plastic deformation leads to macroscopic
transient effects on the stress–strain curve. Wagoner and Laukonis (1983) have
characterized two types of these transients. Type 1 transient corresponds to a lower
initial flow stress accompanied by an increased hardening rate. Type 2 transient
corresponds to an increased initial flow stress accompanied by a reduced hardening
rate. However, since a lower yield stress has been also associated with lower hard-
ening (Rauch and Schmitt, 1989), these characterization criteria can be relaxed, and
the two kinds of stress–strain curves obtained after strain path changes can be sim-
ply labeled as (Wilson et al., 1990):

1st kind: Transient increase in hardening rate


2nd kind: Transient reduction in hardening rate

1.2. Strain path changes and dislocation structures

The effect of change in strain path on the dislocation substructure has generally
been observed in steel (Rao and Laukonis, 1983; Fernandes and Schmitt, 1983;
F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244 1217

Rauch and Schmitt, 1989; Wilson and Bate, 1994), in aluminum and its alloys
(Hasegawa et al., 1975; Sang and Lloyd, 1979; Lloyd and Sang, 1979; Li and Bate,
1991; Bate, 1993; Wilson et al.; 1990) and in other fcc metals (Gracio et al., 1989;
Zandrahimi et al., 1989; Vieira et al., 1990; Schmitt et al., 1991; El-Danaf et al.,
2001). In all of these materials, the dislocation microstructure induced by the pre-
strain becomes unstable. It is disrupted and dissolved, and a new dislocation
structure typical of the new strain path forms. These transient phenomena
may lead to premature failure, particularly for materials that exhibit the 2nd
kind of behavior (Sang and Lloyd, 1979). These observations show that large
fluctuations of the stress–strain behavior after strain path changes are asso-
ciated with drastic modifications of the dislocation microstructure. Recently, this
type of behavior has been modeled (Teodosiu and Hu, 1997; Mollica et al., 2001),
incorporated into crystal plasticity frameworks (Hiwatashi et al., 1997; Peters et al.,
2001a, Hoc and Forest, 2001; Kalidindi, 2001) and finite element analysis (Li et al.,
in press).
Copper and steel exhibit a similar mechanical behavior after strain path changes
(Rauch and Schmitt, 1989; Fernandes et al., 1993). For cross-loading, the yield
stress after strain path change is higher than the corresponding flow stress on the
monotonic stress–strain curve and the hardening rate is lower. At the micro-
structural scale, in addition to dislocation dissolution, micro-bands have been
observed to develop in both materials. However, micro-band propagation through a
grain boundary appears to occur more frequently for steel, probably because steel
possesses a larger number of slip systems than copper. This may facilitate the con-
dition for micro-band propagation and make steel more prone to microscopic plas-
tic flow localization. Aluminum appears to have a different behavior, especially
when it is alloyed with other elements. In that case, due to the interaction with solute
elements, the dislocation structure does not appear to be well defined (Hughes,
1993). In addition, the effects of reorientation of the back-stress in alloys containing
non-shearable precipitates appear to be the dominant factor for the mechanical
response of the alloy. In this case, the material behavior is strongly linked to the
evolution of dislocation pile-ups at hard obstacles.
It is worth noting that, as defined in Eq. (1),  is a very simple parameter that
characterizes a change of strain path and appears to correlate nicely with the
change observed in the microstructure. However, there are cases where 
alone does not appear to be sufficient. For instance, for biaxial stretching
followed by uniaxial tension or the reverse sequence in pure aluminum, sequences
for which  is the same, it has been reported by Baudelet et al. (1978) that the
microstructure was preserved when uniaxial tension followed biaxial stretching
whereas, for the reverse strain path sequence, it was dissolved. Finally, all the dis-
location structure observations performed above were done on a single material
cross-section (2-dimensional). However, the recent work by Lopes et al. (in press)
shows that observations done on several material cross-sections (3-dimensional) can
present different features.
Because the present paper deals with commercial purity aluminum, the effects of
strain path changes for aluminum alloys are reviewed is more detail.
1218 F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244

1.3. Strain path changes in aluminum alloys

Hasegawa et al. (1975) reported the effects of strain path changes on the disloca-
tion substructures for polycrystalline aluminum. The structure developed during the
first deformation step becomes unstable and dissolves while a new structure typical
of the second deformation step builds up. For pure aluminum or commercially pure
aluminum, Baudelet et al. (1978) reported that the dislocation microstructure after
uniaxial tension consisted in a mixture of rectangular and equiaxed cells. For
balanced biaxial stretching, only equiaxed cells were observed. For a sequence of
balanced biaxial stretching followed by uniaxial tension, the equiaxed cell structure
remained, even after substantial amounts of uniaxial deformation. However, for the
reverse sequence, the heterogeneous dislocation structure developed during the uni-
axial tension pre-strain disappeared, and only equiaxed cells were observed after the
subsequent balanced biaxial deformation. As just mentioned above, it is worth not-
ing that in both sequences, the parameter  was the same (  0:5).
Sang and Lloyd (1979) performed sequences of uniaxial tension in orthogonal
directions, rolling (RD) and transverse (TD), on aluminum. Pre-strains larger than
0.1 resulted in initiation of diffuse strain localization immediately after the strain
path changes. Consistent with these experiments, Wilson (1989) observed that a
reorganization of the dislocation distribution caused a transient decrease in strain
hardening for aluminum AA1050. Li and Bate (1991) conducted strain path changes
for an 1100 aluminum sheet processed to obtain a cube texture. They observed that
strain path changes produced effects more persistent than the obvious transient
effects as the strain hardening rate following a path change was significantly less
than that for monotonic straining at a given flow stress over large deformation
intervals.
Gracio et al. (2000b) studied the mechanical behavior, texture and dislocation
substructure of AA1050 sheet rolled 40% in different orientations and subsequently
deformed in shear parallel to the transverse direction of the as received sheet. Their
experiments corresponded to three values of  (0.5, 0, 0.5). For  ¼ 0:5 and
 ¼ 0:5, the yield stress after reloading was the same as that obtained on the
monotonic curve for the same equivalent strain. A transient period followed by a
flow stress increase occurred for the  ¼ 0:5 strain path change; whereas, softening
occurred after reloading for  ¼ 0:5. For  ¼ 0, the yield stress after reloading was
lower than the corresponding stress on the monotonic curve. This was followed by a
short period of hardening and finally by softening. The dislocation microstructure
consisted of equiaxed, closed cells for all values of , suggesting that the stress–strain
curves did not appear to depend on dislocation structure evolution. This was con-
firmed by the results of a self-consistent polycrystalline model, which showed that
the mechanical behavior could be explained by crystallographic texture evolution. It
has to be mentioned, though, that the value of the pre-strain in this particular work
(about 0.5) was relatively large.
Hasegawa and Yakou (1980) compared the effect of stress reversal (tension fol-
lowed by compression) and recovery thermal treatment (180  C) on an aluminum
single crystal because they noticed that these two processes exert a similar influence
F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244 1219

on stress–strain curves for aluminum. They noted that the dislocation density
decreased after both the recovery treatment and the stress reversal. However, for the
reversal, the density decreased only when the stress level reached the stress achieved
during the initial loading path prior to unloading. Moreover, recovery led to a
coarsening of the dislocation structure; whereas, stress reversal resulted in the dis-
solution of the dislocation structure; i.e., to an increased mean free-path.

1.4. Objectives of present work

In a previous work (Gracio et al., 1999, 2000a; Lopes et al., in press), a commer-
cial purity aluminum sheet 1050-O (AA1050-O) was deformed monotonically in
uniaxial tension and simple shear in different directions with respect to the material
symmetry axes. It was observed that AA1050-O exhibits a strong anisotropic flow
behavior during these two deformation modes, particularly between specimens tes-
ted in directions at 0 (or 90 ) and 45 from the sheet rolling direction (RD). For the
material investigated, the observed anisotropy in strain hardening was explained by
the crystallographic texture and its evolution with strain. The variation in disloca-
tion microstructure was found to be minor and did not appear to have a strong
influence on the strain hardening anisotropy.
The aim of this work is to assess the relative contribution of the dislocation struc-
ture and crystallographic texture on the anisotropy of work-hardening for this 1050-
O aluminum alloy subjected to non-linear deformation paths. Recently, Peeters et al.
(2001a,b) have developed an integrated substructure and texture evolution model to
predict the mechanical and structural behaviors of an IF steel subjected to non-lin-
ear deformation paths. In the present work, such an analytical description of the
dislocation structure evolution was not attempted. The approach used here is based
on a combined analysis of mechanical test results, dislocation substructure obser-
vations, and classical full constraints or self consistent polycrystal predictions.
The behavior of this AA1050-O sheet sample was assessed for non-linear defor-
mation paths consisting of uniaxial tension followed by simple shear. Depending on
the orientation of the specimens with respect to the loading axes, these tests can
describe the entire spectrum of strain path changes from (pseudo-) monotonic to
(pseudo-) reverse loading. The results obtained in the condition of pseudo-reverse
loading were verified by comparison with the condition of reverse loading, using
forward and reverse simple shear tests. The static recovery effect taking place
between the first and second loading steps was estimated using interrupted uniaxial
tension tests.
In Section 2, the materials, experimental procedures and test results are presented.
In Section 3, the material microstructure and texture evolutions are described. As
explained in Section 4, a full-constraint and a visco-plastic self-consistent (VPSC)
polycrystal models are used to simulate stress–strain curve anisotropy and texture
evolution. In Section 5, the mechanical behavior of the material for the different
loading conditions is discussed and analyzed based on dislocation microstructure
observations, crystallographic texture measurements and polycrystal modeling.
Finally, Section 6 summarizes the main results of this work.
1220 F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244

2. Experimental procedures and results

2.1. Material and characterization

An aluminum alloy 1050 sheet sample, 3 mm thick, was investigated in this work.
The sheet was received in the H24 condition and annealed at 343  C (650  F) for
about 45–60 min to ensure that the entire sheet was at temperature. Then, the
material was pulled out of the furnace and allowed to cool in air. Optical micro-
scopy showed that the material was fully recrystallized with equiaxed grains of
about 30 mm diameter. The chemical composition was obtained by re-melt analysis
(Table 1). The material was characterized using monotonic uniaxial tension and simple
shear tests. The crystallographic texture was measured using X-ray techniques and
orientation imaging microscopy (OIM). The dislocation microstructure was observed
after plastic deformation on a transmission electron microscope (TEM). Standard uni-
axial tension tests using specimens oriented at 0, 45 and 90 from the rolling direction
(RD) were conducted to measure the monotonic stress-strain behavior up to a strain
corresponding to the uniform elongation. The r values (width to thickness strain
ratio) were measured and found to be constant during the deformation. Experi-
mental details for this characterization work are given in Lopes et al. (in press).

2.2. Interrupted tension tests

TD uniaxial tension tests were conducted up to strains of about "  0:07 and " 
0:14 with standard specimens at a strain rate of about 0.001 s1. These specimens
were then unloaded and deformed again up to the uniform elongation after sitting
for 2 days at room temperature. This corresponds to the amount of time between the
two steps of the non-linear strain path experiments. These tests were carried out to
assess any possible recovery effect that might be occurring at room temperature. The
curves presented in this and subsequent sections were duplicated and the results were
found to be very reproducible.
Fig. 1a shows the duplicate TD uniaxial tension hardening curves for specimens
that were reloaded 2 days after being strained to " ¼ 0:07 and 0.14. These curves are
compared to the duplicate monotonic tension curves measured without interruption.
This plot indicates that, on reloading, the yield stress does not reach the flow stress
of the reference curve. Then, the strain hardening that follows yielding is relatively
high during a short transient period after which, it becomes similar to the strain
hardening of the reference curve. However, the flow stress does not quite reach the
flow stress of this reference curve.

Table 1
Chemical composition of AA1050 measured by remelt analysis

Si Fe Cu Mn Mg Cr Ni Zn Ti Ga V

0.089 0.28 0.002 0.001 0.001 0.001 0.003 0.005 0.011 0.016 0.007
F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244 1221

Fig. 1. Hardening curves in TD uniaxial tension for AA1050-O, Monotonic deformation and interrupted
deformation with unloaded specimens for 2 days at two different strain levels, 0.07 and 0.14. (a) True
stress vs. true strain curve; (b) True stress vs. plastic work per unit volume.
1222 F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244

2.3. Plastic work analysis

In order to compare the hardening curves either after monotonic deformation or


after a pre-strain in a different path, it is necessary to transform the stress and strain
data using effective quantities  and d". However, the definition of equivalence is
not a simple matter. In most of the work described in Section 2, the Von Mises
effective strain was used, but this quantity is arbitrary. It is well known, for instance,
that the Von Mises criterion does not fit the behavior of aluminum alloys very
accurately (Hosford, 1993). Moreover, the anisotropy effects due to the initial ani-
sotropy (or crystallographic texture) are not taken into account. A better equiva-
lence can be obtained by representing the stress as a function of the plastic work per
unit volume (Khan and Huang, 1995), and by converting the pre-strain to a corre-
sponding pre-work. The plastic work for any deformation mode can be calculated
by numerical integration of the differential expression

dw ¼ d" ¼ ij d"ij ð2Þ

In this manner, for any initial anisotropy, the reloading curve for a given strain
path should exactly follow the monotonic curve if the material was to deform under
the condition of isotropic work-hardening. Therefore, using the flow stress as a
function of the plastic work allows a direct comparison between the isotropic hard-
ening assumption and the real hardening exhibited by the material under study. The
choice of a different equivalent quantity would mostly affect the position of the
reloading yield stress with respect to the monotonic hardening curve. Transient
effects in strain hardening would still be observable even with a shift in the calcu-
lated effective pre-strain. It can be noticed that the stress–work curve and the stress–
strain curve have very similar, although not identical shapes, as illustrated by Fig. 1a
and b relative to the interrupted tension test. In other words, the relationship
between strain and plastic work is almost linear.

2.4. Tension—simple shear tests

In order to conduct non-linear deformation experiments, uniaxial tension tests in


the transverse direction (TD) were carried out to pre-strain the sheet before con-
ducting the shear tests. Large uniaxial tension specimens of 300 mm in gauge length
and 60 mm in width were deformed to one of the two pre-strain levels, "  0:07 or
"  0:14. The cross-head speed for the large tension specimens was 20 mm/min,
resulting in approximately the same strain rate as that used for the standard tension
specimens (0.001 s1). The subsequent shear tests were conducted at 45, 90 and 135
from the tension direction (Fig. 2a) using specimens with dimensions of 40403
mm (Fig. 2b and c). In these tests, shear strains are effectively imposed in a 4083
mm region of the shear specimen. Three levels of subsequent shear strain (" ¼ =2)
were achieved for these subsequent tests, "  0:015, "  0:15 and "  0:30, and all of
these testing conditions were duplicated.
F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244 1223

2.4.1. Pseudo-monotonic loading


Fig. 3 shows the stress–work curves obtained after uniaxial tension (TD) pre-
strains of 0.07 and 0.14 followed by shear at 45 , i.e., in condition of pseudo-
monotonic deformation for which, =0.85. These curves are compared to the
monotonic 45 shear hardening curve. This figure shows that, on reloading, for both
pre-strains, the material yields at a lower stress and subsequently hardens at a higher
rate compared to the monotonic curve. Then, after a small amount of plastic work
(corresponding to a few percent strains), the strain hardening becomes similar to
that of the monotonic curve. For the smaller pre-strain, the flow stress reaches the
level of the monotonic curve, whereas, it does not for the larger pre-strain.

2.4.2. Cross loading


Fig. 4 shows the stress–work curves obtained after TD uniaxial tension pre-strains
of 0.07 and 0.14 followed by shear at 90 , i.e., in the condition of cross-loading
deformation ( ¼ 0). 90 throughout this paper denotes the direction of shear with
respect to the tensile direction. Because the latter is always the TD, the direction
denoted by 90 is, in fact, superimposed with the RD. Again, Fig. 4 shows that on
reloading, for both pre-strains, the material yields at a lower stress and subsequently
hardens at a higher rate compared to the monotonic curve. However, the transient

Fig. 2. (a) Shear test configuration (RD is the rolling direction, TD is the transverse direction) and shear
specimens. (b) Schematic; (c) perspective (first specimen type); (d) perspective (second specimen type).
1224 F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244

period is longer than that for pseudo-monotonic loading and the reloading curve
reaches or even gets to higher levels than the monotonic curve.

2.4.3. Pseudo-reverse loading


Fig. 5 shows the stress–work curves obtained after uniaxial tension (TD) pre-
strains of 0.07 and 0.14 followed by shear at 135 , i.e., in the condition of pseudo-
reverse deformation for which, =0.85 (r value is 0.87). Again, for both pre-
strains, the material yields at a lower stress and subsequently hardens at a higher
rate compared to the monotonic curve. However, for both pre-strains, after a short
transient period of high strain hardening, another transient period occurs where the
strain hardening becomes lower than that of the monotonic curve. Then, the strain
hardening tends to increase again to the level of the monotonic stress–work curve.
However, in both pre-strain cases, the flow curves obtained after the strain path
changes remain below the monotonic flow curve.

2.4.4. Pseudo-loading conditions


Fig. 6 represents the stress–work curves in simple shear after reloading for pseudo-
monotonic and pseudo-reverse loadings at two pre-strain levels (0.07 and 0.14). As
discussed later in this paper, because grain rotation is not significant within a strain
range of a few percent, the differences observed in the two simple shear tests at each
pre-strain level are necessarily due to the different types of dislocation interaction in

Fig. 3. Stress vs. plastic work curves obtained in TD uniaxial tension at strains of 0.0, 0.07 and 0.14
followed by 45 simple shear. Upper horizontal axis indicates approximate shear strains.
F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244 1225

these tests. This is experimental evidence showing the influence of the dislocation
structure on the mechanical behavior of a material after a strain path change.
Beyond a certain strain, texture evolution can be different for simple shear tests at 45
and 135 from the RD because these directions are not material symmetry axes.
Nevertheless, Fig. 6 shows that the pseudo-reverse curve tends to recover to the level
of the pseudo-monotonic curve at larger subsequent plastic work (strains). More
issues concerning this curve will be discussed in Section 4.

2.5. Forward and reverse simple shear tests

To confirm the work hardening stagnation for pseudo-reverse loading, 45 and
TD forward and reverse simple shear tests were conducted with reversals at shear
strains of 0.07, 0.15 and 0.22. In order to decrease the influence of the free ends, a
longer specimen (60 mm) with a narrower deformation zone (3 mm wide) was used
(Fig. 1d). Previous studies had shown that the width to thickness ratio of the shear
zone must be larger than about 3 in the simple shear test. Therefore, the 3 mm thick
sheet was thinned down to 1 mm by material removal on each side of the sheet.
Fig. 7 shows the stress–work curves for forward and reverse simple shear experi-
ments conducted in the TD direction. These curves confirm the material behavior
observed in pseudo-reverse loading, characterized by a lower yield stress and a
transient effect characterized by a stagnation of the work hardening. This type of

Fig. 4. Stress vs. plastic work curves obtained in TD uniaxial tension at strains of 0.0, 0.07 and 0.14
followed by 90 simple shear. Upper horizontal axis indicates approximate shear strains.
1226 F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244

strain hardening stagnation is usually not observed for stronger aluminum alloys
containing solutes and shearable precipitates (Rauch et al., 2001), but a similar
phenomenon has been observed for high strength aerospace alloys containing non-
shearable precipitates (Stoltz and Pelloux, 1976).

3. Microstructure and texture evolutions

3.1. Dislocation structure for monotonic loading

All the TEM observations were made on grains representative of the texture of the
material, i.e., near the cube orientation. Although based on the experience of the
TEM operator only, the micrographs shown in this paper are statistically repre-
sentative of the numerous observations made in this work.
Fig. 8a and b show the dislocations structures after monotonic uniaxial tension
and simple shear, respectively. It was shown in the previous work by Lopes et al. (in
press) that the cells are equiaxed for conditions close to axisymmetric deformation,
i.e., when two of the principal strains are equal, like TD uniaxial tension for which
the r value (width to thickness plastic strain ratio) is 0.87, Fig. 8a. For shear or near
plane strain conditions (i.e., uniaxial tension with low r value), the material exhibits
straight dislocation walls, leading to rectangular cells with a high aspect ratio

Fig. 5. Stress vs. plastic work curves obtained in TD uniaxial tension at strains of 0.0, 0.07 and 0.14
followed by 135 simple shear. Upper horizontal axis indicates approximate shear strains.
F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244 1227

(Fig. 8b). It is worth noting that, although there is no systematic rule, the orienta-
tion of the wall generally depends on the slip plane position and the nature of the
applied load.

3.2. Dislocation structure for non-monotonic loading

For large shear strains of the order of "  0:30, it was observed that, whatever the
direction of the subsequent simple shear direction (45, 90 or 135 ), the dislocation
structure obtained after uniaxial tension was completely erased, and that the final
dislocation structures was that of simple shear (Fig. 8b). Fig. 9 a–c show dislocation
cells after uniaxial tension (TD) up to "  0:14 followed by simple shear up to
strains of " ¼ =2  0:15. After the sequence TD tension followed by 45 shear up
to a strain of about 0.15, the new dislocation structure is already in place (Fig. 9a),
suggesting a reorganization into walls on the most active planes in simple shear that
were already active in the build-up of the initial cell structure. After TD uniaxial
tension followed by 90 simple shear (Fig. 9b), up to a strain of about 0.15, the dis-
location structure is made of the superimposition of the previous structure devel-
oped in uniaxial tension (Fig. 8a) with a new structure that is typical of the final
deformation mode (shear, Fig. 8b), requiring the production of dislocations with
new Burgers vectors. Finally, after tension followed by 135 shear to the same strain
level of 0.15, the relatively sharp cells or walls which develop in monotonic defor-

Fig. 6. Stress vs. plastic work curves for 45 and 135 simple shear obtained after TD uniaxial tension at
strains of 0.07 and 0.14.
1228 F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244

mation (Fig. 9a) appear to be blurred, and many single dislocations between the
walls or inside the cells can be observed, suggesting the dissolution of the structure
that was developed during the pre-strain. In both 90 and 135 subsequent simple
shear, the final dislocation structure is not establish after a subsequent strain of 0.15,
in contrast with 45 subsequent simple shear. These observations can be correlated
with the transient hardening stage, which appears to last for a shorter strain (plastic
work) range for subsequent simple shear tests at 45 compared to the tests con-
ducted at 90 and 135 . The micro-localizations reported, for instance by Lopes et al.
(1999) and Nesterova et al. (2001) in steel, were not observed in this AA1050-O
sheet.

3.3. Texture evolution

Fig. 10 shows the (111) pole figures measured by OIM for the non-deformed
material (Fig. 10a) and for the specimens that were subjected to TD uniaxial tension
up to a strain of 0.14, followed by 45 and 135 simple shear up to a shear strain of
0.3, Fig. 10b and c, respectively. All these figures were represented in the plane
containing the shear direction (SD parallel to the axis at 45 from the RD) and the
sheet normal direction, which remain orthogonal under shear deformation. Despite
the fact that the reference frame is not conventional, Fig. 10a shows that the initial
texture consists mainly of the cube {100} < 001 > component with other minor

Fig. 7. Stress vs. plastic work for forward and reverse 45 simple shear tests with different reversal strains
(0.08, 0.15 and 0.22).
F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244 1229

components. Fig. 10b and c shows that the (111) pole clusters have shifted with
respect to their initial position and occupy about the same position. However, the
texture resulting from 135 simple shear appears more diffuse than the texture
resulting from 45 simple shear.

4. Polycrystal modeling

The full constraint, rate independent, polycrystal model (Gambin and Barlat,
1997) and the visco-plastic self-consistent (VPSC) polycrystal code, based on the
theory of Lebensohn and Tomé (1993), were used to predict the stress–strain curves
as crystallographic texture evolves with increasing plastic deformation.

Fig. 8. TEM micrographs of typical dislocation structures developed after plastic deformation in mono-
tonic loading: (a) uniaxial tension with T.A. as tensile axis, " ¼ 0:12; (b) simple shear " ¼ 0:30.
1230 F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244

Fig. 9. TEM micrographs of typical dislocation structures developed after plastic deformation in TD
uniaxial tension up to a strain of 0.14, followed by simple shear up to a strain of 0.15 in different direc-
tions; (a) 45 shear illustrating cell reorganization; (b) 90 shear showing cell superimposition; (c) 135
shear showing cell dissolution.
F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244 1231

Fig. 10. Experimental (111) pole figures for (a) undeformed material and for specimens deformed in TD
uniaxial tension up to a strain of 0.14, followed by simple shear up to a shear strain of about 0.3; (b) 45
shear; (c) 135 shear.
1232 F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244

4.1. Full-constraint model

The rate independent formulation is based on the single crystal behavior descrip-
tion by a yield function and its associated flow rule (Gambin, 1991, 1992)
( )1=a
X s a
¼’ 1=a
¼   ¼1 ð3Þ
 s 
s c

s ¼ bsi nsj ij0 is the shear stress on a system ‘‘s’’ defined by the slip plane n and the
slip direction b, resulting from the deviatoric stress tensor 0 . ‘‘a’’ is an exponent
which regularizes the singularity of the Schmid description on the single crystal. cs is
called the critical resolved shear stress of the system ‘‘s’’, although it is different from
a threshold stress. cs reduces to a threshold stress in the limit case a!1, which
corresponds to the classical  Schmid
 crystal. The associated flow rule gives
  the rate of
deformation tensor D ¼ dij tensor and the plastic spin tensor X ¼ !ij 1
: !
l @ @
dij ¼ þ
2 @ij0 @ji0
: !
l @ @
!ij ¼  ð4Þ
2 @ij0 @ji0

Note that, because the single crystal yield function


: is a first degree homogeneous
function, it is simple to show that the coefficient l is equal to ij0 dij and that the shear
rate on each system is given by
:  
: : @ l s  s a2
 s ¼ l s ¼ s s  s  ð5Þ
@ c c c

In this work, the critical resolved shear stress cs was assumed to be the same for all
the systems and to increase isotropically as a function of the accumulated shear
strain according the Voce law

cs ð Þ ¼ c ð Þ ¼ A0  B0 expðC0 Þ ð6Þ

For the simulations, the coefficients A0 , B0 and C0 were determined by a trial-and-


error method, repeating the polycrystal simulations until the experimental TD mono-
tonic simple shear stress–strain curve was satisfactorily predicted. These coefficients
were found to be 35.0, 26.0 and 3.0 MPa, respectively, which was slightly under-
estimating hardening at large strains. The exponent ‘‘a’’ was chosen to be 32, which
makes the single crystal behavior very similar to that described by the Schmid law.

1
Note: derivatives obtained without assuming the symmetry ij0 ¼ ji0 .
F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244 1233

4.2. Visco-plastic self-consistent model

The visco-plastic self-consistent method was used with a strain rate sensitivity
:
coefficient  ¼ 20 to relate the shear strain rate  s to the shear stress s on a slip
system


: : s
s ¼ 0 s ð7Þ
c
:
 0 is a reference shear strain rate and cs reduces to the classical critical resolved
shear stress when the material becomes rate independent ( ! 1). Assuming the
same critical shear stress on all the systems, the single crystal shear stress-accumu-
lated shear strain relationship ( c  ) proposed by Tomé et al. (1984) is used in this
work


0
c ¼ 0 þ ð 1 þ 1 Þ 1  exp  ð8Þ
1

The coefficients 0 , 1 , 0 and 1 that were determined in the previous work (Lopes
et al., in press) to simulate monotonic tensile and shear stress–strain curves were also
used in the present simulations.

0 ¼ 8:75 MPa; 1 ¼ 22:0 MPa; 0 ¼ 240:0 MPa; 1 ¼ 11:5 MPa

The tangent modulus method was selected to describe hardening for the polycrystal
model. This method leads to predicted pole figures in monotonic deformation that
are in good qualitative agreement with experimental pole figures (Lopes et al., in
press).

4.3. Symmetry considerations

Rauch (1998) reviewed in details the implications of material and loading sym-
metries on simple shear test results. In general, due to the stress tensor symmetry,
simple shear at  (the reverse direction of ) is equivalent to simple shear at  þ 90 ,
which is called the dual test. As a result, from a polycrystalline point of view, shear
at 45 is the same as shear at 135 . Therefore, in this work, simple shear tests at 45,
45 and 90 from the tensile direction were considered. Because of the orthotropic
nature of the sheet, shear at 0 and 0 are equivalent. This is also true for shear at 90
and 90 . In order to represent the material texture, a sampling of about 500 OIM
measured orientations was selected. In order to enforce strict orthotropic symmetry,
four variants of each individual orientation were considered, giving a total of about
2000 orientations for the simulations. The velocity gradient tensors used to simulate
TD uniaxial tension, and 45, 45 (135) and 90 simple shear are listed in Table 2. The
simple shear stress formula that can be compared to the experimental shear stress is
given in the Appendix.
1234 F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244

4.4. Simulation results

Fig. 11 a shows the simulated hardening curves using the full constraints model
for uniaxial tension (TD) followed by 45, 90 and 135 simple shear tests. These
results show that the reloading yield stress is similar to the yield stress for the
monotonic test at an equivalent amount of plastic work. 45 and 135 simple shear
test simulations lead to about the same stress–strain behavior. This was expected at
a small amount of shear strain from considerations mentioned earlier (same initial
starting material pre-strained in TD tension) but not at larger strains because simple
shear at  is generally not the same as simple shear at  (the reverse direction of ).
However, this result is reasonable because, due to the strong cube texture, the
material exhibits almost othotropic symmetry with respect to the axes at 45 from
the RD and TD. Compared to monotonic simple shear, 45 and 135 simple shear
tests after TD tension lead to slightly higher hardening, contrasting with 90 simple
shear test after TD tension, which leads to slightly lower hardening. Fig. 11b shows
the simulation results using the VPSC model for the same loading conditions. This
figure shows that the VPSC model leads to the same trends as the full constraint
model, and that the choice of a homogenization technique or a viscosity effect for
polycrystal calculations does not seem to influence the outcome of this study. This
means also that grain interactions with the average matrix behavior has little influ-
ence on the anisotropic hardening behavior, although during a strain path change,
local grain interactions might influence the way dislocations reorganize.
Fig. 12 shows the simulated (111) pole figures using both polycrystal models for
monotonic TD uniaxial tension and for cases where the TD tension is followed by
45 and 135 simple shear. All of these predicted pole figures compare favorably with
the experimental pole figures (Fig. 10), although the experimental texture appears to
be more diffuse, in particular for 135 shear in reloading.
Fig. 13 a shows the full constraint simulated stress–strain behavior in 45 and 90
forward and reverse shear for a reversal strain of 0.21. The flow stress before and
immediately after reversal is exactly the same for the 45 test, as expected. However,
the reverse loading results in a lower hardening rate compared to monotonic hard-
ening rate, which corresponds to different texture evolution. For forward and
reverse 90 simple shear, the flow stress before and after reloading is also the same,

Table 2
Boundary conditions imposed on the polycrystal for the simulation of TD uniaxial tension (r90 ¼ 0:87)
and simple shear in different direction; note that 90 shear means 90 from the TD tensile pre-strain
direction

Test TD tension 45 shear 90 shear 45 shear (135 )

2 r90 3
 0 0 2 3 2 3 2 3
6 ð1 þ r90 Þ 7 1 1 0 0 2 0 1 1 0
6 7 4
Velocity gradient 6 0 1 0 7 1 1 05 40 0 0 5 4 1 1 05
4 1 5
tensor (proportional) 0 0  0 0 0 0 0 0 0 0 0
ð1 þ r90 Þ
F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244 1235

Fig. 11. Predicted stress vs. plastic work curves obtained after TD uniaxial tension at a strain of 0.14
followed by 45 and 135 simple shear up to a strain of 0.3: (a) Taylor model predictions; (b) visco-plastic
self-consistent model predictions.
1236 F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244

Fig. 12. Predicted (111) pole figures for specimens deformed in: TD uniaxial tension up to a strain of 0.14,
and followed by 45 and 135 simple shear tests up of a strain of 0.3. Full constraint predictions (a–c).
VPSC predictions (d–f).
F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244 1237

Fig. 13. (a) Predicted and (b) experimental stress vs. plastic work curves obtained in 45 and TD simple
shear for forward and reverse (0.21 reversal strain) loading conditions.
1238 F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244

although the reverse flow curve tends to be higher than the forward flow curve after
reversal.

5. Discussion

Since the main factors affecting the behavior of this AA1050-O sheet sample are
the dislocation structure and the crystallographic texture, the discussion will focus
on dislocation storage, recovery (static and dynamic) and texture evolution. All the
stress–strain curves presented in this work were duplicated and results were very
reproducible.
Fig. 1 shows that reloading a tensile specimen under the same deformation mode
two days after unloading leads to a certain amount of recovery. The material has a
much lower flow stress than it had just before unloading, hardens at a faster rate for
a small amount of plastic work (few percent strain), and then exhibits the same
hardening rate as that for the non-interrupted test without reaching the same flow
stress. Therefore, the interruption of a test for 2 days, the amount of time between
tensile pre-deformation and subsequent simple shear tests, leads to some amount of
static recovery, part of which becomes permanent. When comparing flow stresses
between monotonic and non-monotonic loading, this difference needs to be taken
into account. Note that, in Fig. 1, the monotonic curve and the reloading segments
translated by a proper amount would superimpose well.
For the pseudo-monotonic case (TD tension followed by 45 shear, Fig. 3), the
reloading flow stress reaches the monotonic simple shear flow stress for the small
TD tensile pre-strain (0.07). However, for the larger pre-strain (0.14), the reloading
flow stress remains well below the monotonic curve. It was expected from the inter-
rupted tensile test (Fig. 1) that the same amount of static recovery would also occur
for the pseudo-monotonic curve at the two pre-strain levels. Therefore, this result
indicates that plastic work might not be as good as an equivalent measure of defor-
mation as expected, or that systematic errors from the experimental procedure were
not eliminated. Nevertheless, the results collected in this work are still very valuable.
For instance, the comparison of the pseudo-monotonic and pseudo-reverse curves
for a given pre-strain (Fig. 6) is not affected by the amount of static recovery
because the initial starting condition (pre-straining in tension) is the same. More-
over, clear trends on the reloading flow stress levels seen in Figs. 3–5 can be
observed. Finally, the shapes of the reloading curves, with or without stagnation, are
not affected by the choice of the equivalent measure of plastic deformation.
The pseudo-reverse hardening curves in Fig. 5, which show stagnation after
reloading, are consistent with the curves of Fig. 7 for which forward and reverse
simple shear tests were carried out, confirming the validity of the pseudo-reverse
test. This stagnation of work hardening can be explained by the rapid dissolution of
the dislocation structure after reloading (Fig. 9c). Fig. 6 compares the hardening
curves for pseudo-monotonic and pseudo reverse tests. Based on crystal plasticity
results, these curves should be identical. This is a clear indication that the disloca-
tion structure plays an important role in the hardening behavior of materials sub-
F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244 1239

jected to non-linear loading. Just after strain hardening stagnation for the reloading
curve, the flow stress for the pseudo-monotonic loading is larger than that for
pseudo-reverse loading at the same amount of work. Because, for each pre-strain,
both curves tends towards the same flow stress again at larger deformation, the
accumulation of dislocations is higher in pseudo-reverse than in pseudo-monotonic
deformation after the stagnation stage. This indicates that, as suggested by Hase-
gawa and Yakou (1980), reverse loading and recovery have the same effect on the
hardening behavior of aluminum, i.e., the flow stress decreases, but the hardening
rate increases. In fact, the symbols in Fig. 5 show that, if the reloading curves were
translated by the appropriate amount, all the curves would superimpose very well.
Note that the transient stagnation of strain hardening on reverse loading that is
usually reported in the literature is a rather inaccurate description of the material
behavior. As illustrated by the curves shown in this work (Figs. 5 and 7), after
reverse loading, the material exhibits a low yield stress followed by a very high strain
hardening within a strain range of approximately 1–2% before strain hardening
stagnation. The phase of high strain hardening, or ‘‘true’’ Bauschinger effect, is
more likely due to the reorientation of the internal stresses caused by grain incom-
patibilities and dislocation pile-ups. The hardening stagnation is more associated
with the polarization aspect of dislocation assemblies, i.e, the excess of dislocations
of one sign at either side of a dislocation wall (Kocks et al., 1980). The strain hard-
ening stagnation corresponds to the dissolution of the polarized part of dislocation
assemblies.
Fig. 4 shows that 90 simple shear reloading exhibits a lower yield stress compared
to the monotonic curve. However, due to the higher hardening rate, the flow stresses
of the reloading curves reach and become even higher than the monotonic curve, in
contrast to 45 and 135 subsequent shear loadings. In the absence of static recov-
ery, whose effect is shown in Fig. 1 for uniaxial tension, the yield stress of the
reloading curve would be approximately equal to the flow stress of the monotonic
curve at an equivalent amount of pre-strain. After a small amount of plastic defor-
mation, the reloading flow stress would even be higher than the flow stress of the
monotonic curve. This analysis is in agreement with and validates the polycrystal
simulations of forward and reverse simple shear (Fig. 13), which do not account for
static recovery. This behavior is consistent with the TEM observations (Fig. 9b),
which show that cross loading leads to a transition to the new dislocation structure,
typical of the second path (Fig. 8b), without dissolution of the initial structure
(Fig. 8a), making the material harder to deform.
Experimental (Fig. 10b and c) and predicted (Fig. 12b and c or e and f) (111) pole
figures for pseudo-monotonic and pseudo-reverse loading are in good agreement
although, as noted before, the experimental pole figures appear to be more diffuse
compared to the predicted pole figures. Although final textures for simple shear at
45 and its reverse direction 45 (135 ) are generally different, they are very similar
for AA1050-O because cube texture (100) < 001> , the main component in this
material, exhibits symmetry with respect to axes oriented at 45 from the rolling
direction. This is the reason why the stress–strain behavior for these two cases is
very similar (Fig. 11).
1240 F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244

Fig. 13a shows the predicted stress–strain behavior in 45 and TD for monotonic
simple shear tests and forward–reverse tests at a reversal strain of 0.21. In the first
case (45 ), the monotonic curve stays above the reloading curve, whereas, in the
second case (TD), the opposite is observed. It is interesting to note that these trends
are consistent with those calculated for pseudo-reverse loading (Fig. 11). Since the
polycrystal model does not predict the stagnation due to cell dissolution, both pre-
dicted reverse loading curves (Fig. 13a) are, relative to their respective monotonic
curves, higher than the experimental curves (Fig. 13b). The predicted 45 reverse
loading curve, contrary to the 90 curve, exhibits some stagnation due to texture
evolution. Because, experimentally, this effect is combined with the influence of dis-
location dissolution, the experimental 45 simple shear reverse flow curve exhibits
material softening (Fig. 13b), while the 90 curve exhibits only stress stagnation.
Note that the monotonic TD shear curve and the TD forward shear curve do not
match perfectly, the only instance where reproducibility was not satisfactory in this
entire work. However, because the TD forward curves at two other pres-train levels
(0.07 and 0.14) are in perfect agreement (Fig. 7), it is assumed that the TD mono-
tonic shear curve is slightly overestimated. Keeping in mind that neither static nor
dynamic recovery effects are captured by the polycrystal model, the predicted
(Fig. 13a) and experimental (Fig. 13b) curves seem very consistent. All these obser-
vations suggest that beyond a relatively small strain (plastic work) range following a
stress reversal, the stress–strain behavior becomes mostly influenced by crystal-
lographic texture evolution.

6. Conclusions

This paper presents a combined analysis involving dislocation microstructure


observations and polycrystal modeling to interpret the mechanical behavior of alu-
minum alloy 1050-O deformed in sequences of two linear strain paths, simulating
pseudo-monotonic, cross and pseudo-reverse loading conditions. The hardening
behavior was described by curves representing the stress as a function of plastic
work in order to eliminate the influence of plastic anisotropy. The effect of recovery
was assessed using interrupted uniaxial tension test results. Forward and reverse
simple shear tests were conducted to confirm the results of the pseudo-reverse tests
and validate the experiments consisting of tension followed by simple shear.
In general, the anisotropic hardening behavior of the material after a strain path
change is characterized by three stages. The first stage, up to 1 or 2% strains, cor-
responds to the ‘‘true’’ Bauschinger effect, i.e. grain incompatibilities and reorienta-
tion of the back stress. The second stage, for a few percent strains (corresponding,
for instance, to the work-hardening stagnation in reverse loading), is connected to
the reorganization of dislocation structures. Finally, for larger strains, the third
stage is controlled by texture evolution.
The mechanical tests showed that the hardening behavior after re-loading is
affected by the recovery taking place between loading and re-loading and that tran-
sient effects in the stress–work curves are observed for conditions of cross and
F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244 1241

reverse loadings. This behavior was correlated to the dislocation microstructure


evolution after these abrupt changes in strain paths were performed. For pseudo-
reverse or cross loading deformation, the structure typical of the second loading
condition appears in the grains. However, in the first case (pseudo-reverse loading),
a dissolution of the previous structure is the first step of the structure evolution
while in cross loading, it seems that closed cells evolve towards dislocation walls by
superimposition of the two types of structures, without noticeable evidence of cell
dissolution.
An important conclusion of the present work combined with the previous paper of
Lopes et al. (in press) is that strain hardening is not isotropic for linear or non-linear
loading. For monotonic loading, or at strains far from a strain path change, the
differences observed in the hardening rate in different testing orientations appear to
be controlled mainly by crystallographic texture evolution. However, for sequences
of two linear strain paths, the anisotropic hardening behavior just after reloading
appears to be strongly dependent on the dislocation microstructure evolution.

Acknowledgements

The authors would like to thank the contribution of the anonymous reviewers for
their very constructive comments, and of Mr. D.J. Lege (Alcoa Technical Cen-
ter) for his critical review of the manuscript and his excellent suggestions. The
authors are also grateful to Mr. E. Llewellyn (Alcoa Technical Center) for OIM
texture measurements and analysis. Finally, the authors express their sincere grati-
tude to Dr. C. Tomé (LANL) for providing the visco-plastic self-consistent poly-
crystal software (Lebensohn and Tomé, 1993) with many valuable advices about its
use.

Appendix. Velocity field gradient, stress and rate of deformation

@v
Given a velocity field v and its gradient @x in a local reference frame x, its expres-
sion in a global reference frame X is obtained by tensor transformation

@V @v
¼ P PT
@X @x

In the particular case of simple shear in a direction of a sheet orientated at an angle 


from the rolling direction (RD), the transformation matrix P is given by
2 3
cos sin 0
P ¼4 sin cos 05
0 0 1
1242 F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244

The stress deviator S can be obtained using polycrystal calculations carried out in
the global frame X. Transformed in the local frame x, S becomes

s ¼ PT SP

For simple shear the velocity gradient tensor expressed in the two frames is
2 3 2 3
@v 4
0 k 0
@V k sin  cos  k cos2  0
¼ 0 0 0 5 and ¼ 4 k sin2  k sin  cos  0 5
@x @X
0 0 0 0 0 0

The deviatoric shear stress tensor component expressed in the local frame is
 
s12 ¼ Pi1 Pj2 Sij ¼ ðS22  S11 Þ sin  cos  þ S12 cos2   sin2 

Assuming the plastic work and its rate as ‘‘equivalent’’ measures of plastic defor-
mation and rate of deformation, respectively, the plastic work rate should be equal
in tension and shear and should produce the same viscous effects
 : :
w ¼ 22 "22 ¼ 22 v2;2
: :
w ¼ 212 "12 ¼ s12 v1;2

This implies that, if the monotonic shear test is conducted with a velocity gradient
component v1,2, the tension pre-strain should be conducted with v2;2 ¼ v1;2 s12 =22 ,
with both stresses measured at the same amount of plastic work.

References

Bate, P.S., 1993. The effect of combined strain path and strain rate change in aluminum. Metall. Trans.
24A, 2679–2689.
Baudelet, B., Deguen, M., Felgères, L., Parnière, P., Rondé-Oustau, F., Sanz, G., 1978. Analyze micro-
structurelle de l’influence des trajectoires de déformation. Mém. Scient. Revue Métall 75, 409–422.
Bay, B., Hansen, N., Hugues, D.A., Kuhlman-Wilsdorf, D., 1992. Evolution of fcc deformation structures
in polyslip. Acta Metall. Mater. 40, 205–219.
El-Danaf, E., Kalidindi, S.R., Doherty, R.D., 2001. Influence of deformation path on the strain hard-
ening behavior and microstructure evolution of low SFE FCC metals. Int. J. Plasticity 17, 1245–1265.
Fernandes, J.V., Gracio, J.J., Schmitt, J.H., Grain, 1993. Size effect on the microstructural evolution of
copper deformed in rolling-tension. In: Teodosiu, C., Raphanel, J.L., Sidoroff, F. (Eds.), Proceedings of
the MECAMAT’91 Symposium on Large Plastic Deformations: Fundamental Aspects and Applica-
tions to Metal Forming. Balkema, Rotterdam, p. 219.
Fernandes, J.V., Schmitt, J.H., 1983. Dislocation microstructures in steel during deep drawing. Philos.
Mag. 48, 841–870.
Gambin, W., 1991. Plasticity of crystals with interacting slip systems. Engineering Transactions 39, 303.
Gambin, W., 1992. Refined analysis of elastic-plastic crystals. Int. J. Solids Structures 29, 2013–2027.
Gambin, W., Barlat, F., 1997. Modeling of deformation texture development based on rate independent
crystal plasticity. Int. J. Plasticity 13, 75–85.
F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244 1243

Geng, L., Wagoner, R.H., 2002. Role of plastic anisotropy and its evolution on springback. Int. J. Mech.
Sci. 44, 123–148.
Gracio, J.J., Barlat, F., Ferreira Duarte, J., 1999. Plasticity of aluminum alloy 1050 during non-propor-
tional loading. In: Advanced Technology Of Plasticity, Vol. II, Proc. 6th ICTP, pp. 1365–1370.
Gracio, J.J., Barlat, F., Rauch, E.F., 2000a. Strain hardening anisotropy of aluminum: A self-consistent
approach. In: Martin, P., MacEwen, S., Verreman, Y., Liu, W.J., Goldak, J. (Eds.), CD ROM Math.
Modeling In Metal Processing And Manufacturing. Nucleus Design, St-Hubert, Québec, Canada.
Gracio, J.J., Fernandes, J.V., Schmitt, J.-H., 1989. Effect of grain size on substructural evolution and
plastic behavior of copper. Mater. Sci. Eng. A118, 97–105.
Gracio, J.J., Lopes, A.B., Rauch, E.F., 2000a. Analysis of plastic instability in commercially pure alumi-
nium. J. Mater. Proc. Technol. 103, 160–164.
Hansen, N., 1992. Microstructure and flow stress of cell forming metals. Scripta Metall. Mater. 27, 947–
950.
Hasegawa, T., Yakou, T., 1980. The effect of strain reversal and thermal recovery on stress vs strain
behavior in aluminum. Scripta Metall. 14, 1083–1087.
Hasegawa, T., Yakou, T., Karashima, S., 1975. Deformation behavior and dislocation structures upon
stress reversal in polycrystalline aluminum. Mater. Sci. Eng. 20, 267–276.
Hatherly, M., 1983. Deformation at high strains. In: Gifkin, R.C. (Ed.), Strength of Metals and Alloys.
Pergamon Press, Oxford, pp. 1181–1195.
Hiwatashi, S., Van Bael, A., Van Houtte, P., Teodosiu, C., 1997. Modeling of plastic anisotropy based on
texture and dislocation structure. Computational Materials Science 9, 274–284.
Hoc, T., Forest, F., 2001. Polycrystal modeling of IF-Ti steel under complex loading path. Int. J. Plasti-
city 17, 65–85.
Hosford, W.F., 1993. Continuum analyses of anisotropic plasticity. In: The Mechanics of Crystals and
Textured Polycrystals. Oxford University Press, New York, p. 139.
Hughes, D.A., 1992. Microstructure and flow stress of deformed polycrystalline metals. Scripta Metall.
Mater. 27, 969–974.
Hughes, D.A., 1993. Microstructural evolution in a non-cell forming metals: Al-Mg. Acta Metall. Mater.
41, 1421–1430.
Juul Jensen, D., Hansen, N., 1990. Flow stress anisotropy in aluminum. Acta Metall. Mater. 38, 1369–
1380.
Juul Jensen, D., Hansen, N., 1991. Flow stress anisotropy in cross-rolled aluminum. In: Proc. 9th Intern.
Conf. Strength of Materials (ICSMA 9), Israel. Freund, London, pp. 179–186.
Kalidindi, S.R., 2001. Modeling anisotropic strain hardening and deformation textures in low stacking
fault energy fcc metals. Int. J. Plasticity 17, 837–860.
Khan, A.S., Huang, S., 1995. Continuum Theory of Plasticity. John Wiley & Sons, New York.
Kocks, U.F., Hasegawa, T., Scattergood, R.O., 1980. On the origin of cell walls and of lattice mis-
orientation during deformation. Scripta Metall. 14, 449–454.
Kuhlmann-Wildsorf, D., 1992. Fundamentals of cells and subgrains structures in historical perspective.
Scripta Metall. Mater. 27, 951–956.
Lebensohn, R.A., Tomé, C.N., 1993. A self-consistent anisotropic approach for the simulation of plastic
deformation and texture development of polycrystals: application to zirconium alloys. Acta Metall.
Mater. 41, 2611–2624.
Li, F., Bate, P.S., 1991. Strain path changes effects in cube textured aluminum. Acta Metall. Mater. 39,
2639–2650.
Li, S., Hoferlin, E., Van Bael, A., Van Houtte, P., Teodosiu, C., in press. Finite element modeling of
plastic anisotropy induced by texture and strain-path change. Int. J. Plasticity.
Lloyd, D.J., Sang, H., 1979. The influence of strain path on subsequent mechanical properties—ortho-
gonal tensile paths. Metall. Trans. 10A, 1767–1772.
Lopes, A.B., Barlat, F., Gracio, J.J., Ferreira Duarte, J., Rauch, E.F., in press. Effect of texture and
microstructure on strain hardening anisotropy for aluminum deformed in uniaxial tension and simple
shear. Int. J. Plasticity.
1244 F. Barlat et al. / International Journal of Plasticity 19 (2003) 1215–1244

Lopes, A.B., Rauch, E.F., Gracio, J.J., 1999. Textural vs structural plastic instabilities in sheet metal
forming. Acta Mater. 47, 859–866.
Mollica, F., Rajagopal, K.R., Srinivasa, A.R., 2001. The inelastic behavior of metals subjected to loading
reversal. Int. J. Plasticity 17, 1119–1146.
Nesterova, E.V., Bacroix, B., Teodosiu, C., 2001. Microstructure and texture evolution under strain path
changes in low-carbon interstitial-free steel. Metall. Mater. Trans. A 32A, 2527–2538.
Peeters, B., Bacroix, B., Teodosiu, C., Van Houtte, P., Aernould, E., 2001a. Work hardening/softening
behaviour of b.c.c. polycrystals during changing strain path: II. TEM observations of dislocations
sheets in an IF steel during two-stage strain paths and their representations in terms of dislocation
densities. Acta Mater. 49, 1621–1632.
Peeters, B., Seefeldt, M., Teodosiu, C., Kalindindi, S.R., Van Houtte, P., Aernould, E., 2001b. Work
hardening/softening behaviour of b.c.c. polycrystals during changing strain path: II. An integrated
model based on substructure and texture evolution, and its predictions of the stress–strain behaviour of
an IF steel during two-stage strain paths. Acta Mater. 49, 1607–1619.
Rao, V.N., Laukonis, J.V., 1983. Microstructural mechanisms for the anomalous behavior of aluminum-
killed steel prestrain in plane strain tension. Mater. Sci. Eng. 60, 125–135.
Rauch, E.F., 1998. Plastic anisotropy of sheet metals determined by simple shear tests. Mater. Sci. Eng.
A241, 179–183.
Rauch, E.F., Gracio, J.J., Barlat, F., Lopes, A.B., Ferreira Duarte, J., 2002. Hardening behaviour and
structural evolutions upon strain reversal of aluminum alloys. Scripta Materialia, 46, 881–886.
Rauch, E.F., Schmitt, J.-H., 1989. Dislocation substructures in mild steel deformed in simple shear.
Mater. Sci. Eng. A113, 441–448.
Ronde-Oustau, F., Baudelet, B., 1977. Microstructure and strain path in deep-drawing. Acta Metall. 25,
1523–1529.
Sang, H., Lloyd, D.J., 1979. The influence of biaxial prestrain on the tensile properties of three aluminum
alloys. Metal. Trans. 10A, 1773–1776.
Schmitt, J.H., Aernould, E., Baudelet, B., 1985. Yield loci for polycrystalline metals without texture.
Mater. Sci. Eng. 75, 13–20.
Schmitt, J.-H., Fernandes, J.V., Gracio, J.J., Vieira, M.F., 1991. Plastic behavior of copper sheets during
sequential tension tests. Mater. Sci. Eng. A147, 143–157.
Stoltz, R.E., Pelloux, R.M., 1976. The Bauschinger effect in precipitation strengthened aluminum alloys.
Metall. Trans. 7A, 1295. 1306.
Teodosiu, C., Hu, Z., 1998. Microstructure in the continuum modeling of plastic anisotropy. In: Carten-
sen, J.V., Leffers, T., Lorentzen, T., Pedersen, O.B., Sørensen, B.F., Winther, G. (Eds.), Proceedings of
the Risø International Symposium on Material Science: Modelling of Structure and Mechanics of
Materials from Microscale to products. Risø National Laboratory, Roskilde, Denmark, p. 149.
Tomé, C.N., Canova, G.R., Kocks, U.F., Christodoulou, N., Jonas, J.J., 1984. The relation between
macroscopic and microscopic strain hardening in FCC polycrystal. Acta Metall. 32, 1637–1653.
Vieira, M.F., Schmitt, J.-H., Gracio, J.J., Fernandes, J.V., 1990. The effect of strain path change on the
mechanical behavior of copper sheets. J. Mater. Process. Technol. 24, 313–322.
Wagoner, R.H., Laukonis, J.V., 1983. Plastic behavior of aluminum-killed steel following plane-strain
deformation. Metall. Trans. 14A, 1487–1495.
Wilson, D.V., 1989. Effects of changes in strain path on work-hardening in cubic metals. Metal. Trans.
20A, 2471–2482.
Wilson, D.V., Bate, P.S., 1994. Influence of cell walls and grain boundaries on transient responses of an
IF steel to changes in strain path. Acta Metall. Mater. 42, 1099–1111.
Wilson, D.V., Zandrahimi, M., Roberts, W.T., 1990. Effects of changes in strain path on work-hardening
in CP aluminum and an Al-Cu-Mg-Si alloy. Acta Metall. Mater. 38, 215–226.
Zandrahimi, M., Platias, S., Price, D., Barrett, D., Bate, P.S., Roberts, W.T., Wilson, D.V., 1989. Effects
of changes in strain path on work-hardening in cubic metals. Metall. Trans. 20A, 2471–2482.

You might also like