A constitutive model for casting magnesium alloy ZL101 based

on the analysis of spherical void evolution

B. Chen
a,
*
, X.H. Peng
a
, J.H. Fan
a
, J.G. Wang
b
, S.T. Sun
a
a
Department of Engineering Mechanics, College of Resource and Environment Science, Chongqing University, Chongqing 400044, China
b
Center for Protective Technology, National University of Singapore, Singapore 117543, Singapore
Available online 4 March 2008
Abstract
Since casting magnesium alloys contain numerous spherical microvoids, the aggregate of microvoids and matrix can be analyzed using
a representative volume element. The representative element can be idealized as a cell containing a spherical void. Through the analysis
on the velocity eld of the spherical void-cell model, the strain eld of the spherical void-cell model was obtained. Dening an intrinsic
time that involves the hardening due to plastic deformation and the softening due to voids, a new endochronic model was derived for the
elastoplastic and damage behavior of casting magnesium alloys. The corresponding numerical algorithm and nite element procedure
were developed and applied to the analysis of the elastoplastic response and the porosity of casting magnesium alloy ZL101. The com-
puted results show satisfactory agreement with experiments.
2008 Elsevier B.V. All rights reserved.
PACS: 61.66.Dk; 61.72.Qq; 64.30.Ef; 62.20.F-; 81.40.Jj; 07.05.Tp
Keywords: Casting magnesium alloy; Spherical void; Void volume fraction; Elastoplasticity; Constitutive model
1. Introduction
Due to being lightweight, easily shaped and recycled,
casting magnesium alloys have been preferred choices for
lightweight constructions of components for the automo-
tive industry, for example, steering wheels, door structures
and oil sumps [1]. However, because casting alloys contract
when solidied, they inevitably contain a certain amount of
microscopic voids. As the microscopic voids grow under
loading, the walls or ligaments between voids thin down
and ductile fracture may occur due to coalescence of voids
[2]. Constitutive theories for such kind of material should,
therefore, take into account the eect and evolution of
microvoids. Gurson [3] proposed a phenomenological rela-
tionship for progressively cavitating solids based on a thick
walled spherical shell using approximate limit analysis.
Tvergaard and Needleman [4] extended the Gurson model
to account for the nal failure of materials at a realistic
value of the void volume fraction. Fan et al. [5] put forward
a formulation of constitutive relations to describe the
behavior of void-damaged materials based on a nite ele-
ment analysis. Siruguet and Leblond [6] modeled the eect
of inclusions on void growth in a porous ductile metal.
Baaser and Gross [7] analyzed the growth of microvoids
in a crack tip of a ductile material loaded by a remote K
I
eld. Horstemeyer et al. [8] investigated the internal state
variable rate equations in continuum framework to model
void nucleation, growth, and coalescence in a cast AlSi
Mg aluminum alloy. In this paper, assuming a casting mag-
nesium alloy is an elastoplastic material containing numer-
ous spherical microvoids, such a material was analyzed
based on its representative volume element (RVE). The
RVE can be idealized as a spherical void-cell model.
Through the analysis of the velocity and strain elds of
the spherical void-cell model, an endochronic constitutive
equation including void evolution was obtained. The corre-
sponding nite element procedure based on the constitutive
0927-0256/$ - see front matter 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.commatsci.2008.01.015
*
Corresponding author.
E-mail address: bchen@cqu.edu.cn (B. Chen).
www.elsevier.com/locate/commatsci
Available online at www.sciencedirect.com
Computational Materials Science 44 (2008) 190194
equation was developed and applied to analyze the consti-
tutive behavior and the porosity of casting magnesium
alloy ZL101 subjected to elastoplastic deformation. Com-
parison between the computed results and experimental
data show satisfactory agreement.
2. Constitutive description for casting magnesium alloy and
void evolution equation
The general physical model used here is a representative
macroscopic volumetric element (RVE) of a casting mag-
nesium alloy material (an aggregate of microvoids and duc-
tile matrix) (Fig. 1). The representative volumetric element
is by denition large enough to be statistically representa-
tive of the properties of the aggregate. The RVE can be
simplied as a spherical void-cell model shown in Fig. 2.
The radii of the void and the cell are a and b, respectively.
The radius of an arbitrary point in the matrix of void-cell is
r. The matrix of the model is assumed homogeneous and
incompressible. Suppose there is a microscopic velocity
eld v
r
, v
u
and v
h
in the matrix of the void-cell model under
spherical coordinates, the corresponding microscopic
strain rate eld _ e
ij
in the matrix can be expressed as
_ e
r

om
r
or
; _ e
h

1
r
om
h
oh

m
r
r
; 1a; b
_ e
u

1
r sin h
om
u
ou

m
h
r
ctgh
v
r
r
; _ e
rh

om
h
or

m
h
r

1
r
om
r
oh
:
1c; d
In the case of axisymmetrical deformation
v
u
0; _ e
ru
_ e
hu
0; 2a; b
where v
r
and v
h
are the function of the r and h. Noticing
that the matrix of the void cell is incompressible, one
obtains
_ e
r
_ e
h
_ e
u

o
or
r
2
v
r
sin h
o
oh
rv
h
sin h 0: 3
Suppose the boundary condition at the outer surface of
the spherical void cell can be expressed as
v
i

_
E
ij
x
j
; r b; 4
where
_
E
ij
is the macroscopic strain rate of the matrix. In the
case that the global axis 3 coincides with the local spherical
axis and the deformation is axisymmetrical
_
E
11

_
E
22
;
_
E
ij
0; ij: 5
One can obtain the boundary velocity eld in a spherical
coordinate system
v
r
b
_
E
11
sin
2
h
_
E
33
cos
2
h;
v
h
b
_
E
11

_
E
33
sin h sin u;
v
u
0:
6
Combining Eqs. (1) with (6) one can obtain
_ e
r

2A
0
r
3

_
E
e
6
1 3 cos 2h; _ e
h

A
0
r
3

_
E
e
6
1 3 cos 2h;
7a; b
_ e
u

A
0
r
3

1
3
_
E
e
; _ e
rh

_
E
e
sin2h; A
0

b
3
3
2
_
E
11

_
E
33
b
2
_
E
m
;
7ce
where
_
E
e
and
_
E
m
are, respectively, the macroscopic equiva-
lent and bulk strain of the matrix,
_
E
e

_
E
33

_
E
11
;
_
E
m

_
E
11

_
E
22

_
E
33
=3: 8a; b
The microscopic intrinsic time measure is denoted by f,
which is dened as the Euclidean norm of the deviatoric
increment of microscopic strain [9]. Due to the assumption
of incompressibility of the matrix, one has
df de
0
ij
de
0
ij

1=2
de
ij
de
ij

1=2
; 9
where de
0
ij
is the increment of microscopic deviatoric strain.
The intrinsic time scale can be dened as
dz df=F fgf : 10
F(f) is the hardening function which reects the harden-
ing of the material subjected to plastic deformation, g(f) is
the softening function which reects the softening behavior
of material due to void growth during plastic deformation,
f is the current void volume fraction of material. For sim-
plicity, to F(f) and g (f) are given the following simple
forms without considering the strain rate eect
F f 1 b
1
f
c
1
; gf 1 b
2
f
c
2
; 11a; b
p
Solid(Vs)
Vs =Solid Volume
Void(Vv)
VVv=Vs
p
p
Fig. 1. Material element with voids.
2
1
3

a
r
b
Fig. 2. A spherical void-cell model.
B. Chen et al. / Computational Materials Science 44 (2008) 190194 191
where b
1
, b
2
, c
1
and c
2
are material constants, which can be
determined from the curve of a uniaxial experiment.
Substituting Eq. (10) into the incremental form of endo-
chronic constitutive equation [10], which can describe the
constitutive behavior of materials without a yield surface
with favorable precision, yields the following incremental
constitutive equation
Ds
r
ij
s
r
ij
s
r
ij
z
n

k
r
X
3
r1
C
r
De
p
a
r
s
r
ij
z
n
Df=F fgf ; 12
where
De
p
ij
de
ij
ds
ij
=2G; 13a
k
r

1 expa
r
Dz
a
r
Dz
: 13b
s
ij
denotes the microscopic deviatoric stress, z
n
is the
intrinsic time scale after n th increments of loading and
s
r
ij
z
n
represents the r th component of s
ij
at z
n
, C
r
and a
r
(r = 1, 2, 3) are material constants, and G is the elastic
shear modulus. By setting
A
X
3
r1
k
r
C
r
; B
ij

X
3
r1
k
r
a
r
s
r
ij
z
n
; 14
one can derive the following expression of the incremental
form of the endochronic constitutive equation involving
voids:
Ds
ij
ADe
p
ij
B
ij
Dz: 15
Keeping Eq. (13a) in mind Eq. (12) can be re-written as
Ds
ij
2G
p
De
ij
T
p
B
ij
Dz; 16
where
T
p

2G
2G A
; 2G
p
AT
p
: 17
If the eect of voids is not considered (f = 0, g(f) = 1),
one can prove that Eqs. (12) and (15) reduce to the incre-
mental form of the constitutive equation given by Peng
and Fan [10]. Moreover, the Chaboche s constitutive law
for back stress can also be obtained as a special case when
g(f) = 1 and F(f) is constant.
Assuming homogeneous matrix, the homogenization
principle can be used in the transition between microscopic
and macroscopic quantities [3]. Letting U and / be the mac-
roscopic and the microscopic potential functions, respec-
tively, the macroscopic stresses can be expressed as [3]
R
ij

oU
oE
ij

1
V
Z
V m
o/
oE
ij
dV
1
V
Z
V m
s
kl
oe
kl
oE
ij
dV : 18
Substituting Eqs. (7) and (16) into Eq. (18), the macro-
scopic constitutive equation of casting magnesium alloy
can be obtained.
The void volume fraction may change during deforma-
tion, which is contributed by both the growth of existing
voids and the nucleation of new voids, i.e.,
_
f
_
f
growth

_
f
nucleation
: 19
Keeping in mind that the matrix is incompressible, the
increment of the void volume fraction due to the growth
of void can be given by
_
f
growth

_
V
v
V
1 f _ e
kk
: 20
The new voids, nucleated either by cracking of the par-
ticles or by decohesion of the particlematrix interface, can
be described with
_
f
nucleation
K
_
f; 21
where K follows a normal distribution with mean value f
N
and standard deviation s
N
. Similarly to the formula by Chu
and Needleman [11], K is given as
K
f
N
s
N

2p
p exp
1
2
f f
N
s
N

2
" #
: 22
where s
N
= 0.1, f
N
= 0.4, f
N
= 0.2. The addition of the void
nucleation term would more fully reect the eect of the
void evolution on material behavior and improve the pre-
dictive ability of the constitutive relation.
The material parameters C
r
, a
r
in the constitutive equa-
tion Eq. (15) control the stress response and can be deter-
mined by a simple tensile test [5]. Taking the plastic
incompressibility condition, the tensile stress can be
expressed in terms of C
i
, a
i
as
r

3=2
p X
3
r1
C
r
a
r
1 exp

3=2
p
a
r
e
p
: 23
Given a set of measured r
i
; e
p
i
, we can determine the
parameters C
r
and a
r
with the least squares approach.
3. Application and verication
The corresponding numerical algorithm and FE
approach were developed based on the constitutive equa-
tion presented. The proposed constitutive equation and
the corresponding nite element procedure were applied
to the analysis of the relationship between stress and strain
and the porosity of the casting magnesium alloy ZL101
specimens (with and without notch) under elastoplastic
deformation. The material parameters were identied as
C
1;2;3
7:853 10
4
; 5:463 10
3
; 1:326 10
3
MPa;
a
1;2;3
1527; 278:4; 13:8; 24a
b
1
1:03; b
2
1:1; c
1
1:05; c
2
9:8: 24b
The geometry of the specimens without a notch is
200 mm in length, and 10 mm in diameter. The size of
the specimen with a notch is shown in Fig. 3. The upper
right quarter of the two kinds of cylindrical specimens were
taken for the analysis due to the symmetry of the problems.
The eight-node isoparametric element with 2 2 Gaussian
points was adopted. The axial displacements were imposed
192 B. Chen et al. / Computational Materials Science 44 (2008) 190194
at the end of the specimens with the incremental step of
0.02 mm. The computed results were veried by both mac-
roscopic experiment and microscopic observation.
For the specimen without a notch, the relationship
between stress and strain was investigated. Fig. 4 shows
that the computed stressstrain relationship agrees satisfac-
torily with the experimental data. From Fig. 4 it can also
be found that the constitutive equation can properly reect
the main character of the material stressstrain curve with-
out a distinct yield point. In order to verify the relationship
between porosity and deformation, specimens without a
notch were also tested on an Instron 1342 servo-hydraulic
material testing system. They were subjected to dierent
deformations, and then unloaded. The tested specimens
were cut along the radius at minimum section. After polish-
ing these sections, the changes of porosity in the center
parts of the specimens were carefully observed with a
microscopic photointerpreter. Since the void distribution
is stochastic, a quantitative metallographic method was
adopted. Fig. 5 gives the comparison of the change of
porosity during deformation, which shows satisfactory
agreement between the computed and experimental results.
Fig. 5 also shows that the porosity changes with the devel-
opments of plastic deformations according to exponential
law. Fig. 6 is the SEM photograph at the fractured section
of the tested casting specimen, where many dimples can be
seen in the fracture section. These dimples are the roots of
voids and reect ductile fracture of the material. For the
specimen with a notch the distributions of the stress and
porosity on the smallest section of the specimen were stud-
ied. Fig. 7 shows the contours of the void volume fraction
at dierent applied strain. It can be seen from Fig. 7 that
0
50
100
150
200
250
0 5 10 15 20
/%

/
M
P
a
Experimental
Computed
Fig. 4. Experimental and computed re curve.
0
0.5
1
1.5
2
2.5
3
3.5
4
0 0.2 0.4 0.6 0.8

p
/%
Experimental
f
/
%
Computed
Fig. 5. Relation between void volume fraction f and e
p
.
Fig. 6. Scanning electron micrograph of the fractured section of ZL101.
100
200
R
1
2

1
0
1
Fig. 3. Cylindrical tensile specimen with notch.
0.03 0.02
0.035 0.03
0.02
Fig. 7. Contours of void volume fraction (a) e
a
= 0.012, and (b)
e
a
= 0.025.
B. Chen et al. / Computational Materials Science 44 (2008) 190194 193
voids rstly occur at the root of the notch, where strain is
relatively larger, and the porosity takes its maximum in the
region near to the root of the notch and decreases with the
increase of the distance away from the notch root. Figs. 8
and 9 show the distributions of the axial stress and the
porosity along the notch line at 30 kN, where a is the dis-
tance from the surface, coordinate 0 denoting the surface
and 5 mm denoting the center of the smallest cross section
of the specimen. Fig. 8 shows that maximum axial stress
arises at the notch root of the specimen, but maximum
stress considering voids is lower than that without consid-
ering voids. Fig. 9 shows that maximum porosity also
appears at the notch root of the specimen and agrees well
with experimental results.
4. Conclusions
Casting magnesium alloys contain numerous spherical
microvoids, the representative void-matrix cell was isolated
and idealized as a single spherical cell containing a spheri-
cal void. The void volume fraction of the void-matrix cell
model equals that of the material. Through the analysis
of the velocity and strain of the cell, an elastoplastic consti-
tutive description for casting magnesium alloys was
obtained. The change of the void volume fraction during
deformation was considered as a combination of the
growth of existing voids and the nucleation of new voids.
The eect of void evolution on the constitutive equation
was considered using a softening function. The correspond-
ing FE procedure was developed and applied to analyze the
constitutive behavior and the evolution of porosity of the
casting magnesium alloy ZL101 specimens, with and with-
out a notch. It was found that, in the specimen without a
notch, the calculated result for the stressstrain relation-
ship for the specimen agrees with experimental data. The
variation of porosity follows plastic deformation and is
also consistent with experimental results conforming to
the exponential function. Study of the notched specimen
shows that the maximal axial stress appears at the notch
root and that the maximal stress value considering voids
is lower than without considering voids. The porosity also
reaches the maximum in the vicinity of the notch root and
decreases toward the center of specimen, which is reason-
able compared with the experimental results.
Acknowledgement
The nancial support from Natural Science Foundation
of China (Grant No. 10572157) is fully acknowledged.
References
[1] G. Eisenmeier, B. Holzwarth, H.V. Hopped, H. Mughrabi, Mater.
Sci. Eng. A 319 (2001) 578582.
[2] D.M. Tracey, Eng. Fract. Mech. 3 (1971) 301316.
[3] A.L. Gurson, J. Eng. Mater. Technol. 99 (1977) 215.
[4] V. Tvergaard, A. Needleman, Acta Metall. 32 (1984) 157169.
[5] J. Fan, J. Huang, X. Zeng, Nucl. Eng. Des. 116 (1989) 307313.
[6] K. Siruguet, J.B. Leblond, Int. J. Plastic. 20 (2004) 225254.
[7] H. Baaser, D. Gross, Comput. Mater. Sci. 26 (2003) 2835.
[8] M.F. Horstemeyer, J. Lathrop, A.M. Gokhale, M. Dighe, Theor.
Appl. Fract. Mech. 33 (2000) 3147.
[9] K.C. Valanis, Arch. Mech. 25 (1971) 517551.
[10] X. Peng, J. Fan, Comput. Struct. 47 (1993) 313320.
[11] C.C. Chu, A. Needleman, J. Eng. Mater. Technol. 102 (1980) 249
256.
0
10
20
30
40
50
60
70
80
0 5 4 3 2 1
a/mm
Considering void
No considering void

z
/
M
P
a
Fig. 8. Distributions of axial stresses along notch line.
0
0.5
1
1.5
2
2.5
3
3.5
f
/
%
Experimental
Calculated
0 5 4 3 2 1
a/mm
Fig. 9. Distribution of void volume fraction along notch line.
194 B. Chen et al. / Computational Materials Science 44 (2008) 190194