You are on page 1of 22

International Journal of Plasticity 18 (2002) 1661–1682

www.elsevier.com/locate/ijplas

Anisotropic plasticity for sheet metals using


the concept of combined isotropic-
kinematic hardening
Han-Chin Wu*
Department of Civil and Environmental Engineering, The University of Iowa, Iowa City, IA 52242, USA

Received in final revised form 18 August 2001

Abstract
Hill’s 1948 anisotropic theory of plasticity (Hill, R., 1948. A theory of yielding and plastic
flow of anisotropic metals. Proc. Roy. Soc. London A193, 281–297) is extended to include the
concept of combined isotropic-kinematic hardening, and the objective of this paper is to
validate the model so that it may be useful for analyses of sheet metal forming. Isotropic
hardening and kinematic hardening may be experimentally observed in sheet metals, if yield-
ing is defined by the proportional limit or by a small proof strain. In this paper, a single
exponential term is used to describe isotropic hardening and Prager’s linear kinematic hard-
ening rule is applied for simplicity. It is shown that this model can satisfactorily describe both
the yield stress and the plastic strain ratio, the R-ratio, observed in tension test of specimens
cut at various angles measured from the rolling direction of the sheet. Kinematic hardening
leads to a gradual change in the direction of the plastic strain increment, as the axial strain
increases in the tension test; while in the traditional approach for sheet metal, this direction
does not change due to the use of isotropic hardening. # 2002 Elsevier Science Ltd. All rights
reserved.
Keywords: Anisotropy; Sheet metal; Yield stress; R-ratio; Plastic deformation

1. Introduction

Using the phenomenological approach, Hill (1948) developed an anisotropic theory


of plasticity, which uses a quadratic yield function. The yield function was originally

* Fax: +1-319-335-5660.
E-mail address: hcwu@uiowa.edu

0749-6419/01/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S0749-6419(01)00060-2
1662 H.-C. Wu / International Journal of Plasticity 18 (2002) 1661–1682

Nomenclature

x, y, z rolling, transverse and normal directions of the metal sheet


x0 ; y0 axial and width directions of a flat tension specimen
x ; y ; xy stress components
1 ; 2 principal stresses
"pw ; "pt plastic strains in width and thickness directions in tension test
rx ; ry ; rxy back stress components representing kinematic hardening
rD D D
x ; ry ; rxy deviatoric part of back stress components
f isotropic hardening function
F, G, H and N coefficients of Hill’s 1948 anisotropic yield function
2 yield function
Dx ; Dy ; Dxy components of rate of deformation
Dx0 ; Dy0 ; Dx0 y0 rate of deformation along the axial, width, and thickness
directions
Dpx ; Dpy ; Dpxy plastic part of rate of deformation
 angle of tension specimen measured counter-clockwise from
the rolling direction
d increment of the accumulated plastic strain
0 ; 45 ; 90 yield stress along directions of 0, 45, and 90 degrees
R0 ; R45 ; R90 plastic strain ratio along directions of 0, 45, and 90 degrees
s-direction a direction bisecting the right angle formed by the x and  y
axes

derived for a three-dimensional anisotropic solid, but it takes the following form if
expressed in the two-dimensional stress space for a rolled sheet metal:

ðG þ HÞx2  2Hx y þ ðF þ HÞy2 þ 2Nxy


2
¼ 1: ð1Þ

The material parameters F, G, H and N represent the state of anisotropy of the


metal sheet, while x , y and xy are the in-plane components of the Cauchy stress.
Hill’s criterion considers the axes of anisotropy as the axes of reference, thus making
x the rolling direction and y the transverse direction of the metal sheet. Eq. (1) can
also be written as:
 
2 2 2R
1 þ 2  1 2 ¼ u2 ; ð2Þ
Rþ1
when considering in-plane isotropy. In Eq. (2), u is the yield stress in uniaxial ten-
sion. The parameter R is known as the plastic strain ratio, and defined by:

d"pw
R¼ ; ð3Þ
d"pt
H.-C. Wu / International Journal of Plasticity 18 (2002) 1661–1682 1663

where d"pw is the plastic strain increment in the width direction, and d"pt that in the
thickness direction. The R-value indicates anisotropy of plastic deformation. When
R ¼ 1 the material is isotropic, and Eq. (2) reduces to the von Mises yield criterion for
isotropic metals. Eq. (2) is a case of expressing the coefficients of the yield function in
terms of R. There are two aspects of anisotropy in a metal sheet, i.e. anisotropy of
the yield stress and anisotropy of the plastic flow. By expressing the coefficients of
the yield function in terms of R, the two kinds of anisotropy have been combined.
If Eq. (2) is applied to the case of equibiaxial tension, with 1 ¼ 2 ¼ B , it reduces
to:
 1
1þR 2
B ¼ u ; ð4Þ
2

where B is the equibiaxial yield stress. Thus, Eq. (4) results in B > u when R > 1
and B < u when R < 1. This relationship is further illustrated by plotting Eq. (2) in
Fig. 1. Most sheet metals show a yield locus with B > u in the tension quadrant of
the two-dimensional principal stress space. It is seen from the figure that Eq. (2)
does not agree with observed results in the case of R < 1.

Fig. 1. Yield surfaces with coefficients expressed in terms of R.


1664 H.-C. Wu / International Journal of Plasticity 18 (2002) 1661–1682

To remedy the discrepancy, Hill (1979, 1990, 1993) and other investigators, such
as Barlat and Lian (1989), Gotoh (1977a,b), Hosford (1972), Kreissig and Schindler
(1986) proposed more complex yield functions. These are generally non-quadratic in
form and have additional anisotropic parameters. A brief summary of the proposed
yield functions is given in Wu et al. (1999) and in Lademo et al. (1999). In a recent
work, Lademo et al. (1999) compared the models of Hill (1948), Barlat and Lian
(1989), and Karafillis and Boyce (1993) with their own experimental data of two
aluminum materials and found that, for the former two models, calibrations based
on yield stresses gave a good description of the yield stress anisotropy, but the same
calibrations gave poor descriptions of the anisotropy in the R-ratio. Similarly, a
calibration based on R-ratios gave a good description of the anisotropy in the R-
ratio but a poor description of the yield stress anisotropy. The model of Karafillis
and Boyce (1993) with zero back stress was unable to describe both kinds of aniso-
tropy. It was then concluded that none of the models considered were capable of
simultaneously describing both the anisotropy in the yield stress and the R-ratio for
the alloys considered. The model of Karafillis and Boyce with nonzero back stress
was not considered by Lademo et al. (1999).
The existing anisotropic theory of plasticity for sheet metals proposes a yield func-
tion (quadratic or non-quadratic), assumes the normality condition for the plastic
strain increments, and considers isotropic hardening. This theory is exceedingly dif-
ferent from the classical flow theory of plasticity. It is different in many aspects as
follows: First, instead of a quadratic (initial) yield function for all metallic materials,
this anisotropic theory for sheet metals uses a non-quadratic yield function with the
exponent of the function depending on the material being considered. Furthermore,
instead of the usual isotropic-kinematic hardening concept, this theory is entirely
based on the concept of isotropic hardening. Another discrepancy is related to the
experimental determination of the coefficients of the yield function. Traditionally, in
the flow theory of plasticity, the yield function has been determined independent of
the flow rule. In this theory, however, the flow rule has been extensively used in the
determination of the coefficients of the yield function through the use of R. It should
be pointed out that these different assumptions of the theory are not fully compa-
tible. They have together created inconsistencies which have been demonstrated by
the findings of Lademo et al. (1999) that none of the models considered were capable
of describing both aspects of anisotropy in the yield stress and the R-ratio for the
alloys considered. The existing theory also leads to an undesirable situation that the
sheet metal would require a different theory of plasticity from that used for the same
material in other product forms.
The objective of the present paper is to demonstrate that the traditional flow theory
of plasticity is capable of describing both aspects of anisotropy (yield stress and R-ratio)
for sheet metals and it may be useful for analyses of sheet metal forming. The proposed
initial yield surface is Hill’s quadratic criterion, but it undergoes isotropic (expansion of
the yield surface) as well as kinematic (motion of the yield surface) hardening. Some
previous works using the same idea were reported in Wu et al. (1999), Wu and
Loureiro (1999) and Wu and Hong (1999). Although this model has been previously
used for computation, validation of the model is rare in the literature. In this paper,
H.-C. Wu / International Journal of Plasticity 18 (2002) 1661–1682 1665

the hardening rules were specified by an isotropic hardening function previously pro-
posed by Wu and Yip (1981) and Prager’s kinematic hardening rule, Prager (1955).
In the case of the existing theory, the isotropic-hardening fixes the shape of the
yield surface and also the direction of the plastic strain increment (associated flow rule)
as the load increases, leading to a poor description of the anisotropic flow properties.
Lademo et al. (1999) pointed out that the anisotropic flow properties may be better
described by use of a non-associated flow rule. In the proposed theory, reasonable
results are obtained by use of associated flow rule together with the concept of kine-
matic-hardening. In the case of kinematic-hardening, the direction of the plastic strain
increment changes as the loading increases, and it is not necessary to use a non-
associated flow rule as suggested by Lademo et al. (1999). This effect will be further
explained later in the text. The definition of yield is significantly affecting the
amount of kinematic hardening and, thus, is the topic of discussion in Section 2.
All aforementioned papers, including the present paper, address anisotropic metal
sheets subjected to biaxial tension with proportional stress- (strain-) paths. Non-pro-
portional stress- (strain-) paths and shear deformation are frequently addressed topics
in recent plasticity literatures. They are associated with the re-orientation of material
textures, and detailed discussions may be found in Kim (1992), Kim and Yin (1997),
Tugcu et al. (1999), Tugcu and Neale (1999), Kuroda (1997), Huang et al. (2000), Cao
et al. (2000), Yoon et al. (2000), Dafalias (2000), and Wu (2002). These problems are
important in the forming of sheet metals.
The proportional stress- (strain-) paths are still important in that they are experi-
mentally used to characterize the anisotropy of the yield criterion and the flow rule of
sheet metals. However, the characterization of the material is not complete without the
consideration of the non-principal state of stress. The non-principal state of stress
has been addressed by the author in Wu (2002) and further efforts will be devoted to
addressing this problem in the near future.

2. The definition of yield and kinematic hardening in sheet metals

There is an extensive literature addressing the issue of experimental determination


of yield surfaces in plasticity, e.g. Naghdi et al. (1958), Ivey (1961), Bertsch and Findley
(1962), Mair and Pugh (1964), Phillips and Tang (1972) and Wu and Yeh (1991). Most
of these works used tubular specimens and determined the yield surfaces in the xy
versus x stress subspace. From the experimental results of these investigations, it was
clear that the definition of yield significantly influenced the characteristics of strain
hardening. Yield defined by proportional limit leads to kinematic hardening and dis-
tortion of the yield surface; yield defined by a 0.2% proof strain leads to a combined
isotropic-kinematic hardening; and yield defined by backward linear extrapolation or
by a large proof strain leads to isotropic hardening.
The picture is not as clear in the case of sheet metals or metal plates. In this case, the
experimental points on the yield surface are usually determined from the uniaxial ten-
sion test, the bulge test (balanced biaxial tension), the biaxial test using cruciform spe-
cimens, the plane-strain compression test, and the through-thickness compression
1666 H.-C. Wu / International Journal of Plasticity 18 (2002) 1661–1682

test. Test data using the uniaxial tension test, the bulge test, and the cruciform speci-
men test are obtained only in the tensile stress region of the stress space. Test data of
the plane-strain compression test and the through-thickness test can include infor-
mation for the compression region, but, these must be supplemented by the assump-
tion that a superimposed hydrostatic stress does not influence yielding. Therefore,
they are not direct experimental results.
Szczepinski (1963) investigated the effect of definition of yield by use of uniaxial
tension specimens cut out from metal sheets making various  angles with the rolling
direction. The yield stress
 determined
 from each specimen is transformed into a set
of stress components x ; y ; xy by the use of Mohr’s circle. These sets of stress
components can then be plotted in a three-dimensional stress space, with x and y
forming two perpendicular axes on the horizontal plane, and xy the vertical axis
normal to the horizontal plane. The initial yield surface for an isotropic material is
an ellipsoid in this space, if the Mises yield criterion is used. The subsequent yield
surfaces are represented by the motion, distortion, and/or expansion of the ellipsoid
and can be best viewed by two projections. The first projection is onto the (x ; y )
plane (the plane view) and the second projection plane (the elevation view) is onto a
plane that passes through the xy axis and the s-direction which is bisecting the right
angle formed by the x and  y axes and is perpendicular to the major axis of the
ellipsoid. The kinematic hardening, which is difficult to identify based on data in the
tension quadrant of the two-dimensional principal stress space, can be visualized in
the aforementioned projections.
Experimental results which show kinematic hardening behavior of sheet/plate
metals can be found in the literature. By testing specimens cut from 2008-T4 alumi-
num sheet, Lege et al. (1989) found a difference in tensile and compressive yield
stresses, hinting the existence of an initial back stress. Tests by use of cruciform
specimens were conducted by Kreissig and Schindler (1986). The experimental
results show a significant kinematic hardening effect on the yield surface of sheet
metal materials.

3. Development

In this development, the elastic deformation of the sheet material is assumed to be


isotropic. In the case of plane stress, the equations are
   
"ex ¼ E1 x 
y ; "ey ¼ E1 y 
x ; "exy ¼ 2 1
xy ð5Þ

where the elastic strain components are

"ex ¼ Dex dt; "ey ¼ Dey dt; "exy ¼ Dexy dt ð6Þ

and Dex ; Dey ; and Dexy are components of the elastic part of the rate of deformation
Dx ; Dy ; and Dxy , respectively. In (5), E is the elastic modulus, is the shear modulus
and
is Poisson’s ratio.
H.-C. Wu / International Journal of Plasticity 18 (2002) 1661–1682 1667

The new yield function, like Hill’s 1948 yield criterion, considers the axes of aniso-
tropy as the axes of reference and is represented by Eq. (7). Hence, x denotes the rolling
direction and y the transverse direction of the sheet metal. The yield function is:
   2
2 ¼ ðG þ HÞðx  rx Þ2 2Hðx  rx Þ y  ry þ ðH þ FÞ y  ry
 2 ð7Þ
þ 2N xy  rxy ¼ f 2 ;

where the material parameters F, G, H and N are determined by the initial state of
anisotropy of the metal sheet. The components of the back stress, rx, ry and rxy, specify
the center of the yield surface and are directly related to the kinematic hardening of the
material. The function f specifies the size of the yield surface, and it is an expression of
isotropic hardening. The function f increases with the accumulated plastic strain and
will be further discussed. In the case of a material without isotropic hardening, f=1;
and, in the case of an annealed or an as-received material where the initial state of
anisotropy is not apparent, rx=ry=rxy=0, and the initial value of f is f 0=1. When
there is a significant presence of anisotropy in the as-received material at initial
yielding, the initial values for rx, ry and rxy are not zero. In this situation, it is then
desirable to fit Eq. (7) to the experimental yield surface through an optimization
procedure. In this way, it is possible to determine the values of coefficients (F, G, H
and N), and the initial back stress components.
Using (7) and the normality rule, the flow rule is:
: 
Dpx ¼ l ðG þ HÞx  Hy
: 
Dpy ¼ l Hx þ ðH þ FÞy ð8Þ
:
Dpxy ¼ l Nxy

where the components of the ‘‘effective stress vector’’ are

x ¼ x  rx ; y ¼  y  r y ; xy ¼ xy  rxy ð9Þ

The ‘‘effective stress vector’’ is defined by Eq. (9) and this definition has been used
by other authors, such as Cho and Dafalias (1996). It should not be confused with
the ‘‘equivalent stress’’ to be defined later which is sometimes referred to as the
‘‘effective stress’’ in the plasticity literature.
: Eq. (8) may be applied to calculate the
rate of plastic deformation. Note that l is the plastic multiplier which is a positive
scalar that may vary during the straining process, and Dpx ; Dpy ; and Dpxy are the
plastic parts of the rate of deformation, such that

Dx ¼ Dex þ Dpx ; Dy ¼ Dey þ Dpy ; Dxy ¼ Dexy þ Dpxy ð10Þ

In the case of uniaxial tension test using a specimen cut out from a metal sheet at
an angle , measured counter-clockwise from the rolling direction, the stress com-
ponents are
1668 H.-C. Wu / International Journal of Plasticity 18 (2002) 1661–1682

x ¼  cos2 ; y ¼  sin2 ; xy ¼  sin cos ð11Þ

where  is the stress along the axial direction (x0 ) of the specimen. Let y0 be along the
width direction of the specimen, the x0 –y0 –z0 axes thus form a rectangular specimen
coordinate system with x0 and y0 lying in the x–y plane and z0 =z.
In this case, the rate of deformation in the specimen coordinate system are:

Dx0 ¼ Dx cos2  þ Dy sin2  þ 2Dxy sincos


Dy0 ¼ Dx sin2  þ Dy cos2   2Dxy sincos ð12Þ
1 
Dxy0 ¼  Dx  Dy sinð2Þ þ Dxy cosð2Þ;
2
and the corresponding plastic parts are related by

Dpx0 ¼ Dpx cos2  þ Dpy sin2  þ 2Dpxy sincos


Dpy0 ¼ Dpx sin2  þ Dpy cos2   2Dpxy sincos ð13Þ
1 
Dpxy0 ¼  Dpx  Dpy sinð2Þ þ Dpxy cosð2Þ;
2

The equivalent stress, , is similarly defined as in Hill’s (1950) formulation, but it


includes the effect of kinematic hardening:
" #1
ðG þ HÞ2x  2Hx y þ ðH þ FÞ2y þ 2N2xy 2
¼ : ð14Þ
K

where
2
K ¼ ðF þ G þ HÞ: ð15Þ
3
Combining Eqs. (7) and (14), the equivalent stress can also be expressed by:
f
 ¼ pffiffiffiffi : ð16Þ
K

The equivalent plastic strain-rate, D p , is defined so that the rate of plastic work
per unit volume is:
:
Wp ¼  D p ð17Þ

To obtain the expression for D p , the rate of plastic work is first written as
:     @ :
Wp ¼ ij  rij Dpij ¼ ij  rij l ð18Þ
@ij
H.-C. Wu / International Journal of Plasticity 18 (2002) 1661–1682 1669

Upon the differentiation of the yield function, 2, with respect to the stress com-
ponents and with the appropriate substitutions, Eq. (18) then reduces to:
: : :
Wp ¼ f 2 l ¼ 2 l ð19Þ

Using Eqs. (19) and (7), it can then be shown that Eq. (17) is satisfied if D p is given
by the following expression:
rffiffiffiffiffi  2 2M  2 12
K  p 2
D p ¼ p p p
ðF þ HÞ Dx þ2H Dx Dy þ ðG þ HÞ Dy þ p
Dxy ð20Þ
M N
where

M ¼ ðG þ HÞðH þ FÞ  H2 : ð21Þ
It can also be found that

: D p
l ¼ pffiffiffiffi ð22Þ
f K

In this model, a combined isotropic-kinematic hardening is considered. The evolu-


tion of isotropic hardening is accomplished by the evolution of the isotropic function
f; and the evolution of kinematic hardening is accomplished through the evolution of
the back stress (rx, ry, rxy). In the isotropic hardening, the effective plastic strain rate,
D p , is directly related to a monotonic increasing parameter such that:
:
¼ D p : ð23Þ

where represents the accumulated effective plastic strain. Following Hill (1950) the
isotropic hardening is:
ð 
f¼f  p
D dt ¼ f ð Þ; ð24Þ

which may be further expressed by a modified exponential function proposed by Wu


and Yip (1981) as:

f ð Þ ¼ D  ðD  1Þe : ð25Þ

Note that D and  are material constants with values depending on the angle ,
i.e. the isotropic hardening is direction dependent. The parameter D , where D 51,
represents the asymptotic amount of isotropic hardening as ! 1, while 
represents the rate of hardening. In addition, it is noted that fð Þ ¼ 1 for a material
without isotropic hardening. In the case of loading, (25) may be differentiated with
D and  remaining constant for each specified direction, and, by use of (22), the
following expression may be obtained:
1670 H.-C. Wu / International Journal of Plasticity 18 (2002) 1661–1682

: pffiffiffiffi :
f ¼ K  ðD  fÞf l ð26Þ

Prager’s (1955) kinematic hardening rule is used. According to which, the rate of
translation of the center of the yield surface is in the direction of the plastic strain
rate. Consequently, the deviatoric part of back stress rates are:
: : 
rD p
x ¼ C Dx ¼ Cl ðG þ HÞx  Hy
: : 
rD p
y ¼ C Dy ¼ Cl Hx þ ðH þ FÞy ð27Þ
: :
rD p
xy ¼ C Dxy ¼ Cl Nxy

and
: :D :D
rDx þ ry þ rz ¼ 0 ð28Þ

:
Knowing that rz ¼ 0 for plane stress, it may be shown that

: : :D : : :D
r x ¼ 2rDx þ ry and ry ¼ rDx þ 2r y ð29Þ

The scalar parameter C characterizes the material behavior. In this study, C is


considered as a constant, which leads to a linear-kinematic strain hardening.
The plastic strain ratio, R, for a tension specimen cut at an angle  from the
rolling direction, is defined by:

 p 
Dpy0 D y0
R ¼ ¼  Dp þD p ð30Þ
Dpz0 x0 y0

where the plastic incompressibility of the material is assumed. By the use of Eq. (13),
Eq. (30) is reduced to:
    
ðG þ HÞ  cos2   rx  H  sin2   ry sin2 
R ¼    
Gð cos2   rx Þ þ F  sin2   ry
    
ðH þ FÞ  sin2   ry  H  cos2   rx cos2 
þ    ð31Þ
Gð cos2   rx Þ þ F  sin2   ry
 
2N  sincos  rxy sincos
  
Gð cos2   rx Þ þ F  sin2   ry

Note that the isotropic hardening function, f, does not appear in Eq. (31), but the
back stress components do. Therefore, the R-ratio is significantly influenced by
kinematic hardening. From this equation, the following expressions can be found:
H.-C. Wu / International Journal of Plasticity 18 (2002) 1661–1682 1671

0 H  Hrx þ ðH þ FÞry
R0 ¼
0 G  Grx  Fry
   
1 2N 45  2rxy
R45 ¼ 1 ð32Þ
2 45 ðG þ FÞ  2Grx  2Fry
90 H þ ðG þ HÞrx  Hry
R90 ¼ ;
90 F  Grx  Fry

where 0 ; 45 ; 90 ; R0, R45, and R90 represent the yield stress and plastic strain
ratio at orientations 0, 45, and 90 degrees from the rolling direction, respectively. In
the case of rx=ry=rxy=0, Eq. (32) reduces to Hill’s (1948) definition of plastic
strain ratios given by:
 
H 1 2N H
R0 ¼ ; R45 ¼ 1 and R90 ¼ : ð33Þ
G 2 GþF F

For a given direction , the yield stress   may be found by the substitution of (11)
into (7). The resulting equation is

 
2 ðG þ HÞcos4   2H cos2  sin2  þ ðF þ HÞsin4  þ 2Nsin2  cos2 
"  #
2ðG þ HÞrx cos2  þ 2H rx sin2  þ ry cos2 
þ  ð34Þ
2ðH þ FÞry sin2   4N rxy sin cos
þ ðG þ HÞr2x  2H rx ry þ ðH þ FÞ r2y þ 2N r2xy  f2 ¼ 0

4. Results and discussion

It is shown in this section that the theory of the previous section is capable of
predicting experimentally observed data for sheet metals. In particular, tension test
of specimens cut at different  angles measured from the rolling direction is of
interest. The theoretical results are compared with two sets of experimental data for
different aluminum materials. The return-mapping algorithm, Simo and Hughes
(1998), is used in the computation to ensure that the updated stress point is on the
current yield surface. The following procedure is followed.
(1) Input time increment dt and angle  to determine the initial yield stress   and
the initial yield strain by use of (5), (11) and (34).
(2) Use the return-mapping algorithm to determine the increment dl, so that the
plastic part of rate of deformation and the plastic strain-ratio R may be found from
(8), (13) and (30).
(3) Update  ; f; rx ; ry and rxy by use of (26)–(29) and the following equations
of return-mapping algorithm:
1672 H.-C. Wu / International Journal of Plasticity 18 (2002) 1661–1682

Trial stress ¼ Tðnþ1Þ ¼ ðnÞ þ E Dx0 ðnÞ dt


ðnþ1Þ ¼ Tðnþ1Þ  E Dpx0 ðnÞ dt ð35Þ

The first example is related to the experimental data of Lademo et al. (1999). These
authors tested two aluminum materials in uniaxial tension. Only AA7108-T1 is being
considered here, and the other material can be similarly considered. The present writer
agrees with Lademo et al. (1999) that the uniaxial tension test is a desirable test to use
for the calibration of a material model. At the present stage of knowledge, the tension
test does not provide all needed information, however. It does not, for example, provide
information about the initial back stress which is needed in the present model. Due to
the thinness of the sheet material, a compression test is not practical. Therefore, more
research needs to be conducted to determine the initial back stress components (rx,
ry, rxy) which are present due to material processing.
The experimental data have been reproduced from Table 2 of Lademo et al. (1999).
Only specimen No. 1 for each  angle is considered. The equation to reproduce the
experimental stress–strain curve is
X
Y ¼ Y þ Qk ð1  expðCk " p ÞÞ ð36Þ
k

which was given in Lademo et al. (1999); Y is the flow stress; Y is the yield stress; " p
is the plastic strain; Qk and Ck are material parameters given in Table 2 of Lademo
et al. (1999). Note that, in most cases, three exponential terms of Eq. (36) were used,
and this equation reproduces the experimental data very well.
In the present computation by use of the proposed theory, only one exponential
term with two parameters D and  for each loading direction is used in the iso-
tropic hardening function, Eq. (25), and one material parameter C is used in the
kinematic hardening rule, Eq. (27). The idea is to keep the number of parameters at
its minimum so that the model can be used to predict material behavior under var-
ious loading conditions. Based on the experimental data, the variations of D and 
as a function of  have been determined. If more terms are used in Eqs. (25) and
(27), better agreement with the experimental results in discussion can be attained
but, at the same time, the parameters need to be adjusted case by case as in Lademo
et al. (1999) and the model will lose its power of prediction.
From Table 2 of Lademo et al. (1999), the elastic modulus E and Poisson’s ratio
vary slightly from one  angle to the other. In the interest of simplicity, the average
values of E=66,586 MPa and
=0.345 are used throughout the computation. The
following constants have been used in this computation: F=4.919 106 (MPa)2,
G=8.862 106 (MPa)2, H=5.309 106 (MPa)2, N=0.000035 (MPa)2, D =2.4
for all ’s, C=300 MPa, and  varies as a function of  as

 ¼ 8:50464 þ 29:4053  32:7892 þ 11:797 þ 0:6 ð37Þ

This functional relation has been found by use of experimental data for the cases
of  ¼ 0; 35; 45; 55; and 90 . An improved relation would have been obtained if
H.-C. Wu / International Journal of Plasticity 18 (2002) 1661–1682 1673

Fig. 2. The tensile stress–strain curves for AA7108–T1 aluminum: (a) =0 , (b) =45 , and (c) =90 .
1674 H.-C. Wu / International Journal of Plasticity 18 (2002) 1661–1682

Lademo et al. (1999) had reported data for 15 and 75 . The author has tried to
express (37) by simple mathematical expressions without success. The initial back
stress has been determined to be r0x =11 MPa, r0y =44 MPa, and r0xy =0. Constants
(F, G, H, N) and back stress (r0x ; r0y ; r0xy ) have been determined to fit the experi-
mental data for all measured  , i.e. 0 ; 35 ; 45 ; 55 ; 90 ; and to satisfy the con-
straint imposed by (34).
The theoretical stress–strain curves (the solid curves) are compared with the
experimental data of Lademo et al (1999) (the dotted curves) in Fig. 2. Only the
cases of =0, 45, 90 are shown. Curves for =35 and 55 have also been obtained
with a similar degree of agreement with the experimental data.
Fig. 3 shows the flow stress plotted against  at the initial yielding and at constant
strain levels of 0.01, 0.02, 0.03, 0.1 and 0.14. Both theoretical curves and experi-
mental data are shown. The theoretical curves show maxima occurring at about 15
and 75 , but no experimental data in these regions have been reported by Lademo et
al. (1999). Fig. 4 shows the R-ratio plotted against . The experimental R-ratio was
determined by applying linear regression to the plastic width strain versus thickness
strain curve for each , which did not deviate much from a straight line. The R-ratio
has only changed slightly as strain increases for this material. Two theoretical curves
are shown: one for initial yielding and the other for strain at 0.14. Although there is
a general agreement between the theory and experiment, the poor agreement for
=90 can not be explained.
Fig. 5 shows the stress-paths for tension tests for all  angles as the longitudinal
strain of the specimen increases. All are linear paths going from A to B, where B is
the far end of each path. These paths do not stay in the (x ; y ) plane except for the

Fig. 3. Flow stress versus  for AA7108–T1 aluminum at constant strain levels of initial yielding, 1, 2, 3,
10 and 14%.
H.-C. Wu / International Journal of Plasticity 18 (2002) 1661–1682 1675

Fig. 4. R-ratio versus  for AA7108–T1 aluminum at initial yielding and 14% strain.

cases of  ¼ 0 and 90 , but they are linear paths defined by equation (11). Both the
plane view [Fig. 5(a)] and the elevation view [Fig. 5(b)] are shown. As it was men-
tioned earlier, the plane view is the projection of a three-dimensional figure onto the
(x ; y ) plane; while the elevation view is the projection of the same figure onto a
plane which passes through the xy axis and the s-direction. The number marked at
the far end of each path corresponds to the  angle for that path. The corresponding
back-stress-paths may be drawn and they are generally nonlinear paths. But, they
are not shown due to the smallness of these paths.
Finally, Fig. 6 shows the gradual change in direction of the ‘‘effective stress vec-
tor’’ for the case of =45 . Similar figures may be drawn for other  angles. Fig. 6(a)
shows the plane view and Fig. 6(b) the elevation view. The effective stress vector is the
quantity x ; y ; xy defined by Eq. (9); it is represented by the vector from C to A
when the axial strain is 0.005, and by the vector from D to B when the axial strain is 0.1.
It is seen from the figures that this vector changes direction as the axial strain increases.
When this vector changes direction, the corresponding stress point moves along the
yield surface, resulting in the change of direction of the corresponding plastic strain-
rate vector.
The second example is related to a set of uniaxial tension experiments conducted
in the writer’s laboratory by Y.P. Shiao and M. Loureiro. The experimental results
are reported in Shiao (2000). The material was a 6061-O aluminum sheet tested at
the as-received condition. Tension specimens were cut at 0, 15, 30, 45, 60, 75 and 90
from the rolling direction. Three samples were tested for each case. Post-yield strain
gauges, one along the axial direction and one along the width direction of the spe-
cimen, were used to measure strains for each specimen.
The theory of the previous section is used to model the tests. The same procedures
as in the first example are followed in the determination of constants and in the
1676 H.-C. Wu / International Journal of Plasticity 18 (2002) 1661–1682

Fig. 5. Stress-paths for tension tests of AA7108–T1 aluminum with  shown: (a) plane view, (b) elevation
view.

computation. The following constants are used in the calculation: E=70 GPa,

=0.33, F=0.975 104 (MPa)2, G=1.2546 104 (MPa)2, H=1.0474 104


(MPa)2, N=2.6611 104 (MPa)2, C=0.2 MPa, =2.7 for all ’s, and D are
given by
H.-C. Wu / International Journal of Plasticity 18 (2002) 1661–1682 1677

Fig. 6. Direction of effective stress vector for tension test of AA7108–T1 aluminum at =45 . (a) Plane
view, (b) elevation view.

Fig. 7. The tensile stress–strain curves for 6061–O aluminum, =0 .


1678 H.-C. Wu / International Journal of Plasticity 18 (2002) 1661–1682

Fig. 8. Width strain versus axial strain for 6061–O aluminum, =0 .

Fig. 9. R-ratios for 6061–O aluminum: (a) =0 , (b) =30 , (c) =60 , and (d) =90 .

 m
0
D ¼ D0  ð38Þ

where D0=2.9 is the value of D at =0 and m=0.5. Note that   are initial yield
stresses and are known for each . In effect, (38) is expressing D as a function of .
The initial back stress has been determined to be r0x =20.8 MPa, r0y =13 MPa,
and r0xy =0.
The theoretical stress–strain curves (the solid curves) are now compared with the
experimental data (the dotted curves). Only the case of =0 is shown in Fig. 7. The
stress-strain diagrams indicate only a small influence of anisotropy for this material,
even though a significant anisotropy is found in the initial yield stress and R-ratio,
which will be further discussed. Fig. 8 shows the calculated width strain plotted against
the axial strain for the case of =0 . The dotted curve indicates the experimental
H.-C. Wu / International Journal of Plasticity 18 (2002) 1661–1682
1679
Fig. 10. Theoretical R-ratio for 6061–O aluminum at constant strain levels of 1, 2, 4 and 7%.
1680 H.-C. Wu / International Journal of Plasticity 18 (2002) 1661–1682

Fig. 11. The initial yield stress versus  for 6061–O aluminium.

data. The calculated R-ratios are plotted against the axial strains in Fig. 9 for the cases
of =0, 30, 60 and 90 , where the dots denote experimental data. Reasonable agree-
ment between theory and experiment has been achieved. Fig. 10 shows the theoretical
R-ratios plotted against  at constant strain levels of 1, 2, 4 and 7%. No experimental
data for larger strain levels have been obtained due to necking of the specimens. Fig. 10
is constructed directly from Fig. 9(a)–(d). No experimental data are shown in Fig. 10
due to scatter, but a direct comparison between theory and experiment can be made
in Fig. 9. The initial yield stress is plotted against  in Fig. 11 with dots denoting
again the experimental data.

5. Summary and conclusion

A simple anisotropic theory of plasticity has been proposed. This theory uses
Hill’s (1948) quadratic yield criterion, the normality condition for the flow rule, and
a combined isotropic-kinematic hardening rule. This theory is, in fact, the same as
that usually used in the classical plasticity. The present research merely confirms that
the flow theory of plasticity may be extended to the case of anisotropic sheet metals.
Although this is not surprising, the theory has not been previously verified against
experimental data related to sheet metals. The results of this research indicate that
the conventional non-quadratic yield functions are not necessary for sheet metals,
and the proposed theory may be useful for analyses of sheet metal forming.
In both examples considered, the present theory has achieved a general agreement
with experimental results in both aspects of anisotropy (yield stress and R-ratio).
The agreement is not perfect due to the simple expressions used to describe isotropic
and kinematic hardening given by Eqs. (25) and (27), and there are only two func-
tions D and  and a parameter C in the equations. The agreement shown in this
H.-C. Wu / International Journal of Plasticity 18 (2002) 1661–1682 1681

paper is very satisfactory, if one realizes that the experimental data of Lademo et al.
(1999) have been reproduced by Eq. (36) using three Qk s and three Ck s for each  to
describe the axial stress- axial strain relation and an additional set of three Qk s and
three Ck s to describe the axial stress-width strain relation.
The anisotropy has two folds (yield stress and R-ratio) as previously mentioned.
In the two examples considered, the materials are very different. They are different
in the magnitude of yield stress and also in its variation with . They are also dif-
ferent in the magnitude of the R-ratio. In fact, the R-ratio for AA7108-T1varies
from 0.3 at =0 to 2.0 at =55 ; while that for 6061-O aluminum remains less
than 1.0 for all s. In order to describe such a wide range of variations of material
behavior and still using the same constitutive equation, the material functions are
different for the two materials. For AA7108-T1, D are taken as constant for all s
and  vary with  as in Eq. (37); and, for 6061-O aluminum,  are taken as con-
stant for all s, but D are made to vary with  as in Eq. (38). The two variations are
not the same due to different materials in the two examples.
The definitions of yield and their effect on the hardening rules of the sheet material
have been discussed. It has been shown that strain-hardening plays an important
role in the determination of the R-ratios and yield stresses, which are obtained from
uniaxial tension specimens, cut at an  angle measured from the rolling direction of
the metal sheet. Furthermore, kinematic hardening makes it possible to have the
directional change in the plastic strain-rate without introducing the non-associated
flow rule into the formulation as suggested by Lademo et al. (1999).

Acknowledgements

This research is supported by the US National Science Foundation (DMI97-


00103).

References

Barlat, F., Lian, J., 1989. Plastic behavior and stretchability of sheet metals. Part I: a yield function for
orthotropic sheets under plane stress conditions. Int. J. Plasticity 5, 51–66.
Bertsch, P.K., Findley, W.N., 1962. An experimental study of subsequent yield surfaces-corners, normal-
ity, Bauschinger effect and allied effects. Proc. 4th US Nat. Cong. Appl. Mech. 893–907.
Cao, J., Yao, H., Karafillis, A., Boyce, M.C., 2000. Prediction of localized thinning in sheet metal using a
general anisotropic yield criterion. Int. J. Plasticity 16, 1105–1129.
Cho, H.W., Dafalias, Y.F., 1996. Distortional and orientational hardening at large viscoplastic deforma-
tions. Int. J. Plasticity 12, 903–925.
Dafalias, Y.F., 2000. Orientation evolution of plastic orthotropy in sheet metals. J. Mech. Phys. Solids 48,
2231–2255.
Gotoh, M., 1977a. A theory of plastic anisotropy based on a yield function of fourth order (Plane stress
state)—I. Int. J. Mech. Sci. 19, 505–512.
Gotoh, M., 1977b. A theory of plastic anisotropy based on a yield function of fourth order (Plane stress
state)—II. Int. J. Mech. Sci. 19, 513–520.
Hill, R., 1948. A theory of yielding and plastic flow of anisotropic metals. Proc. Roy. Soc. London A193,
281–297.
1682 H.-C. Wu / International Journal of Plasticity 18 (2002) 1661–1682

Hill, R., 1950. The Mathematical Theory of Plasticity. Clarendon Press, Oxford. pp. 317–324.
Hill, R., 1979. Theoretical plasticity of textured aggregates. Math. Proc. Camb. Phil. Soc. 85, 179–191.
Hill, R., 1990. Constitutive modelling of orthotropic plasticity in sheet metals. J. Mech. Phys. Solids 38,
405–417.
Hill, R., 1993. A user-friendly theory of orthotropic plasticity in sheet metals. Int. J. Mech. Sci. 35, 19–25.
Hosford, W.F., 1972. A generalized isotropic yield criterion. J. Appl. Mech. 607–609.
Huang, H.-M., Pan, J., Tang, S.C., 2000. Failure prediction in anisotropic sheet metals under forming
operations with consideration of rotating principal stretch directions. Int. J. Plasticity 16, 611–633.
Ivey, H.J., 1961. Plastic stress–strain relations and yield surfaces for aluminum alloys. J. Mech. Eng. Sci.
3, 15–31.
Karafillis, A.P., Boyce, M.C., 1993. A general anisotropic yield criterion using bounds and a transfor-
mation weighting tensor. J. Mech. Phys. Solids 41, 1859–1886.
Kim, K.-H., 1992. Evolution of anisotropy during twisting of cold drawn tubes. J. Mech. Phys. Solids 40,
127–139.
Kim, K.-H., Yin, J.J., 1997. Evolution of anisotropy under plane stress. J. Mech. Phys. Solids 45, 841–851.
Kuroda, M., 1997. Interpretation of the behavior of metals under large plastic shear deformations: a
macroscopic approach. Int. J. Plasticity 13, 359–383.
Kreissig, R., Schindler, J., 1986. Some experimental results on yield condition in plane stress state. Acta
Mechanica 65, 169–179.
Lademo, O.-G., Hopperstad, O.S., Langseth, M., 1999. An evaluation of yield criteria and flow rules for
aluminum alloys. Int. J. Plasticity 15, 191–208.
Lege, D.J., Barlat, F., Brem, J.C., 1989. Characterization and modeling of the mechanical behavior and
formability of a 2008-T4 sheet sample. Int. J. Mech. Sci. 31, 549–563.
Mair, W.V., Pugh, H.L.D., 1964. Effect of prestrain on yield surfaces in copper. J. Mech. Eng. Sci. 6, 150–
163.
Naghdi, P.M., Essenburg, F., Koff, W., 1958. An experimental study of initial and subsequent yield sur-
faces in plasticity. J. Appl. Mech. 25, 201–209.
Phillips, A., Tang, J.L., 1972. The effects of loading path on the yield surface at elevated temperatures.
Int. J. Solids Structures 8, 463–474.
Prager, W., 1955. The theory of plasticity: a survey of recent achievements (James Clayton Lecture). Proc.
Instn. Mech. Engrs. 169, 41–57.
Shiao, Y.P., 2000. A Study on Multiaxial Cyclic Loading and Anisotropic Plasticity of Structural Metals.
PhD thesis, National Taiwan University, Taiwan.
Simo, J.C., Hughes, T.J.R., 1998. Computational Inelasticity. Springer, New York.
Szczepinski, W., 1963. On the effect of plastic deformation on yield condition. Arch. Mechanics 15, 275–295.
Tugcu, P., Wu, P.D., Neale, K.W., 1999. Finite strain analysis of simple shear using recent anisotropic
yield criteria. Int. J. Plasticity 15, 939–962.
Tugcu, P., Neale, K.W., 1999. On the implementation of anisotropic yield functions into finite strain
problems of sheet metal forming. Int. J. Plasticity 15, 1021–1040.
Wu, H.C., in press. On finite plastic deformation of anisotropic metallic materials. Int. J. Plasticity.
Wu, H.C., Hong, H.K., 1999. Endochronic description of plastic anisotropy in sheet metal. Int. J. Solids
Structures 36, 2735–2756.
Wu, H.C., Loureiro, M., 1999. An anisotropic plasticity for sheet metals. In: Zabaras, N., Becker, R.,
Lalli, L., Ghosh, S. (Eds.), The Integration of Material, Process and Product Design. Balkema, Rot-
terdam, pp. 93–102.
Wu, H.C., Yeh, W.C., 1991. On the experimental determination of yield surfaces and some results of
annealed 304 stainless steel. Int. J. Plasticity 7, 803–826.
Wu, H.C., Yip, M.C., 1981. Endochronic description of cyclic hardening behavior for metallic materials.
J. Eng. Mater. Technol. 103, 212–217.
Wu, H.C., Hong, H.K., Shiao, Y.P., 1999. Anisotropic plasticity with application to sheet metals. Int. J.
Mech. Sci. 41, 703–724.
Yoon, J.W., Barlat, F., Chung, K., Pourboghrat, F., Yang, D.Y., 2000. Earing predictions based on
asymmetric nonquadratic yield function. Int. J. Plasticity 16, 1075–1104.

You might also like