You are on page 1of 20

International Journal of Plasticity, Vol. 7, pp. 693-712, 1991 0749-6419/91 $3.00 + .

00
Printed in the U.S.A. Copyright © 1991 Pergamon Press plc

A SIX-COMPONENT YIELD FUNCTION


FOR ANISOTROPIC MATERIALS

FR~Dt~RIC BARLAT, DANIEL J. LEGE, a n d JOHN C . B ~ r a

Alcoa Laboratories

Abstract- In classical flow theory of plasticity, it is assumed that the yield surface of a mate-
rial is a plastic potential. That is, the strain rate direction is normal to the yield surface at the
corresponding loading state. Consequently, when the yield surface is known, it is possible to
predict its flow behavior and, associated with some failure criteria, to predict limit strains above
which failure occurs. In this work a new six-component yield surface description for orthotro-
pic materials is developed. This new yield function has the advantage of being relatively sim-
ple mathematically and yet is consistent with yield surfaces computed with polycrystal plasticity
models. The proposed yield function is independent of hydrostatic pressure. So, except for such
cases, strain rates can be calculated for any loading condition. Applications of this new cri-
terion for aluminum alloy sheets are presented. The uniaxial plastic properties determined for
2008-T4 and 2024-T3 sheet samples are compared to those predicted with the proposed consti-
tutive model. In addition, for 2008-T4, the predictions of the six-component yield function are
compared to those made with the plane stress tricomponent yield criterion proposed by Barlat
and Lian. Though rather good agreement between experiments and predicted results is obtained,
some discrepancies are observed. Better agreement could result if the isotropic work-harden-
ing assumption associated with the yield criterion were relaxed. Nevertheless, the proposed yield
function leads to plastic properties similar to those computed with polycrystalline plasticity mod-
els and can be very useful for describing the behavior of anisotropic materials in numerical
simulation of forming processes.

!. INTRODUCTION

This paper deals with the plastic behavior of metals subjected to deformation conditions
such that temperature and strain rate effects can be neglected. In particular, this study
can be applied to most aluminum alloys deformed at room temperature. When a ten-
sion test specimen is subjected to uniaxial loading, there is a given critical stress level
under which the material behavior is elastic. Above this critical stress or yield stress, ir-
reversible plastic deformation occurs. If the specimen is unloaded and reloaded, the same
phenomenon can be observed, but the yield stress on reloading corresponds to the stress
just prior to unloading. The questions which must be addressed in order to describe the
elasto-plastic behavior of a material are (ZYczXowsra [1981]): What stress results in in-
elastic response (yield condition)? How are plastic deformation increments related to
stress (flow rule)? How does the yield condition change with deformation (hardening)?
The last two questions relate to the postyield conditions. Experimentally, yield stress
measurement is not an easy task. Several different definitions of the yield stress have
been proposed in the past (for instance, see HECI~R [1976]). This difficulty can be ex-
plained by considering the two scales of microscopic and macroscopic yielding. Plastic
flow in metallic materials occurs by dislocation glide on given crystallographic planes
and directions. On a particular slip system, the movement of a dislocation can be re-
versible with or without energy dissipation. In addition, slip does not occur at the same
time in all the grains of the polycrystalline aggregate because of their different orien-
tations. Therefore, plastic deformation is nonhomogeneous at the small strain levels.

693
694 F. BARLATe t al.

In this work, the relatively large plastic strain range is of interest and the conventional
0.2% offset plastic strain is adopted as a definition of yielding. At this deformation level,
all of the grains are assumed to be subject to plastic deformation.
When multiaxial loading is considered, the questions discussed for the uniaxial case
are still valid. What are the yield and the postyield conditions? However, the yield con-
dition does not reduce to a critical number, the yield stress. Instead, it is expressed by
a function describing the yield surface. In the most general case, this function • is de-
fined in the six-dimensional space of the stresses, and the material is

elastic if ~ ( o ~ ) < c, c = constant (la)

or

plastic if ~ ( o ~ ) = c. (Ib)

• ( o ~ ) > c has no physical meaning. For multiaxial loading, the postyield conditions
are not given by a simple stress-strain curve. One has to know the relationship between
the stress and the strain-rate tensor (flow rule) and how the yield surface shape evolves
during deformation. In the following section, the yield conditions and flow rules that
characterize the behavior of isotropic and anisotropic materials are succinctly described.

I!. YIELD CONDITIONS AND FLOW RULES

The yield surface represents the limitation of the elastic range. Experimental evidence
and theoretical considerations concerning the plastic behavior of materials have led to
some restrictions on the mathematical representation of the yield surface. It is well
known that yielding in metals does not depend on a superimposed hydrostatic pressure
and that plastic deformation occurs with no volume change. These observations are not
true for materials exhibiting high porosity, for materials subjected to structural trans-
formations during plastic straining, for polymers, or for materials subjected to very high
pressure. However, for heavily cold worked materials, the relative density change due
to the multiplication of dislocations from the undeformed state is 1012b 2 ~ 10 -3, where
b is the Bfirgers vector (FRANCOIS [1976]). For commercial purity sheet metals, the rel-
ative density change due to void nucleation and growth at hard particles during plastic
deformation is of the same order of magnitude, 10 -3 (ScHMITT & JALrNmR [1982]), and
for plasticity flow rules, these corresponding volume changes can be neglected. For pres-
sure independent material behavior, the yield surface is assumed to be a function of the
stress deviator S defined by the tensor equation:

S = O - - Om I (2)

where Om is the mean pressure and I the identity tensor. Thus, for isotropic materials,
the yield function can be defined as a function of two stress deviator invariants only.
Two different approaches can be used to derive the yield surface (ZYCZKOWSra [1981]).
One approach assumes that the limitation of the elastic range is the result of a physical
quantity which reaches a critical value. For instance, the isotropic von Mises and Tresca
yield criteria are obtained if the elastic range is bounded by a certain value of the elas-
tic distortion energy or by a critical level of the maximum shear stress, respectively. The
von Mises yield function can be derived, as well, if yielding is assumed to occur when
Six-componentyield function for anisotropic materials 695

the octahedral shear stress reaches a given value. The yield surface o f a single crystal
results from the assumption o f a critical resolved shear stress on certain slip systems.
Using the other approach, a yield function can be chosen as a good approximation
for a set o f data generated either by experimental techniques or by physics-based mod-
eling. For instance, the following phenomenological yield function proposed by HER-
SHEY [1954] and HOSFORn [1972]:

,/i = IS l _ S2I m + IS= - S3l m + - s,I m = 2~m (3)

is a particularly good approximation of yield surfaces calculated with polycrystal models


for isotropic BCC and FCC metals when the exponent m is equal to 6 and 8, respec-
tively (LOGAN & HOSFORD [1980]). In this equation, # is the effective stress. It is inter-
esting to note that the function defined by eqn (3) reduces to the von Mises criterion for
m = 2 and m = 4, and to the Tresca condition for m = 1 and m = 0o.
The plastic behavior of materials is not linear. So, in the classical flow theory of plas-
ticity, the flow rule gives a relationship between the stress deviator and the strain-rate
tensor. For instance, the Levy-Mises relations assume the identity between these two ten-
sor quantities. If the von Mises yield condition # is associated with the Levy-Mises re-
lation, then the flow rule can be written as follows:

~ij = A --.0# (4)


o/j

The yield function is considered a plastic potential or, in other words, the strain rate
vector is orthogonal to the yield surface. The associated flow rule given in eqn (4) can
be generalized to represent the behavior of other isotropic and anisotropic materials.
HECKER [1976], who reviewed numerous critical experiments to assess the yield surface
shape, found that the normality rule was never violated. The yield surface shape and,
consequently, the associated flow rule must be described as accurately as possible in or-
der to model the plastic behavior of materials. BARLAT[1987] has shown that the theo-
retical sheet forming limit strain at necking for balanced biaxial stretching is roughly
five times larger for a material whose yield surface is described by the von Mises crite-
rion than for one which follows the Tresca criterion, when all other properties are iden-
tical. In addition, when using the phenomenological description (eqn (3)) with an
exponent m = 8, the theoretical forming limit strains reasonably agree with experimental
trends, whereas they are unrealistically too high or too low with the von Mises or the
Tresca yield conditions, respectively.
DRtrCKER [1951, 1959] showed that, based on a stability postulate, the yield surface
must be convex. If the yield function is smooth with no vertex, convexity assures the
uniqueness o f the strain rate for a given stress state. Mathematically, the yield surface
is convex if for a given stress and strain rate condition a o and ~o, any stress oi~ in-
side or on the yield surface obeys the following relationship:

(ao - ai~)~ij -> 0. (5)

If the function # is twice differentiable, its Hessian matrix 3£ is defined by:

02#
~cu- aa,a~j" (6)
696 F. BARLATel al.

is convex if the matrix ~ is positive semi-definite, that is, if its eigenvalues are posi-
tive or zero (Eca3I~STON [1958]; ROCr,.A~tJ~R [1972]). In addition, these authors showed
that a linear transformation on the variables does not affect convexity. LIVPU_ANN[1970]
showed that if a yield function is convex in the three-dimensional space of the principal
stresses, then it is also convex in the more general six-dimensional stress space. There-
fore, for isotropic yield functions, the convexity needs to be checked only in the three-
dimensional principal stress space. It can readily be shown that the yield function given
by eqn (3) is convex for m equal or larger than 1.

IIL YIELD FUNCTIONS FOR ISOTROPIC MATERIALS

Phenomenological yield functions do not result directly from microstructure-based


calculations. However, they have some advantages over the yield surfaces calculated
from the crystallographic texture of polycrystalline aggregates. Namely, phenomenolog-
ical yield functions are easy to implement (in FEM codes for instance), and lead to fast
computations. They can describe the complete anisotropy, whereas polycrystal models
only account for crystallographic texture. Finally, they are easy to adapt to different ma-
terials. For instance, using eqn (3), the difference between a FCC and a BCC isotropic
material can be modeled by a change of exponent m. However, using polycrystal models,
the same difference requires profound modifications, such as a change in slip systems,
for instance.
As mentioned previously, eqn (3) gives an excellent approximation of an isotropic
polycrystal yield surface calculated using the TAYLOR [1938J/BisHoP and HILL [1951]
model, or using a self-consistent model (H~Rsrmv [1954]). Therefore, the main purpose
o f this work is to generalize eqn (3) for anisotropic materials. However, eqn (3) does
not possess shear stress terms, and this restricts its possible use for anisotropic gener-
alizations to stress states where the anisotropy axes and principal stress axes are super-
imposed. Therefore, this section describes how shear stress terms can be introduced in
the isotropic formulation. In a first approach, only plane stress states where three out
of the six stress components are equal to 0 (azz = Ozx = Oyz = 0) are considered. The
stress tensor is represented by a 2 x 2 matrix whose characteristic equation is a poly-
nomial of the second order. Therefore, a simple calculation of the eigenvalues leads to
the isotropic, tricomponent, plane-stress yield surface:

= IK, - KE] m + IK1 + K21 m + 12K21 m = 2a '~ (7a)

Kl -- Oxx + ayy (7b)


2

g2---
J/ "~" /2
°xx -- Oryy 2
"l- 0"xy. (7c)

This formulation was proposed by BARLATand RICHMOND [1987] and it was extended
to planar anisotropy by BARLAr et al. [1988], and BARLATand LIAN [1989]. According
to LIVPMANN'S work [1970], the convexity requirement is obviously fulfilled since the
function describes the same yield surface as eqn (3) in another stress space. Figures 1
and 2 illustrate the excellent agreement between this yield function and the TAYLOR
[1938]/BISHOP and HILL [1951] yield surface for isotropic FCC materials (from BARLAT
& LtAN [1989]). In this type of plot, the solid lines represent the intersections of the yield
Six-component yield function for anisotropic materials 697

1.5

~ / - - - p =1
1.0

X Ai
0.5

__i/'-,,, 7--./ 'y'.ff

-0.5

-1.0 I I
.0 -0.5 0 0.5 1.0 1.5
Fig. 1. Tricomponent plane stress yield surface (oxx/#, Oyy/O, Oxy/O) for an isotropic FCC sheet calculated
with the Bishop and Hill model. The solid line represents the section of the yield surface by a plane parallel
to (o~/O, oyy/o) for different values S = o~/O and projected onto the (o=/O, oyy/#) plane. Dashed lines rep-
resent the projection onto (o~/0, ayr/O) of curves defined by points having the same strain ratio o (~yy/~=).

surface with a plane parallel to (Oxx/#, oyy/#) for different values o f S = oxy/# a n d p r o -
j e c t e d o n t o the (ox~/6, Oyy/0) p l a n e .
A n expression o f the yield surface described b y eqn (3) in terms o f all six c o m p o n e n t s
o f the stress tensor can be o b t a i n e d after solving the third-degree characteristic equation:

/~3 - - 312A - 2/3 = 0 (8)

where - 3 / 2 a n d 2/3 a r e t h e second a n d the t h i r d p r i n c i p a l invariants o f the stress devi-


a t o r , respectively. E x p r e s s e d with the BISHOP a n d HILL [1951] n o t a t i o n o f the stresses,

A = Oyy - Oz~ B = o ~ - Oxx C = Oxx - Oyy (9a,b,c)

F = Oyz G = o= H = Oxy (9d,e,f)

the stress d e v i a t o r i n v a r i a n t s t a k e the f o l l o w i n g f o r m :

F2+G2+H 2 (A-C)2+(C-B)2+(B-A)2
12 = + (10a)
3 54
698 F. BARLATet al.

1.5 I ~yy/~ I

1.0

0.5

~xx/~

-0.5 -

-1.0
-1.0 -0.5 0 0.5 .0 1.5

Fig. 2. Tricomponent plane stress yield surface obtained with eqn (3) for m = 8.

(C-B)(A - C)(B -A) (C-B)F 2 + (A - C)G2+ ( B - A ) H 2


13 = + FGH-
54 6
(lOb)

Since eqn (8) is the characteristic equation of a symmetric matrix, the three solutions
(principal stresses) are real. However, it is more convenient to express them in terms of
complex numbers. Let us denote z the complex number of module p and argument 0:

z = 13 + i ( I 3 - 12) 1/2 = pe i° = p(cos 0 + i sin 0) (11)

where i is the imaginary unit number, p = 13/2, and 0 = arc c o s ( I 3 / I 3 / 2 ) . The princi-
pal deviatoric stresses are given by Cardan's solutions of eqn (8):

S l = z l / 3 .~_ ~1/3 (12a)

S2 = o)zl/3 + (.~1/3 (12b)

S3 = ~_oZ1/3 -~ 0oZ 1/3 (12c)


Six-component yield function for anisotropic materials 699

where co = e - 2 i x / 3 and the bar ( - ) stands for conjugate. The yield function described
b y eqn (3) can be rewritten as follows:

= I(1 - co)z ~/3 + (1 - ~)Z1/3lm + I(co - - ~)zl/3 -st- (¢~ - - c°)Zl/3lm


(13)
+ I(~ - 1)Z 1/3 + (co - 1)~ 1/31m = 26 m

or, alternatively:

~i1=5 (312)m/2I 2COS~T]/2~'t-'ff~m-F 2COS(20--371")


"Jt''6m 2COS(206 571")m/
= 26". (14)

It is interesting to examine the geometrical interpretation o f the quantities mentioned


a b o v e in the complex plane (Fig. 3). co, & and 1 are 120 ° apart and their sum is zero.
These properties also hold for the quantities 1 - co, ¢3 - 1 and co - t3.
Since the a r g u m e n t o f the p r o d u c t o f two complex n u m b e r s is equal to the sum o f
their arguments, it is possible, using eqns (12), to visualize o n Fig. 3 the complex n u m -
bers that generate the principal deviatoric stresses $1, $2 and $3. Then it is relatively

Imaginary axisi zl/3(I - co!

zl/3(co 1 ) ~ ;";:"::';~"~ \ zl/3


Real axis

Z I/3(f.l)-- ~)

Fig. 3. Stress deviator invariants in complex plane.


700 F. BARLAT et al.

easy to find the position o f the complex numbers whose real part ((R0 gives the differ-
ences o f the deviatoric stresses.

S1 - S2 = 2(RelZl/3( 1 -- w)} (15a)

$2 - $3 = 2 ( R e l z l / 3 ( o~ - ~)1 (15b)

$3 - $1 = 2(Relzl/3(& - 1)}. (15c)

Noting that sin 0 = (I~ - I 2 )1/2//132/2 is always positive or, equivalently that 0/3 is always
between 0 ° and 60 °, the above mentioned complex numbers are located on the shaded
area on Fig. 3. This implies that $1 - $2 as well as $2 - $3 are always positive, whereas
$3 - SI is always negative.
Therefore, the absolute values in eqn (14) are not necessary and the yield function re-
duces to:

#=(312)m/2[[2COS(20_~r)]m + (2COS(20 - -.3 r ) ] m + [ - - 2 c o s ( 2 0 + 65 7 r ) ]m]

= 2 ~ m. (16)

Keeping in mind that -3•2 is the second invariant of the stress deviator and introduc-
ing a constant k, the yield surface described by eqn (16) can be expressed by:

= 312 q'(0) = k 2. (17)

This is the generic form of the generalized isotropic yield criterion studied recently by
BILLINGTON [1988]. Eqn (16) obviously reduces to the von Mises criterion when m = 2
or m = 4 and to the Tresca criterion when m = 1 or m = 0o.
Finally, it can be shown that the yield surface can be expressed with polar coordinates
R(X) in the ,r-plane by

1/m
2
R = (18)
I cos x - x/3 sin X m cos X + 4-3 sin X
m
+ I~cosxl m 1
with X = 0/3 - 7r/6, when X is between - 7 r / 6 and ~r/6. Figure 4 illustrates this repre-
sentation for different values of m.

IV. YIELD F U N C T I O N S F O R A N I S O T R O P I C M A T E R I A L S

In this section, the isotropic formulation proposed previously is extended to the an-
isotropic case. This study is restricted to anisotropic materials exhibiting orthotropic
symmetry, that is, to materials that possess three mutually orthogonal planes of sym-
metry at every point. Actually, this restriction is not a very significant constraint because
most mechanically processed materials are orthotropic. Typical examples are roiled
sheets or plates.
It is interesting to note that the definition of symmetry is related to a point in the ma-
terial and not related to the whole part. Moreover, the three-dimensional material space
Six-component yield function for anisotropic materials 701

Von Mises
(m-2 or 4) m-8

Tresca
(m=l or oo)
Z

Fig. 4. Isotropic yield functions for m = 1 or ~, m = 2 or 4, and m = 8 in r-plane.

defined by x, y, a n d z should not be confused with the six-dimensional stress space which
reduces to a t h r e e - d i m e n s i o n a l space in the p l a n e stress case. T h e o r t h o t r o p i c symme-
try o f the material space implies some s y m m e t r y o f the yield surface in stress space, b u t
this s y m m e t r y is n o t o r t h o t r o p i c s y m m e t r y .
Table 1 lists some o f the m a j o r yield f u n c t i o n s p r o p o s e d to describe the a n i s o t r o p i c
behavior o f orthotropic materials. Some o f the isotropic functions are also included for
reference. B ~ A T [1987] has discussed the limitations o f each o f these criteria. The ba-
sic p r o b l e m can be summarized by considering the two yield conditions proposed by HrtL
[1948, 1979]. The first one can be applied to a n orthotropic material subject to a n y stress

Table 1. Phenomenological yield functions

Yield Criterion Type Shear Dimension

Tresca lsotropy - -
Von Mises lsotropy - -
HttL [1948] Planar anisotropy yes 6
HE,stray [1954] lsotropy -- --
HOSFORD[1972]
GOrOH [1977] Planar anisotropy yes 3
BAss~I [1977] Planar isotropy -- --
HILL [1979] Planar anisotropy no 3
Locxt~ & HOSFOgD[1980] Planar anisotropy no 2
BUDIANSKI[1984] Planar isotropy no 2

Shear indicates if shear terms appear in the formulation for anisotropic


materials. Dimension indicates the number of stress components used in
the formulation for anisotropic materials.
702 F. BARLATet al.

state. However, it gives a rather crude approximation of the yield surface shape com-
puted from polycrystal models and, therefore, does not provide good predictions of
material behavior. The second anisotropic yield function proposed by HILL can lead to
better results. However, this function does not include shear stress terms and is confined
to a three-dimensional stress space. As a result, orthotropic material axes and princi-
pal stress axes must be superimposed. This restricts the use of this function consider-
ably. In addition, this formulation has been shown to lead to nonconvex and sometimes
to nonbounded yield surfaces (Znu et al. [1987]; LIAN et al. [1989]).
In order to describe the plastic flow behavior of orthotropic polycrystals, the yield
function must be expressed in six-dimensional stress space, be convex, and must be in-
dependent of hydrostatic pressure. In addition, according to previous findings, it must
reduce to eqn (3) or another equivalent formulation in the isotropic case. BARLATand
LIA~ [1989] proposed a yield condition for the three-dimensional plane stress case, which
is very often assumed in sheet forming problems:

eI, = a[Ki -- Kzl m + alKl + K z [ " + (2 - a)[2K2[ m = 2 0 m (19a)

Kl - O×x + h o y y (19b)
2

K2 ~ / ( ° x x -- htYyY)2
+ p 2 Oxy
2 (19c)

where a, h, p and m are material constants. This anisotropic yield function has been ob-
tained by summing two convex functions, a IK1 - K21 m + a ]KI + K2 [m and ( 2 - a)] 2K2 ]m,
and using linear transformations of the stresses. Consequently, it is a convex function
which generalizes the plane stress yield function described by eqn (7). LEGE et al. [1989]
showed that this yield criterion was particularly accurate for the description of the con-
stitutive behavior of a 2008-T4 aluminum alloy sheet sample. Hu~L [1990] has concur-
rently proposed another description of the constitutive behavior of sheet metals. The
implications of this description have not been considered in this paper. However, Hn.L's
[1990] yield function applies to the three-dimensional plane stress case and, therefore,
has the same limitations as the function defined by eqns (19).
In order to generalize the isotropic yield function (eqn (16)) for the six-dimensional
case, it is necessary to consider the matrix representing the stress deviator, to express
it in the Bishop and Hill notation:

C-B

S =
Sxx Sxy
Sxy Syy Syz
Sz~] =
3

H
H

A-C
G

F (20)
3
Sz,, Syz Szz
B-A
G F
3

and to replace A , B, C, F, G, H, by aA, bB, cC, f F , gG, h H in this matrix where a, b,


c, f, g, and h are constants:
Six-componentyield function for anisotropic materials 703

cC- bB
hH gG
3

aA - c C
hH fF (21)
3

bB - aA
gG fF

This last matrix is still symmetrical and, therefore, has three real eigenvalues. In addi-
tion, its trace is zero and Cardan's solution used in the previous section for the isotro-
pic case is still valid. As a consequence, due to this particular choice of the constant
coefficients, z can always be defined. All the results obtained for the isotropic case ap-
ply for the anisotropic case when a A , bB, cC, f F , gG, h H are substituted for A, B, C,
F, G, H in 12 and 13 (eqns (10)):

[fF] 2 -.[- [gG} 2 + [ h i l l 2 (aA - cC) 2 + (cC - bB) 2 + (bB - aA) 2


= + (22a)
3 54

(cC - bB)(aA - cC)(bB - aA)


I~ = + fghFGH
54
(22b)
(cC- b B ) [ f F I 2 + (aA - c C ) { g G I 2 + (bB - a A ) [ h H I 2
6

o_ arccos( )
[2cos(20 +]m/ /
2oos(2065
2~ m ,
(22d)

The resulting anisotropic yield function is orthotropic and independent of hydrostatic


pressure. It reduces to the isotropic case when the constant coefficients a, b, c, f, g, and
h, are all equal to 1 (and particularly to Tresca for m = 1 or 2 and von Mises for m =
2 or 4). It accounts for the small radius of curvature of the yield surface near the uniaxial
or the biaxial tension range that is observed with polycrystalline yield surface calcula-
tions. This yield function is quite general and can be applied to orthotropic materials
for any stress state (except a hydrostatic pressure). Three of the six anisotropic coeffi-
cients can be obtained from the three uniaxial yield stresses in the directions of the sym-
metry axes using a Newton-Raphson numerical procedure. The other three coefficients
are derived from the three shear yield stresses, each related to two of the symmetry axes.
The exponent m can take any real value larger than 1 but practically, m should be larger
than 6, depending on the severity o f the texture, m can be determined such that it op-
timizes the predictions. Finally, it should be mentioned that the six-component aniso-
tropic yield function (eqn (22)) does not reduce to the tricomponent one (eqn (19))
because the linear transformations of the stresses, which must be made to get the an-
704 F. BARLATet al.

isotropic mathematical expression from the isotropic one, are not the same for the two
cases.
Calculation of the strain rates is somewhat lengthy but straightforward when using
the rule o f composition of derivation:

o~ - I o~ os~ o_~_~os____? o~ os3] o12


Oct~ . OS 1 012 + OS2 OIz + OS 3 012. Ooc,~
(23)
[ 0__~_~OS_._2 0___~_~0S_..~2 0~I' 0S3] 013
-t- OSl 013 @ 0S2 013 + 0S3 Oh 00,~"

The case z = ~ (or 0 = 0 ° and 0 = 180°), which corresponds to shear in the isotropic case,
is singular for the partial derivatives OSi/OIk. However, for this particular case, it is
possible to calculate the strain rates as a limit o f the general case when z - ~ --, 0.

V. APPLICATIONTO ALUMINUMALLOYS
In the present section, x, y, and z denote the rolling, transverse and normal directions
in a rolled sheet. The six-component yield function is used to predict uniaxial proper-
ties for loading directions in the plane of the sheet for two different materials: 2008-T4
and 2024-T3. For each alloy, uniaxial yield stresses are used to calculate the material
constants in the yield function. Then, R and F values defined by

~22
R = -- (24)
~33

p = --612 (25)

are predicted as a function of the tension loading direction. 1, 2, and 3 denote the lon-
gitudinal, width, and thickness directions of a tensile specimen. Comparisons with ex-
periments are made when data are available.
The chemical compositions (remelt analyses for the 2008 sample and typical data for
the 2024 sample) and thicknesses of the samples are given in Table 2. For each of the
sheet samples, specimens were taken at 0 °, 15 °, 30 °, 45 ° , 60 °, 75 °, and 90 ° to the rolling
direction. Standard uniaxial tension tests were conducted to determine 0.2% offset yield
strength, ultimate tensile strength, and total elongation. Single or duplicate (Alloy 2008)
tests were also conducted in the seven sheet orientations at a constant crosshead speed
of 0.1 in/rain to determine stress-strain behavior. Throughout these constant crosshead

Table 2. Thickness and compositiona of aluminum alloy sheets


Alloy and Temper Thickness(mm) Si Fe Cu Mn Mg
2008-T4 1.24 0.60 0.13 0.93 0 . 0 6 0.40
2024-T3 0.30 <0.50 < 0 . 5 0 4.40 0 . 6 0 1.50

aComposition in weight%.
Six-component yield function for anisotropic materials 705

speed tests, extensometers measured strains in both the loading (~H) and width (d22)
directions of the specimens. The plastic strain ratio (defined experimentally as R =
C22//E33) w a s derived from the slope of the true width strain/true length strain curve
evaluated from length strains of 0.01 to the uniform elongation assuming volume con-
stancy in the specimen gage section.
A square grid (approximately 0.75 mm x 0.75 mm) was applied to each 2008-T4 ten-
sile specimen using a photo-resist technique. After testing, grids in uniformly deformed
regions of the specimens were analyzed with a View Engineering 1220 video measure-
ment system to determine both the plastic strain ratio (R = c22//~33) and the shear strain
ratio (/" = e12/~11 ). The grids were not analyzed prior to deformation because they were
assumed to be of constant dimensions. This was found to be an erroneous assumption,
however, and resulted in some uncertainty in the parameters calculated from the grid
measurements. Nevertheless, the R values determined at a single point along the tensile
specimen strain path with this suspect procedure showed the same directionality trends
as those determined with extensometer measurements over the entire strain path.
X-ray techniques were used to determine the crystallographic texture of the 2008-T4
sheet. (l 11) and (200) poles figures were used to calculate the crystallographic orienta-
tion distribution function (CODF) using the expansion of the CODF in a series of gen-
eralized spherical harmonics (BuNGE [1965]; ROE [1965]). Corrections were made for the
so-called "ghost peaks" of the CODF using the odd series coefficients resulting from the
CODF positivity condition (BUNGE [1988]).
Figures 5a, 5b, and 5c present experimental and predicted results for the 2008-T4 sheet
sample. Figures 5a and 5b show, respectively, the yield stress and R value variations with
respect to the tension direction predicted with the six-component and the tricomponent
yield functions. The uniaxial yield stresses measured at 0 °, 45 °, and 90 ° from the roll-
ing direction where used to calculate the coefficients for the six-component yield func-
tion (Table 3), whereas the R values in these three directions were utilized to identify
the parameters for the tricomponent yield function (Table 4). This explains why the re-
spective experimental yield stresses and R values are almost superimposed on the theo-
retical curves. The R values predicted from the six-component model are slightly
underpredicted for uniaxial tension in directions close to the transverse direction (90°),
whereas the yield stress variation is underpredicted with the tricomponent yield func-
tion in the same range of tension directions. Perhaps it is not so obvious when looking
at the figures, but from experience, the differences observed between experimental and
theoretical results are significant for the yield stresses, whereas they are not for the R
values. Figure 5c shows the F values predicted with the two models as well as experi-
mentally measured values. Although LEGEet al. [1989] mentioned that the experimen-
tal data were not very accurate, the theoretical predictions result in a good trend. In fact,
it can be assumed that the experimental point F ~ 0.037 in the rolling direction (0 °) is
erroneous. This value should be zero because of symmetry in the rolled sheet.

Table 3. Coefficients of the six-component yield function

Material m a b c f g h

2008-T4 11 1.222 1.013 0.985 1.0 1.0 1.0


2024-T3 8 1.378 1.044 0.955 1.0 1.0 1.210
706 F. BARLATeta/.

1.1 I I I I I I I I

m = 11 (6 - comp. model)
rn = 8 (3 - comp. model)
A Exp.

1.0 •- - ~ - ~
tn
==
¢0

a)

0.9

0.8 I I I I l I I I
0 10 20 30 40 50 60 70 80 90
Angle between tension axis and rolling direction
(a)

1,0 t I = J t = I t

rn = 11 (6-¢ornp. model)
0,8 "_ ~ Z ~ ------ m = 8 (3-cornp. model)
- " ~ . A Exp,

0,6
i
e0
>
m
0,4

0.2 I I i I I I [ I
0 10 20 30 40 50 60 70 80 00
Angle between tenslon sxls and rolling dlrectlon
(b)

0.10 I I I I t so.J,,,~ I I ~m~'~ i


i. t %.
0.08 // x\

o.oe ,/' \
0.04 \\
o.o2 -
0"00 ~ L~
-0.02 m = 11 (6-cornp. model) _
----- rn = 8 (3-cornp. model)
-0.04 L~ Exp.
-0.06 i I I I I I I I
0 10 20 30 40 50 60 70 80 90
Angle between tenslon axls and rolllng dlrectlon
(c)
Fig. 5. (a) Normalized yield stress for 2008-T4 alloy sheet subjected to uniaxial tension in various directions
(values normalized with respect to 0 ° yield stress, 157 MPa). Prediction from the six-component model ( - - ) ;
and from the tricomponent model (--); experiments (4). (b) R value for 2008-T4 alloy sheet subjected to uniaxial
tension in various directions. Prediction from the six-component model ( - - ) ; and from the tricomponent model
(---); experiments (4). (c) F value for 2008-T4 alloy sheet subjected to uniaxial tension in various directions.
Prediction from the six-component model ( - - ) ; and from the tricomponent model (---); experiments (4).
Six-component yield function for anisotropic materials 707

Figure 6 shows the yield surfaces computed with the TAYLOR[1938]/BmHOP and Hr~L
[1951] (TBH) model (Fig. 6a), the tricomponent (Fig. 6b), and the six-component
(Fig. 6c) yield functions. For the 2008-T4 alloy sheet, the TBH yield surface results from
the assumption that each grain in the polycrystal is subjected to the same plastic strain
rate. There are only 56 possible stress states in each grain which can accommodate any
imposed plastic deformation, and the maximum plastic work principle is used to deter-
mine the active one. For an imposed strain rate, the macroscopic stress state is obtained

1.5 I I

1.0
a
I-
._c
m 0.5
o

e
_N
0.0
[ s=o./ y
o S = 0.S-
z
-0.5 - S = 0.4-
S = 0.3"
S =0.2"
b

-1.0 I I !

- 1.0 -0.5 0.0 0.5 1.0 1.5


N o r m a l i z e d s t r e s s in RD
(a)

1.5 I I I ~.3

1.0
a
I.,-
a:
m
m 0.6
o

"O
e
N
'3

o
[
0.0
s:o..1J y S =0.5
Z S=0.4
-0.5 S=0.3
S=0.2 S:0/

-1.0 I I I
-1.0 -0.5 0.0 0.5 1.0 1.5
N o r m a l i z e d s t r e s s in R D
(b)
Fig. 6. (a) Plane stress yield surfaces for 2008-T4 alloy sheet and computed with the Taylor/Bishop and Hill
model. (b) Plane stress yield surfaces for 2008-T4 alloy sheet and computed with the tricomponent yield
function. (Figure continued on the folio wing page.)
708 F. BARLATet al.

1.5 I I
~3yy

t,,- 1.0
===o
+g
10.=-
"0 "0
0.5
[=

s_os y
(~xx
0.0
0
0 e-
Z"
C s---ols
-0.5 S = 0.4 ~ .+.....-----.~
S = 0 . 3 ~ ~
S = 0.2 - - - ~
S=O.O ~
-1.0 I J I
-1 .I -0.5 0.0 0.5 1.0 1.5
Normalized yield stress in the rolling direction
(c)
Fig. 6 continued. (c) Plane stress yield function for 2008-T4 alloy sheet and computed with the six-component
yield function.

by averaging the stress in each grain using the crystallite orientation distribution func-
tion (CODF). These types of computations have shown the existence of rounded verti-
ces in the uniaxial or the biaxial tension range for the different types of crystallographic
textures found in aluminum alloys (BARLAXe, RlCm,tOND [1987]; BARLAX & FRIeKE
[1988]). The TBH yield surface (Fig. 6a) resulting from this computation procedure for
the 2008-T4 sheet exhibits a smaller yield stress in the transverse direction than in the
rolling direction, in agreement with the experiments. The computed TBH yield surface
shows a small radius of curvature, particularly in the biaxial tension range, as compared
to the plane strain or the shear range. The phenomenological models (Figs. 6b and 6c)
are able to precisely reproduce this yield surface shape. Therefore, these functions are
compatible with polycrystal models, and they are much easier to use.
Although both the tricomponent and the six-component anisotropic yield functions
reduce to the same isotropic criterion, they are not identical. The tricomponent model
was proposed after observing that the yield surfaces of textured polycrystals were
roughly approximated by an expansion or a contraction of the isotropic yield surface
in one or both directions ayy/# and Oxy/O. This is illustrated in Fig. 6b. However, there
are many ways to introduce the anisotropic coefficients in the six-component function,
and the particular solution proposed in this work was selected because it makes the func-
tion independent of the hydrostatic pressure. Since a different set of experimental data
have been used to calculate the coefficients of both functions, it is not appropriate to
compare both surfaces. It is clear, however, that the six-component anisotropic yield

Table 4. Coefficients of the tricomponent yield function

Material rn a h p

2008-T4 8 1.240 1.150 1.020


Six-component yield function for anisotropic materials 709

surface (Fig. 6c) does not result from the contraction of the isotropic one in the ors/#
direction. Nevertheless, high curvature still occurs in the biaxial tension range of the yield
surface predicted with the six-component model.
Figures 7a, 7b, and 7c show the yield stress, R value, and _r' value directionality of
2024-T3 sheet. In this case, the experimental results are compared to predictions made
with the six-component model using a value of m that roughly optimizes the predictions
and with HmL's [1948] anisotropic criterion. HmL'S [1948] criterion is compared with the
six-component yield function because it is by far the most widely used for anisotropic
calculations and because it can describe any type of loading condition, like the six-com-
ponent model proposed in this work. Figure 7a shows the yield stress as a function of

1.1 I I I I I I I I I

I
m=8
I ------ Hill's 1948
A Exp.
1.0
i
'13
i A
~" 0.9

0.8 L I I I I I I I
0 10 20 30 40 50 60 70 80 90
Angle between tenslon axls and rolling dlmctlon
(a)

1.4 I I I I I I I I

1.2

1.0 _~ , ~ A A

o.a ..f a

n" 0.6 ] .~,,,,,,,,

0.4 m=8
- - - - - - Hill's 1948
0.2- A Exp.

0.0 I l I I I I I I
0 10 20 30 40 50 60 70 80 90
Angle between tenelon axle and rolllng dlrectlon
(h)
Fig. 7. (a) Normalized yield stress for 2024-T3 alloy sheet versus tensile direction (values normalized with
respect to 0° yield stress, 327 MPa). Prediction with the six-componentmodel for m = 8 (--); and with Hni's
[1948] criterion (--); experiments (z~). (b) R value for 2024-T3 alloy sheet versus tensile direction. Predic-
tion with the six-componentmodel with m = 8 (--); and with Hn.L's [1948] criterion (---); experiments (a).
(Figure continued on the following page.)
710 F. BARLAT et al.

0.20 I I I I I I I I

0.15

o.10

0.05
L.

0.00

-0.05 ~- ..... 111=8


------ Hill's 1948

-0.1o I I I I I I I I
0 10 20 30 40 50 60 70 80 90
Angle between rolling axis end tension direction
(c)

Fig. 7 continued. (c) F value for 2024-T3 alloy sheet versus tensile direction. Prediction with the six-component
m o d e l w i t h m = 8 ( - - ) ; a n d with HILL'S [1948] criterion (---).

the tension direction. As mentioned previously, the yield stresses for different uniaxial
stress states are used to calculate the constants in the yield function. Therefore, as ex-
pected, the agreement between experiment and theory is quite good in both cases. The
theoretical R values (Fig. 7b), which are fully predicted, are in reasonable agreement with
the experiment. The curve calculated with H~L'S [1948] criterion tends to fit the exper-
imental data better than the curve calculated with m = 8. The predicted/" value direc-
tionalities (Fig. 7c) predicted by the two models are nearly identical. For this 2024-T3
sheet sample, the six-component yield function does not seem to result in better predic-
tions than those made with HrLL'S [1948] yield criterion. However, the overall descrip-
tion of this material is more accurate with the proposed six-component yield function
because the relatively high value of m leads to some rounded vertices near the uniaxial
and biaxial range as polycrystal models would predict. For instance, this could result
in better forming limit predictions.
This criterion can be very useful when implemented into FEM codes because it can
account for anisotropy and, more specifically, for crystallographic textures in models
of numerous forming processes. CaUNG and SaAa ]1990] have used a FEM code in con-
junction with this yield function to model the hydraulic bulge test and the cup draw-
ing test for the 2008-T4 alloy sheet. They have obtained results in excellent agreement
with experiments.
It appears that the next step required to improve the predictive capabilities of the phe-
nomenological model is the relaxation of the isotropic hardening assumption. Figure 8
illustrates this point for the extreme case o f a material initially described by the Tresca
criterion (thin full line). The R value is 1 in any tensile direction for this material. How-
ever, if after some thermomechanical treatment the yield function is translated (thick
full line) due to a Bauschinger effect, the R values for tension in the rolling and in the
transverse directions are oo and 0, respectively. However, if the yield surface occupies
a somewhat different position (dashed line), the R values are transposed: 0 in the roll-
ing and ~ in the transverse direction! Of course, this is an extreme example because the
yield surface has sharp vertices. However, this schematic model may explain why yield
Six-component yield function for anisotropic materials 711

1.5 I I I
Oyy
t-
O
0
1.0

"0
4)

0
>
p,
0.5 J
(~xx

j
.--
0.0

"0
-0.5
>-

-1.0 I i
-1.0 -0.5 0.0 0.5 1.0 1.5
Yield stress in rolling direction

Fig. 8. Schematic anisotropic work-hardening.

stresses are quite well described by p h e n o m e n o l o g i c a l or even polycrystal models, whereas


R values a r e not.

VI. CONCLUSIONS

In this w o r k , a new a n i s o t r o p i c yield function for o r t h o t r o p i c materials has been p r o -


p o s e d . This f u n c t i o n reduces to a very g o o d d e s c r i p t i o n o f the yield s u r f a c e c a l c u l a t e d
with p o l y c r y s t a l m o d e l s for i s o t r o p i c materials. T h e r e f o r e , it is believed to be a suitable
m a t h e m a t i c a l expression for a n i s o t r o p i c polycrystals. This f u n c t i o n was used to predict
the uniaxial plastic p r o p e r t i e s for a l u m i n u m alloy sheets, a n d r e a s o n a b l e agreement was
f o u n d between theoretical a n d experimental results. T h e p r o p o s e d yield function can be
a p p l i e d to a n y t y p e o f l o a d i n g c o n d i t i o n a n d can, t h e r e f o r e , be a d v a n t a g e o u s l y substi-
t u t e d for HILL's [1948] f o r m u l a t i o n , which is the six-component general anisotropic yield
f u n c t i o n c u r r e n t l y used in m o s t F E M codes. A s a n o u t c o m e , there are m a n y p o t e n t i a l
a p p l i c a t i o n s o f the new yield f u n c t i o n , p a r t i c u l a r l y in the s i m u l a t i o n o f a n y processes
i n v o l v i n g plastic d e f o r m a t i o n o f a n i s o t r o p i c m a t e r i a l s .

Acknowledgement - T h e authors would like to thank the following people from Alcoa Laboratories: Drs.
R. C. Becker and K. Chung for valuable comments on the manuscript, Dr. S. Panchanadeeswaran for com-
putations related to crystallographic texture analysis, and Dr. O. Richmond for his support to this work.

REFERENCES
1938 TAYLOR,G.I., "Plastic Strain in Metals," J. Inst. Metals, 62, 307.
1948 HILL,R., "A Theory of the Yielding and Plastic Flow of Anisotropic Metals," Proc. Royal Soc. Lon-
don, A193, 281.
1951 Bmnop,J.W.F. and HILL,R., "A Theory of the Plastic Distortion of a PolycrystaUineAggregate under
Combined Stresses," Phil. Mag., 42, 414 and "A Theoretical Derivation of the Plastic Properties of
a Polycrystalline Face-Centered Metal," Phil. Mag., 42, 1298.
1951 DRUCKER,D.C., "A More Fundamental Approach to Plastic Stress-Strain Relations," in Proc. First
U.S. Nat. Congr. Applied Mechanics, ASME, New York, pp. 487-491.
712 F. BARLATet al.

1954 HERSHEY,A.V., "The Plasticity of an Isotropic Aggregate of Anisotropic Face Centered Cubic Crys-
tals," J. Appl. Mech. Trans. ASME, 21, 241.
1958 ECmLESTON,H.G., "General Properties of Convex Functions," in Convexity (Ch. III), University Press,
Cambridge, pp. 45-58.
1959 DRUCKER,D.C., "A Definition of a Stable Inelastic Material," J. Appl. Mech. Trans. ASME, 26, 101.
1965 BUNGE,H.J., "Zur DarsteUung Allgemeiner Texturen," Z. Metallkunde, 56, 872.
1965 ROE,R.J., "Description of Crystallite Orientation in Polycrystalline Materials Ill: General Solution
to Pole Figure Inversion," J. Appl. Physics, 36, 2024.
1970 LIPPMANN,H., "Matrixungleichungen und die Konvexit~it der Fliessfliiche," Zeit. Angew. Math.
Mech., 50, 134.
1972 HOSFORD,W.F., "A Generalized Isotropic Yield Criterion," J. Appl. Mech. Trans. ASME, E39, 607.
1972 ROCKAFELLAR,R.T., "Convex Functions," in Convex Analysis (Section 4), Princeton University Press,
New Jersey, pp. 23-31.
1976 FRANCOIS,D., "Limite d'Elasticit~ des M(~taux non Allies et Crit~res de Plasticit6," in Mise en Forme
des M&aux et Alliages, Editions du CNRS, Paris, pp. 49-57.
1976 HECKER,S.S., "Experimental Studies of Yield Phenomena in Biaxially Loaded Metals," in STRICK-
LIN, J.A. and SACZALSKI,K.J. (eds.), Constitutive Equations in Viscoplasticity: Computational and
Engineering Aspects, ASME, New York, pp. 1-33.
1977 BASSANI,J.L., "Yield Characterization of Metals with Transversely Isotropic Plastic Properties," Int.
J. Mech. Sci., 19, 651.
1977 GOTOH,M., "A Theory of Plastic Anisotropy Based on a Yield Function of Fourth Order (Plane Stress
State)," Int. J. Mech. Sci., 19, 505.
1979 HILL,R., "Theoretical Plasticity of Textured Aggregates," Math. Proc. Cambridge Philos. Soc., 85,
179.
1980 LOGAN,R.W. and HOSVORD, W.F., "Upper-Bound Anisotropic Yield Locus Calculations Assuming
(111)-Pencil Glide," Int. J. Mech. Sci., 22, 419.
1981 ZYCZKOWSKI, M., Combined Loadings in the Theory of Plasticity, Polish Scientific Publishers, War-
saw, Poland.
1982 SCHMITT,J.H. and JALINIER, J.M., "Damage in Sheet Metal F o r m i n g - I . Physical Behavior," Acta
Met., 30, 1789.
1984 BUDIANSKI,B., "Anisotropic Plasticity of Plane-lsotropic Sheets," in DvOlOa¢, G.J. and SHrELD, R.T.
(eds.), Mechanics of Material Behavior, Elsevier, Amsterdam, pp. 15-29.
1987 BARLAT,F., "Crystallographic Textures, Anisotropic Yield Surfaces and Forming Limits of Sheet Met-
als," Mat. Sci. Eng., 91, 55.
1987 BARLAT,F. and RICHMOND,O., "Prediction of Tricomponent Plane Stress Yield Surfaces and Asso-
ciated Flow and Failure Behavior of Strongly Textured FCC Polycrystalline Sheets," Mat. Sci. Eng.,
95, 15.
1987 ZHO, Y., DODD, B., CADDEtL, R.M. and HosroRo, W.F., "Convexity Restrictions on Non-Quadratic
Anisotropic Yield Criteria," Int. J. Mech. Sci., 29, 733.
1988 BARLAT,F. and FmCKE, W.G., JR., "Prediction of Yield Surfaces, Forming Limits and Necking Di-
rections for Textured FCC Sheets," in KALLEND, J.S. and GOTTSTEIN, G. (eds.), Proc. 8th Int. Conf.
Texture of Materials, Santa Fe, NM, 1987, The Metallurgical Society, pp. 1043-1050.
1988 BARL~, F., LtA~, J. and BAUDELET,B., "A Yield Function for Orthotropic Sheets under Plane Stress
Conditions," in Proc. 8th Int. Conf. Strength of Metals and Alloys, Tampere, Finland, 1988 August,
Vol. 1, Pergamon Press, New York, pp. 283-288.
1988 BmLINGTOr~,E.W., "Generalized Isotropic Yield Criterion for Incompressible Materials," Acta Mechan-
ica, 72, 1.
1988 BUNGE,H.J., "Calculation and Representation of Complete ODF," in KALLEND, J.S. and GOTTSaTEIN,
G. (eds.), Proc. 8th Int. Conf. Texture of Materials, The Metallurgical Society, pp. 69-78.
1989 BARLAT,F. and LIA2q, J., "Plastic Behavior and Stretchability of Sheet Metals. Part !: A Yield Func-
tion for Orthotropic Sheets Under Plane Stress Conditions," Int. J. Plasticity, 5, 51.
1989 LEGE,D.J., BARLAT,F. and BREM, J.C., "Characterization and Modeling of the Mechanical Behav-
ior and Formability of a 2008-T4 Sheet Sample," Int. J. Mech. Sci., 31, 549.
1989 LIAN,J., ZHOO, D. and BAUDELET,B., "Application of Hill's New Theory to Sheet Metal Forming -
Part I. Hill's 1979 Criterion and its Application to Predicting Sheet Forming Limit," Int. J. Mech.
Sci., 31,237.
1990 CHUNG,K. and SHAH, K.N., "Finite Element Simulation of Sheet Metal Forming for Planar Aniso-
tropic Metals," accepted for publication in Int. J. Plasticity.
1990 H~L, R., "Constitutive Modelling of Orthotropic Plasticity in Sheet Metals," J. Mech. Phys. Solids,
38, 405.

Alcoa Laboratories
Alcoa Center, PA 15069, USA.
(Received 26 March 1990; in final revised form 28 July 1990)

You might also like