Professional Documents
Culture Documents
Search
Collections
Journals
About
Contact us
My IOPscience
Computational micro to macro transitions for shape memory alloy composites using periodic
homogenization
This content has been downloaded from IOPscience. Please scroll down to see the full text.
2015 Smart Mater. Struct. 24 035009
(http://iopscience.iop.org/0964-1726/24/3/035009)
View the table of contents for this issue, or go to the journal homepage for more
Download details:
IP Address: 203.200.35.21
This content was downloaded on 02/03/2015 at 07:03
doi:10.1088/0964-1726/24/3/035009
In the current manuscript, a homogenization framework is proposed for periodic composites with
shape memory alloy (SMA) constituents under quasi-static thermomechanical conditions. The
methodology is based on the step-by-step periodic homogenization, in which the macroscopic
and the microscopic problems of the composite are solved simultaneously. The implementation
of the framework is examined with numerical examples on SMA composite laminates.
Complexity of the composite nonlinear response and non-proportional stress state in the SMA
appears, even in the case of uniaxial macroscopic boundary conditions. Moreover, under certain
conditions, the composite laminate can exhibit a non-convex transformation surface.
Additionally, the transformation temperatures at various stress levels under isobaric thermal
cycling can be quite different between the composite and the pure SMA.
Keywords: periodic homogenization, shape memory alloys, pseudoelasticity, thermal cycling,
plasticity
(Some gures may appear in colour only in the online journal)
1. Introduction
0964-1726/15/035009+17$33.00
G Chatzigeorgiou et al
microscale.
G Chatzigeorgiou et al
2.2. Macroscale
2.1. Microscale
( x ) :=
div ( x ) = 0
in ,
(6)
(1)
(7)
where D and D are the symmetric fourth-order macrotangent stiffness tensor and the symmetric second-order
macro-tangent thermal modulus tensor respectively.
Finally, the increment of the macro-energy density
function W is written as
(2)
W ( x ) = (x ): (x ).
(8)
( x , x) = D ( x , x): ( x , x)
+ D ( x , x) ( x ),
in ,
= D ( x ): ( x ) + D ( x ) ( x ),
in ,
(5)
{}
1
t
gradu ( x ) + gradu ( x ) in ,
1
2
[ grad u ( x , x)
t
+ [ grad u ( x , x) ]
(4)
(3)
where D is the symmetric fourth-order micro-tangent stiffness tensor, D is the symmetric second-order micro-tangent
thermal modulus tensor and is the macroscopic temperature. Such a relation implies that the temperature inside the
RVE is assumed to be constant when considering the
microscale stress-strain constitutive relations (see Ene 1983,
Maghous and Creus 2003).
( )
(x ) = V x , x dV
1
= 2V
u ( x , x) n ( x)
+ n ( x) u ( x , x) dS ,
(9)
G Chatzigeorgiou et al
( x ) = V ( x , x) dV
1
= V t ( x , x) x dS ,
(10)
(11)
(12)
(13)
At time step n everything is considered to be known. Considering known temporal and spatial macroscopic temperature
distribution throughout the composite, at time step n + 1 is
also known. The task of the macroscale problem is to evaluate
the macroscopic strains at time n + 1.
By introducing the space of the macroscopic test functions
[ grad z ( x , x)
t
+ [ grad z ( x , x) ] .
( x , x) = ( x ) +
W ( x , x) dV = W ( x ).
= : , grad 2 ( ), = 0 on EB ,
(14)
(15)
grad : dV NB t dS = 0,
(16)
(17)
where m denotes the iteration step. In the preceding expression the temperature does not appear, since it is known at time
n + 1 and thus it has no increment during the Newton
Raphson scheme. If the macroscopic stress and the macrotangent modulus D are known from the solution of the
microscale problem, then equation (17) can be solved for u
and thus the macro-strains at time step n + 1 can be
obtained. A detailed algorithmic scheme for the macroscale
problem is provided in table A1 of appendix A.
4
G Chatzigeorgiou et al
periodic on },
dV = 0, ,
(18)
grad : dV = 0,
(19)
z = : + .
grad : [ D : grad + D ] dV = 0,
(20)
grad : D : grad + D dV = 0,
(21)
(26)
1. As a rst step the macro-strain and the macrotemperature are provided exclusively by the macroscale
analysis, which means that the terms and in
equation (21) are zero. Thus
(m*)
+ D (m*) : grad z(m*) dV = 0,
(25)
(24)
*
+ D (m*) (m ) dV = 0,
(23)
D : grad z + D : + D
grad :
= : , grad 2 (),
dV = V1 D : + D dV
1
= V D : grad z + D : + D dV
= V
grad :
[ D : grad + D ] dV :
1
+ V D : grad + D dV .
(22)
=V
(27)
D = V [ D : grad + D ] dV ,
1
D = V D : grad + D dV .
(28)
G Chatzigeorgiou et al
2
vM
= tr
,
tr
tr = ,
> 0,
(33)
< 0,
3
:
2
tr
th
tr
= + + .
H cur
H min,
vM
crit
,
= min sat
vM crit
+ H H min 1 ek ,
H
vM > crit .
(29)
= : +
= S ( ): .
(30)
S = S M S A,
(31)
1
n1
n2
2 a1 [ 1 + [1 ] ] + a 3, > 0,
f=
1 a 1 + n3 [1 ]n 4 a , < 0.
[
] 3
2 2
where the superscripts M and A denote martensite and austenite respectively. For isotropic SMAs the compliance tensor
is a function of the Young moduli EA, EM and the Poisson
ratios A , M . The thermal strain is given by
th = [ 0 ] ,
1
: S: + s0 u 0 f ,
2
(36)
(35)
(34)
(37)
(32)
G Chatzigeorgiou et al
KuhnTucker conditions,
0,
0,
forward = Y 0,
reverse = Y 0,
forward = 0,
reverse = 0,
(38)
Property
Young modulus (GPa)
Poisson ratio
Ms (K)
Mf (K)
As (K)
Af (K)
n1,n2
n3,n4
Phase diagram slope (MPa/K)
H min = H sat
Austenite
Martensite
55
0.33
46
0.33
245
230
270
280
1
1
7.4
7.4
0.056
G Chatzigeorgiou et al
various SMA volume fractions and RVE response at specic macrostrain conditions.
G Chatzigeorgiou et al
Figure 7. Shear versus normal macro-stress at (a) start of forward transformation, (a) nish of forward transformation, (c) start of reverse
G Chatzigeorgiou et al
Figure 11. Laminate composite with 80% SMA volume fraction and
n = 5 (5 SMA layers and 5 metallic material layers). Finite element
mesh in ABAQUS. The normal displacements at the bottom, left,
and behind surfaces are constrained (roller support), while the upper
surface is subjected to traction.
loading-unloading conditions.
G Chatzigeorgiou et al
et al 2012).
EA (GPa)
90
H min
0
Ms (K)
308
n1
0.6
CA (MPa/K)
16
EM (GPa)
63
H sat
0.016
Mf (K)
242
n2
0.2
CM (MPa/K)
10
A
0.3
k (1/MPa)
0.0075
As (K)
288
n3
0.2
(1/K)
1.0E-5
M
0.3
crit (MPa)
12
Af (K)
342
n4
0.3
0 (K)
400
E (GPa)
3.85
0.4
(1/K)
4.4E-5
0 (K)
400
For consistency reasons, it always used the second SMA layer starting
from the bottom.
11
G Chatzigeorgiou et al
Acknowledgments
G Chatzigeorgiou et al
d (k ) dz(k )
Dnn
+ n(k ) = 0 ,
dx1
dx1
NB
(B.1)
with
t (n + 1) dS = 0.
(k )
Dnn
D (k ) D (k ) D (k )
1121
1131
1111
(k )
(k )
(k ) and (k ) =
= D2111
D2121 D2131
n
(k )
(k )
(k )
D 3111 D 3121 D 3131
(k )
11
(k ) . (B.2)
21
(k )
31
dz (k)
dx1
(k )
Dnn
+ n(k ) = m ,
or
(k )
(k ) 1
= Dnn
m n ,
dz (k)
dx1
(B.3)
(k )
(k ) 1
(k )
z(k ) = Dnn
m n x1 + e ,
(B.4)
(N )
e(N ) = Dnn
e(1) = 0,
m n(N ) .
(B.5)
( )
( )
k = 1, 2, , N 1.
(B.6)
nn
(2) (1)
(2)
m n c + e ,
D (2) 1 m (2) c(1) + c(2) + e(2)
n
nn
(3)
= Dnn
RVE problem
(4)
= Dnn
13
G Chatzigeorgiou et al
1.
2.
3.
grad : (n+1)(m+1)(m*) dV = 0.
4.
5.
6.
7.
1. Using the tangent moduli at each point of the RVE, known from
the last solution of the RVE problem algorithm,
compute the correctors and from the equations
D (k ) 1 m (k ) c(r ) + e(k )
n
nn
r=1
grad : D : grad + D dV = 0.
grad : [ D : grad + D ] dV = 0,
D (k+1) 1
nn
1
D = V D : grad + D dV .
r=1
(k +2)
= Dnn
k+1
D (N 1) 1 m (N 1)
nn
n
=
D (N ) 1
nn
m n(N )
N 1
c(r ) + e(N 1)
r=1
N 1
(r )
(N ) 1
Dnn
r=1
m n(N ) .
Dnn (r )
r=1
or
14
m n(r ) c(r ) = 0 ,
N
1
(r ) 1
m = c(r ) Dnn
r = 1
G Chatzigeorgiou et al
(r )
(r ) 1
c(r ) Dnn
. (B.7)
n
r = 1
dz1(k)
dx1
(k )
21
=
kl(k ) = 0
dz 2(k)
1
2 dx1
(k )
31
=
dz 3(k)
1
2 dx1
d n (k)
dx1
d t (k)
dx1
,
(B.8)
D2211
D2221
D2231
= D 3311 D 3321
D 3331
,
D2311
D2321
D2331
Dnn
=
D1122
D1133
D1123
D 3122
D 3133
D 3123
d (k )
D
dx1 nn
d n (k)
dx1
d t (k)
dx1
(k )
= 0,
+ Dnn
+ Dnt (k ) = 0 ,
(k )
(k )
121
131
(k )
(k )
,
221
231
(k )
(k )
321 331
(k )
(k )
133
123
(k )
(k )
.
233
223
(k )
(k )
333 323
d n (r )
dx1
r=1
Dnt =
(B.9)
c(r ) Dnn (r )
d t (r )
dx1
r=1
(r )
+ Dnn
=
+ Dnt (r ) = mt
(r ) 1
(r )
= Dnn
D
c(r ) Dnn
nt
r = 1
Dtn =
=
(B.10)
c(r ) Dtn (r )
r=1
N
d n (r )
dx1
+ Dtn (r )
r=1
mn
T
(r ) 1
r = 1
with
(k )
111
(k )
(k )
n = 211
(k )
311
(k )
122
(k )
(k )
t = 222
(k )
322
c(r ) Dnn (r )
N
1
(r ) 1
mn = c(r ) Dnn
,
r = 1
mt Dnt (k ) ,
N
1
(r ) 1
mt = c(r ) Dnn
r = 1
(r ) 1
(r )
c(r ) Dnn
.
D
nt
r = 1
D1111
D1121
D1131
D 3111
D 3121
D 3131
D2222
D2233
D2223
D2322
D2333
D2323
(k )
= Dnn
N
1
1
(
)
r
(
)
r
= c Dnn ,
r = 1
Dnt
(k )
mn Dnn
,
N
1
1
(
)
r
(
)
r
mn = c Dnn
r = 1
(r ) 1
(r )
c(r ) Dnn
Dnn
r = 1
Dtn
(k )
= Dnn
Dtt =
r=1
N
c(r ) Dtn (r )
d t (r )
dx1
+ Dtt (r )
r=1
(B.11)
(B.12)
G Chatzigeorgiou et al
References
16
G Chatzigeorgiou et al
Mercier S and Molinari A 2009 Homogenization of elasticviscoplastic heterogeneous materials: Self-consistent and MoriTanaka schemes Int. J. Plast. 25 102448
Murat F and Tartar L 1997 H-convergence, in Topics in the
mathematical modelling of composite materials (Progress in
Nonlinear Differential Equations and their Applications Vol
31) ed A Cherkaev and R Kohn 2143 Birkhuser
Nemat-Nasser S and Hori M 1999 Micromechanics: Overall
Properties of Heterogeneous Materials 2nd Edition
(Amsterdam: North-Holland)
Nemat-Nasser S, Sui Y, Guo W and Isaacs J 2005 Experimental
characterization and micromechanical modeling of superelastic
response of a porous NiTi shape-memory alloy Journal of the
Mechanics and Physics of Solids 53 232046
Paiva A and Savi M A 2006 An overview of constitutive models for
shape memory alloys Mathematical Problems in Engineering
2006 130
Paiva A, Savi M A, Braga A M B and Pacheco P M C L 2005 A
constitutive model for shape memory alloys considering
tensile-compressive assymetry and plasticity Int. J. Solids
Struct. 42 343957
Panico M and Brinson L C 2008 Computational modeling of porous
shape memory alloys Int. J. Solids Struct. 45 561326
Patoor E, Eberhardt A and Berveiller M 1996 Micromechanical
Modelling of Superelasticity in Shape Memory Alloys Journal
de Physique IV 6 27792
Patoor E, Lagoudas D C, Entchev P B, Brinson L C and Gao X 2006
Shape memory alloys, part i: general properties and modeling
of single crystals Mech. Mater. 38 391429
Pindera M J, Khatam H, Drago A S and Bansal Y 2009
Micromechanics of spatially uniform heterogeneous media: A
critical review and emerging approaches Composites Part B:
Engineering 40 34978
Piotrowski B, ben Zineb T, Patoor E and Eberhardt A
2012 Modeling of niobium precipitates effect on the
Ni 47 Ti 44 Nb 9 Shape Memory Alloy behavior Int. J. Plast. 36
13047
Ponte-Castaeda P and Suquet P 1997 Nonlinear composites
Advances in Applied Mechanics 34 171302
17