Micromechanics

CONTENTS 2
Contents
Contents...................................................................................2
1Elasticity Equations.................................................................2
1.1Geometric equations............................................................................3
1.1.1Displacements...............................................................................3
1.1.2Strains...........................................................................................3
1.2Static Equations...................................................................................6
1.2.1Principal Stress..............................................................................8
1.3Constitutive Relations..........................................................................9
2Hydrostatic and Deviatoric Components..................................13
2.1Hydrostatic and Deviatoric Stresses..................................................13
2.2Hydrostatic and Deviatoric Strains....................................................14
2.3Constitutive Relations........................................................................15
2.3.1Index Notation, Lams Constants...............................................17
2.3.2Tensorial Notation........................................................................19
2.3.3Engineering Notation...................................................................21
3Homogenization.....................................................................25
3.1Eigenstrain.........................................................................................25
3.1.1Inclusions.....................................................................................25
3.1.2Eshelbys Solution........................................................................25
3.2Inhomogeneities................................................................................29
3.2.1Ellipsoidal Inhomogeneities.........................................................29
3.3Effective Elastic Properties................................................................30
3.3.1Averaging....................................................................................30
3.3.2Effective Elastic Constants..........................................................32
3.3.3Voigt and Reuss Approximation...................................................33
3.3.3.1Voigt Approximation..............................................................34
3.3.3.2Reuss Approximation.............................................................34
3.3.4Dilute (Non-Interacting) Defect Distribution................................35
3.3.5Mori-Tanaka Model......................................................................36
3.4Strength homogenization..................................................................39
1 Elasticity Equations
Equations for calculation of displacements, strains and stresses, are
valid if the structure undergoes only small deformations and the material
behaves in a linearly elastic manner.
CONTENTS 3
Fig. 1.1: Diagram of kinematic a static equations
1.1Geometric equations
1.1.1 Displacements
The displacement of the body is described by 3 components (u, v, w) or
(u
1
, u
2
, u
3
), all of them dependent on the position in the Cartesian
coordinate system (x, y, z) or (x
1
, x
2
, x
3
). In matrix notation the
displacements are arranged as follows:
( )
( )
( )
( )

'

3 2 1 3
3 2 1 2
3 2 1 1
, ,
, ,
, ,
x x x u
x x x u
x x x u
x u
(1.01)
while in the index notation the field of displacements can be expressed as
( )
3 , 2 , 1
3 , 2 , 1

j
i
x u
j i

(1.02)
1.1.2 Strains
Strains describe the deformation of the body. At a point the stretching
can be seen as the differential displacement per unit length, the x-
component of strain is then
( )
x
u
x
z y x u
x
x

, ,
lim
0

CONTENTS 4
(1.03)
therefore the strain is a displacement gradient. The distortion of the
material, which can be described as the change in originally right angles,
is the sum of tilts imparted to vertical and horizontal lines (engineering
strain):
y
u
x
v
xy

+ +
2 1 2 1
tan tan
(1.04)
For other displacement gradients
y
,
z
and distortions
yz
,
zx
the same
reasoning can be applied with cyclic change of coordinates xyzx and
displacements uvwu.
The strain is a second order tensor and therefore the components can
be arranged in the matrix:
1
1
1
1
1
1
1
]
1

,
_

,
_

,
_

,
_

,
_

,
_

z
w
y
w
z
v
z
u
x
w
y
w
z
v
y
v
x
v
y
u
x
w
z
u
x
v
y
u
x
u
2
1
2
1
2
1
2
1
2
1
2
1

(1.05)
where, in the tensorial notation, shear strains (distortions) are in form of
the one half of engineering strains. The difference between vectors (first
order tensors) and second order tensors shows up in how they transform
with respect to coordinate rotations.
The index notation provides a compact description of all the
components of three-dimensional states of strain:
( )
i j j i
i
j
j
i
ij
u u
x
u
x
u
, ,
2
1
2
1
+

,
_

(1.06)
where the comma denotes differentiation with respect to the following
spatial variable (partial derivative). This double-subscript index notation
leads naturally to a matrix arrangement of the strain components, in
which the i-j component of the strain becomes the matrix element in the i
th
row and the j
th
column:
CONTENTS 5
1
1
1
1
1
1
1
1
]
1

,
_

,
_

,
_

,
_

,
_

,
_

1
1
1
]
1

3
3
2
3
3
2
3
1
1
3
2
3
3
2
2
2
1
2
2
1
1
3
3
1
1
2
2
1
1
1
33 32 31
23 22 21
13 12 11
2
1
2
1
2
1
2
1
2
1
2
1
x
u
x
u
x
u
x
u
x
u
x
u
x
u
x
u
x
u
x
u
x
u
x
u
x
u
x
u
x
u

(1.07)
Since the strain matrix (tensor) is symmetric, i.e.
ij
=
ji
, there are six
rather than nine independent strains, as might be expected in 3 3
matrix.
Sometimes it is convenient to arrange the strain components in vector,
or rather pseudovector. Strain is actually a 2
nd
order tensor, like stress or
moment of inertia, and has mathematical properties very different than
those of vectors which must be taken into account while transforming or
calculating the norm of strain. The ordering of the elements in the
pseudovector form is arbitrary, but it is conventional to list them in order
1,1; 2,2; 3,3; 2,3; 1,3; 1,2.
Following the rules of matrix multiplication, the strain pseudovector can
also be written in terms of displacement vector. The strain-displacement
equations can be written in the pseudovector-matrix form:
u
(1.08)

'

1
1
1
1
1
1
1
1
]
1

'

w
v
u
x y
x z
y z
z
y
x
xy
xz
yz
z
y
x
0 / /
/ 0 /
/ / 0
/ 0 0
0 / 0
0 0 /

In so-called Voight-Mandel notation the components of strain are arranged

in the pseudovector and the shear components of strain tensor are
multiplied by 2 as follows:

CONTENTS 6
( ) ( ) ( ) ( )
( ) ( ) ( ) ( )
( ) ( ) ( ) ( )

'

1
1
1
1
1
1
1
1
]
1

'

w
v
u
x y
x z
y z
z
y
x
0 / 2 / 2 / 2 / 2
/ 2 / 2 0 / 2 / 2
/ 2 / 2 / 2 / 2 0
/ 0 0
0 / 0
0 0 /
2
2
2
12
13
23
33
22
11

(1.09)
Such arrangement brings simplified manipulation in some operations.
1.2Static Equations
From the force equilibrium on an infinitesimal cube results in following
equations (Cauchys equations):
0
0
0
+

z
z
yz
xz
y
zy y xy
x
zx
yx
x
b
z y x
b
z y x
b
z y x

(1.10)
where b
i
are body forces. These equations can be written in index notation
as
0
,
+
i j ij
b
(1.11)
In pseudovector-matrix form we can write

'

'

'

1
1
1
1
1
1
1
]
1

0
0
0
0 0 0
0 0 0
0 0 0
z
y
x
xy
xz
yz
z
y
x
b
b
b
x y z
x z y
y z x

(1.12)
CONTENTS 7
From the moment equilibrium on the infinitesimal cube we get:
yx xy
xz zx
zy yz

(1.13)
Due to this moment equilibrium the stress tensor
1
1
1
]
1

33 32 31
23 22 21
13 12 11

ij

(1.14)
is symmetric. The element in the i
th
row and the j
th
column of this matrix is
the stress on the i
th
face in the j
th
direction.
Equilibrium of the stress and surface traction on the boundary can be
expressed by Cauchys formula. It puts into equilibrium the external
traction forces with internal stress. The traction t is associated with any
plane with normal n. It is a stress on the surface of the body
A
A

F
t
0
lim
(1.15)
where the externally applied force F comprises of components in direction
of coordinates. Therefore the traction t is completely defined by three
traction vectors associated with coordinate planes, for instance

'

xz
xy
x
x

) (
t
(1.16)
generally for any normal n
z
z
y
y
x
x
n n n
) ( ) ( ) ( ) (
t t t t
n
+ +
(1.17)
which can be written in compact form as
CONTENTS 8
n t
(1.18)
and in index notation as
i ij
n
j
n t
) (
(1.19)
where n
i
is a multiple of the cosine angle between the investigated plane
and coordinate system and the unit vector (it is a projection onto the
coordinate axes).
1.2.1 Principal Stress
To find the stress on the plane where the corresponding traction vector
is perpendicular to it and shear stresses vanish we put
n n
(1.20)
where the stress on the right hand side of the equation is so called
principal stress. The equation can be expressed as:
( ) 0 n I
(1.21)
where I is the identity matrix. A non-trivial solution is obtained if
( ) 0 det I
(1.22)
Calculation of the determinant leads to the following characteristic
equation:
0
3 2
2
1
3
I I I
(1.23)
where I
1
, I
2
, I
3
are the invariants of the stress tensor (their values remain
the same whatever the rotation of the coordinate system is). The first
invariant is
z y x
I + +
1
CONTENTS 9
(1.24)
the second one can be calculated as
1
]
1

+
1
]
1

+
1
]
1

y yx
xy x
z zx
xz x
z zy
yz y
I






det det det
2
(1.25)
And finally the third stress invariant can be expressed as
1
1
1
]
1

z zy zx
yz y yx
xz xy x
I



det
3
(1.26)
The same reasoning is used for calculation of the principal strain.
1.3Constitutive Relations
The previous sections deal only with the kinematics (geometry) and
static equilibrium of the body; however, they do not provide insight on the
role of the material itself. The kinematic equations relate strains to
displacement gradients, and the equilibrium equations relate stress to the
applied tractions on loaded boundaries and also provide the relations
among stress gradients within the material. Six more equations, relating
the stresses to strains are needed, and these are provided by the
materials constitutive relations. In this section isotropic elastic materials
are dealt with.
In the general case of a linear relation between components of the
strain and stress tensors, we might propose a statement of the form:
( )
t
kl kl ijkl ij
L
(1.27)
where L
ijkl
is a 4
th
order tensor and
t
kl
is initial (or eigen) strain. Because
indices kl do not appear in the equation after summation, they are called
dummy indices. Previous equation constitutes a sequence of nine
equations, since each component of
ij
is a linear combination of all the
components of
kl.
For instance
33 2333 12 2312 11 2311 23
... L L L + + +
CONTENTS 10
(1.28)
Based on each of the indices of L
ijkl
taking on values from 1 to 3, we might
expect a total of 81 independent components in L. However, both the
stress tensor and the strain tensor are symmetric (
ij
=
ji
and
ij
=
ji
), we
must also have L
ijkl
= L
ijlk
and
L
ijkl
= L
jikl
. These relations are called minor symmetries. This reduces the
number of L components to 36, as can be seen from a linear relation
between the pseudovector forms of the strain and stress:

'

1
1
1
1
]
1

'

xy
xz
yz
z
y
x
xy
xz
yz
z
y
x
L L L
L L L
L L L

66 62 61
26 22 21
16 12 11

(1.29)
or, using the Mandel notation:

'

1
1
1
1
1
1
1
1
]
1

'

12
13
23
33
22
11
1313 1323 1312 1333 1322 1311
2313 2323 2312 2333 2322 2311
1213 1223 1212 1233 1222 1211
3313 3323 3312 3333 3322 3311
2213 2223 2212 2233 2222 2211
1113 1123 1112 1133 1122 1111
12
13
23
33
22
11
2
2
2
2 2 2 2 2 2
2 2 2 2 2 2
2 2 2 2 2 2
2 2 2
2 2 2
2 2 2
2
2
2

L L L L L L
L L L L L L
L L L L L L
L L L L L L
L L L L L L
L L L L L L
(1.30)
It can be shown that the L matrix in this form is also symmetric. It
therefore contains only 21 independent elements.
If the material exhibits symmetry in its elastic response, the number of
independent elements in the L matrix can be further reduced. In the
simplest case of an isotropic material, having the same stiffness in all
directions, only two elements are independent namely Youngs modulus
(E) and Poissons ratio (). From these so called shear modulus can be
calculated:
( ) +

1 2
E
G
(1.31)
CONTENTS 11
If a body is loaded by the stress
x
, the resulting deformation
x
=
x
/E
and the other normal components of strain are
y
=
z
= -
x
= -
x
/E. In
the general stress-state the other normal strain components are derived
analogically (however, the material must be isotropic):
( )
z y x x
E

1
(1.32)
( )
z y x y
E
+
1
(1.33)
( )
z y x z
E
+
1
(1.34)
In case of isotropic material each shear deformation is proportional to the
corresponding shear stress with the constant of proportionality 1/G:
( )
xy
xy
xy
E G

+

1 2
(1.35)
( )
xz
xz
xz
E G

+

1 2
(1.36)
( )
yz
yz
yz
E G

+

1 2
(1.37)
The six above equations can be written in the matrix form as
( )
( )
( )

'

1
1
1
1
1
1
1
1
]
1

+
+
+

'

xy
xz
yz
z
y
x
xy
xz
yz
z
y
x
E

1 2 0 0 0 0 0
0 1 2 0 0 0 0
0 0 1 2 0 0 0
0 0 0 1
0 0 0 1
0 0 0 1
1
CONTENTS 12
(1.38)
which can be written in compact form as
M
(1.39)
where M is the elastic compliance matrix. By inversion we get the
generalize Hooks law:
( ) L M
1
(1.40)
where
( )( )
1
1
1
1
1
1
1
1
]
1

5 . 0 0 0 0 0 0
0 5 . 0 0 0 0 0
0 0 5 . 0 0 0 0
0 0 0 1
0 0 0 1
0 0 0 1
2 1 1
E
L
(1.41)
is the elastic stiffness matrix of an isotropic material.
The isotropic constitutive law can also be written using index notation
as
kk ij ij ij
E E

1
(1.42)
where the symbol
ij
is the Kroenecker delta, which is defined as

'

j i
j i
ij
if
if
0
1

(1.43)
ELASTICITY EQUATIONS 13
2 Hydrostatic and Deviatoric
Components
2.1Hydrostatic and Deviatoric Stresses
A state of hydrostatic compression is one in which no shear stresses
exist and where all the normal stresses are equal. For this stress state it is
obviously true that

kk m

3
1
3
33 22 11

+ +

(2.01)
This quantity (mean/hydrostatic stress) is one third of invariant I
1
, which is
a reflection of hydrostatic pressure being the same in all directions, not
varying with axis rotations.
The stress tensor is then composed of the hydrostatic part and the
deviatoric part:
ij ij kk ij
s +
3
1
(2.02)
or in matrix notation the hydrostatic stress state is defined as follows:
1
1
1
]
1

m
m
m

0 0
0 0
0 0

(2.03)
and the deviatoric stress state tensor is then
1
1
1
]
1

m
m
m



33 32 31
23 22 21
13 12 11
s
(2.04)

The hydrostatic (volumetric) stress is related to the change of volume
of a material during deformation, while the deviatoric part is responsible
for the induced distortion. This concept is also convenient because the
ELASTICITY EQUATIONS 14
material responds to these stress components in very different ways. For
instance, plastic and viscous behaviour is driven dominantly by distortional
components, with the hydrostatic component causing only elastic
deformation.
The graphical representation of the stress tensor decomposition is
shown in the following figure:
Fig. 2.1: Decomposition of stress in hydrostatic and deviatoric part
2.2Hydrostatic and Deviatoric Strains
In cubical volume element, originally of dimension abc, is subjected to
normal strains in al three directions, the change in the elements volume is
( ) ( ) ( )
( ) ( ) ( )
z y x z y x
z y x
abc
abc c b a
abc
abc c b a
V
V


+ + + + +

+ + +

1 1 1 1
1 1 1
' ' '
(2.05)
where products of strains are neglected. The volumetric strain is therefore
the sum of diagonal elements in the strain tensor (also called trace of the
matrix, or Tr()). In the index notation, this can be written simply
kk V
V
V

(2.06)
Similarly to mean stress
m
the mean strain is calculated as
kk m

3
1
3
33 22 11

+ +

ELASTICITY EQUATIONS 15
(2.07)
and the strain tensor is then also composed of the hydrostatic and
deviatoric part:
ij ij kk ij
e +
3
1
(2.08)
or in matrix notation the hydrostatic strain is defined as follows:
1
1
1
]
1

m
m
m

0 0
0 0
0 0

(2.09)
and the deviatoric part is obtained by subtraction of
m
from the diagonal
terms in strain tensor:
1
1
1
]
1

m
m
m



33 32 31
23 22 21
13 12 11
e
(2.10)
2.3Constitutive Relations
Since
V
is a relative change of volume it must be independent of the
coordinate system (therefore it is invariant). The relation between
volumetric strain and mean stress can be derived as follows:
( ) ( ) ( )
( )
( )
( )
K E E
E E E
m
z y x
z y x
z y x z y x z y x V

+ +

+ +

+ + + +
3
2 1 3 2 1
1 1 1
(2.11)
where
( ) 2 1 3

E
K
ELASTICITY EQUATIONS 16
(2.12)
is so called bulk modulus. The bulk modulus is the stiffness parameter that
connects the mean hydrostatic stress
m
with the volumetric strain
V
.
Note that as 0.5,
K . That is, the material becomes infinitely stiff as Poisons ratio
approaches 0.5. Values of Poissons ratio greater than 0.5 are not possible
since such values imply that a tensile hydrostatic stress would cause a
volumetric contraction. The volumetric change is proportional to the mean
stress only in case of isotropic material.
The normal component of deviatoric deformation e
x
can be expressed
as
( )
( ) ( ) [ ]
( )
z y x
m m z m y m x
m z y x m x x
s s s
E
E
s s s
E
K E
e

+ + +

1
2 1 1
3
1 1
(2.13)
and since
( ) ( ) ( )
0 3 3
3

+ + + + + +
m m
m z y x m z m y m x z y x
s s s


(2.14)
the sum of the deviatoric normal components is equal to zero. Therefore
z y x
s s s
(2.15)
and
( )
G
s
s
E
s s
E
e
x
x x x x
2
1 1

+
+

(2.16)
The same relation holds for other normal components of deviatoric strain.
The shear strain components are (as mentioned before) related to shear
stress by shear modulus G without factor 2. For instance the first shear
component:
ELASTICITY EQUATIONS 17
( )
G E
xy
xy xy

1 2
(2.17)
2.3.1 Index Notation, Lams Constants
For the diagonal terms in the strain tensor, for instance the strain
11
,
the strain-stress relationship can be expressed as
( )
( )
( )
33 22 11 11 33 22 11 11
1 1 1 1

+ +
+
+
E E E E
(2.18)
which can be in index notation for all normal strains expressed as
( )
j i
E E
kk ij ij

+
if

1
(2.19)
For the shear components of the stress tensor, for instance for instance
the strain
12
, the strain-stress relationship can be expressed as
( )
12 12
1

E
+

(2.20)
which can be in index notation for all normal strains expressed as
( )
j i
E
ij ij

+
if

1
(2.21)
The previous equations (for normal and shear strains) can be written in a
single expression by making use of the Kronecker delta. That gives:
( )
ij kk ij ij
E E

1
(2.22)
The required form of stress-strain relationship (dependence of stress on
strains), using so called Lams constants and has a form:
ELASTICITY EQUATIONS 18
kk ij ij ij
+ 2
(2.23)
In order to establish the relationship between Lams constants, Youngs
modulus and Poissons ratio, it is necessary to compare the two forms of
the constitutive equations for an isotropic elastic material with the
previous equation. However, in order to make that comparison, the
equations for strain-stress relationship must be inverted, because the
previous equation expresses the stress components in terms of the strain
components.
By simple arrangement the following equation can be obtained:
( )
kk kk
E

2 1

(2.24)
and since we know that
( ) ( ) ( )
ij kk ij ij kk ij ij
E
E
E

+
+
+

,
_

+
+

1 1 1
(2.25)
where the
kk
can be then substituted as follows:
( ) ( )( )
ij kk ij ij
E E

2 1 1 1 +
+
+

(2.26)
Therefore the Lams constants can be expressed as
( )

+

1 2
E
G
(2.27)
and
( )( )
G K
E
3
2
2 1 1

+

(2.28)
in terms of E, , K and G.
ELASTICITY EQUATIONS 19
2.3.2 Tensorial Notation
In tensorial notation there is needed the use of the unit fourth-order
tensor, I, with components:
jl ik ijkl
I
(2.29)
with Kronecker delta being called the unit second-order tensor. This tensor
exhibits major symmetry but not minor symmetry, and it has the
important property that I : = : I = for any second-order tensor .
Sometimes it is useful to work with the symmetrized unit fourth-order
tensor, I
S
, which has components:
( )
2
jk il jl ik
S
ijkl
I
+

(2.30)
This tensor exhibits minor and major symmetry but the identity I : = : I
= holds only if the second-order tensor is symmetric.
The stiffness tensor in linear isotropic elasticity is expressed as
S e
I L 2 +
(2.31)
or in index notation as
( )
jk il jl ik jl ik
e
ijkl
+ + L
(2.32)
Using the tensorial notation, the generalized Hookes law can be
presented as
I L 2 3 : 2 : : + +
V S e

(2.33)
where
:
3
1

(2.34)
ELASTICITY EQUATIONS 20
is one third of the trace of the strain tensor, representing the relative
change of volume. The volumetric part of the strain tensor is
V
, and
when we subtract it from the strain tensor we obtain the deviatoric strain:
I I e : :
3
1
:
3
1
D S V

,
_

(2.35)
Therefore the deviatoric projection tensor is
I I
3
1
S D
(2.36)
and the volumetric projection tensor is
I
3
1
V
(2.37)
Then the volumetric-deviatoric decomposition of the strain tensor can be
done as follows:
( ) e I I I I I + + + +
V dev vol D V D V S
: : : :
(2.38)
And the stress tensor can be decomposed in the same way:
( ) s I I I I I + + + +
V dev vol D V D V S
: : : :
(2.39)
where
I :
3
1
:
V V

(2.40)
is the mean stress and
I s
V D
:
(2.41)
ELASTICITY EQUATIONS 21
is the stress deviator.
The elastic stiffness tensor can also be decomposed into its volumetric
and deviatoric part. Realizing that
V
I 3
(2.42)
we can rewrite the stiffness tensor in linear isotropic elasticity as
( )
( )
D V D V
D V V S e
G K I I I I
I I I I L
2 3 2 2 3
2 3 2
+ + +
+ + +


(2.43)
because the coefficient (3 + 2) is recognized as three times bulk
modulus K and = G = shear modulus. The generalized Hookes law:
( ) e I I I I L G K G K G K
V D V D V e
2 3 : 2 : 3 : 2 3 : + + +
(2.44)
is split into the volumetric and deviatoric part:
e s G K
V V
2 3 and
(2.45)
2.3.3 Engineering Notation
While the tensorial notation is useful in theoretical derivations for
developing a numerical algorithm that should be implemented into a
computer code, it is more practical to store the stress and strain
components in one-dimensional arrays (pseudovectors) and stiffness
moduli in two-dimensional arrays (matrices). Such arrangement is called
engineering notation (or Voight notation).
The normal components are usually arranged in a natural order, i.e.
x
followed by
y
and
z
. However, for the shear components several
conventions exist. In principle it is possible to use any of them, but the
selected convention must be used throughout the entire project. One
possibility is to set
ELASTICITY EQUATIONS 22

'

'

'

'

12
13
23
33
22
11
12
13
23
33
22
11
2
2
2

xy
xz
yz
z
y
x
xy
xz
yz
z
y
x
and
(2.46)
The engineering shear component
xy
is twice the tensorial shear strain
because then the energy product
ij ij
:
(2.47)
can be simply replaced in the engineering notation by a simple scalar
product of column vectors,
T
. However, this simple transcription of the
double contraction of two second-order tensors works only if we deal with
one stress-like tensor and one strain-like tensor. When evaluation the
tensorial norm of the stress tensor defined as
:

(2.48)
one has to be careful and realize that
( )
2
12
2
13
2
23
2
33
2
22
2
11
2
33
2
32
2
31
2
13
2
12
2
11
2
... :


+ + + + +
+ + + + + +
ij ij

(2.49)
while
2
12
2
13
2
23
2
33
2
22
2
11
+ + + + +
T
(2.50)
Therefore it is not possible to replace the tensorial norm of the stress
tensor by the Euclidean norm of the pseudovector of stress components.
Instead a diagonal scaling matrix P must be inserted:
P
T

(2.51)
ELASTICITY EQUATIONS 23
where the scaling matrix P has the following form:
1
1
1
1
1
1
1
1
]
1

2 0 0 0 0 0
0 2 0 0 0 0
0 0 2 0 0 0
0 0 0 1 0 0
0 0 0 0 1 0
0 0 0 0 0 1
P
(2.51)
Such scaling matrix leaves the normal components as they are and
doubles the shear components.
When calculating the norm of the strain tensor, it is also necessary to
insert a scaling matrix, but not the same one as for two stress-like tensors.
Since the shear components have already been doubled, the
corresponding scaling factors are now 0.5 instead of 2. Therefore it turns
out that the appropriate scaling matrix is the inverse of P and the tensorial
norm of is in the engineering notation evaluated as
P
1

(2.52)
where the inverse of scaling matrix P is obviously
1
1
1
1
1
1
1
1
]
1

5 . 0 0 0 0 0 0
0 5 . 0 0 0 0 0
0 0 5 . 0 0 0 0
0 0 0 1 0 0
0 0 0 0 1 0
0 0 0 0 0 1
1
P
(2.53)
For the purpose of the volumetric-deviatoric decomposition the
engineering counterpart of the unit second-order tensor (Kronecker delta)
is established as:
{ }
T
0 0 0 1 1 1
(2.54)
The volumetric-deviatoric decomposition is in the engineering notation
based on projection matrices
ELASTICITY EQUATIONS 24
1
1
1
1
1
1
1
1
]
1

0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 1 1 1
0 0 0 1 1 1
0 0 0 1 1 1
3
1
3
1
T
V
I
(2.55)
and
1
1
1
1
1
1
1
1
]
1

1 0 0 0 0 0
0 1 0 0 0 0
0 0 1 0 0 0
0 0 0 3 / 2 3 / 1 3 / 1
0 0 0 3 / 1 3 / 2 3 / 1
0 0 0 3 / 1 3 / 1 3 / 2
3
1
T
V D
I I I I
(2.56)
However, in the engineering transcription of expressions for the elastic
stiffness, the symmetric forth-order unit tensor I
S
must be replaced by the
inverse scaling matrix P
-1
. The elastic stiffness matrix is then:
D V e
G K I P I L
1
2 3

+
(2.57)
The generalized Hooks law
( ) I P I
D V
G K
1
2 3

+
(2.58)
is split into the volumetric and deviatoric part:
e P s
1
2 3

G K
V V
and
(2.59)
HYDROSTATIC AND DISTORTIONAL COMPONENTS 25
3 Homogenization
Many materials are inhomogeneous, i.e., they consist of dissimilar
constituents (phases) that are distinguishable at some small length scale.
The behaviour of inhomogeneous materials is determined, on the one
hand, by the relevant materials properties of the constituents and, on the
other hand, by their geometry and topology (phase arrangement).
Defects in an elastic material give rise to inhomogeneous stress and
strain fields by which the defects can be characterized. Equivalence
between an inhomogeneous material and some homogeneous material
with a certain eigenstrain or eigenstress distribution can be established.
3.1Eigenstrain
Since they are not caused by stress, eigenstrains are also referred to as
stress-free transformation strains (superscript t). Formally, all kinds of
strain which may prevail in a material in the absence of stress can be
interpreted as eigenstrains; typical examples are thermal or plastic
strains. In the framework of infinitesimal deformations the total strains
ij
are the sum of elastic strains
e
ij
= M
ijkl

kl
and the eigenstrains:
ij
=
e
ij
+
t
ij
.
Then the stresses are given by
( )
t
kl kl ijkl ij
L
(3.01)
3.1.1 Inclusions
For instance the phase transformation in solids, where atomic
rearrangements change the geometry of the lattice, gives rise to spatial
distribution of eigenstrain
t
ij
(x).
If nonvanishing eigenstrains prevail only in some bounded subregion
of the homogeneous material this region is called an inclusion and the
surrounding material is called matrix. It has to be emphasized that the
elastic properties of an inclusion and the matrix are the same; otherwise
the region is called inhomogeneity.
3.1.2 Eshelbys Solution
Probably the most important analytical solution of micromechanics has
been found by J.D. Eshelby (1916-1981). It is valid for an unbounded
domain which contains an ellipsoidal inclusion
(r)
with principal axes a
i
.
HYDROSTATIC AND DISTORTIONAL COMPONENTS 26
Fig. 3.1: Ellipsoidal inclusion in unbounded domain
If the eigenstrains in the inclusion are constant (
t
ij
= const.) then the
remarkable result holds that the total strains
ij
inside the inclusion are
constant as well. Via fourth-order Eshelby tensor S
ijkl
they depend linearly
on the eigenstrains:
) (r t
kl ijkl ij
const S in
(3.02)
and therefore
( ) ( )
( )
) (r t
kl
S
mnkl mnkl ijmn
t
kl
t
kl mnkl ijmn
t
kl kl ijkl ij
const I S L
S L L


in

(3.03)
where
( )
nk ml nl mk
S
mnkl
I +
2
1
(3.03)
is the symmetric fourth-order unit tensor, which is represented by the
identity matrix in the Mandel notation.
The Eshelby tensor is symmetric in the first and second pair of indices,
but in general it is not symmetric with regard to an exchange of these
pairs (exhibits the minor but not the major symmetry):
klij ijkl ijlk jikl ijkl
S S S S S ,
HYDROSTATIC AND DISTORTIONAL COMPONENTS 27
(3.04)
In case of an isotropic material its components depend only on Poissons
ratio , the ratios of the principal axes a
i
, and their orientation with respect
to some Cartesian coordinate system. The respective expressions are very
long and can be found in literature (e.g., Mura, 1982; Kachanov et al.,
2003).
The solution by Eshelby (1957) holds for an arbitrary anisotropic
material. Yet, only in case of an isotropic material is a closed-form
representation of the tensor S
ijkl
, and the fields outside the inclusion,
possible. The Eshelby solution for ellipsoidal inclusions is of fundamental
importance for analytical homogenization techniques.
Starting from the general ellipsoid various special cases can be
derived. For instance, the two-dimensional solution for an infinitelly long
cylinder of elliptic cross section in plane strain is obtained from the limit
process a
3
(see the following figure).
Fig. 3.2: Elliptic cylinder
The nonvanishing components of the Eshelby tensor in case of an
isotropic material are, for the orientation of the principal axes according to
previous figure, given by
HYDROSTATIC AND DISTORTIONAL COMPONENTS 28
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( )
( ) ( )
( ) ( )
2 1
1
2233
2 1
2
1313
2 1
2
2233
2 1
2
1133
2
2 1
2
2
2
1
1212
2 1
1
2
2 1
2
1
2211
2 1
2
2
2 1
2
2
1122
2 1
1
2
2 1
2 1
2
1
2222
2 1
2
2
2 1
2 1
2
2
1111
2
,
2
,
2
1 2
,
2
1 2
,
2
2 1
2
1 2
1
, 2 1
1 2
1
, 2 1
1 2
1
, 2 1
2
1 2
1
, 2 1
2
1 2
1
a a
a
S
a a
a
S
a a
a
S
a a
a
S
a a
a a
S
a a
a
a a
a
S
a a
a
a a
a
S
a a
a
a a
a a a
S
a a
a
a a
a a a
S
+

1
]
1

+
+
+

1
]
1

+

+

1
]
1

+

+

1
]
1

+
+
+
+

1
]
1

+
+
+
+

(3.05)
For a spherical inclusion (a
i
= a) in an isotropic material the
dependence on the principal axes and their orientation vanishes
(geometric isotropy) and the Eshelby tensor reduces to

,
_

+
kl ij ijkl kl ij ijkl
I S
3
1
3
1
(3.06)
where
( )
( )
( )
( )
( )

4 3 5
2 6
1 15
5 4 2
4 3
3
1 3
1
+
+

K
K
K
K
and
(3.07)
are scalar parameters. The entire (i.e., elastic and geometric) isotropy of
the problem then allows the decomposition into volumetric and deviatoric
strain which highlights the meaning of the parameters and :
) (r t
ij ij
t
kk kk
in and
(3.08)
HYDROSTATIC AND DISTORTIONAL COMPONENTS 29
3.2Inhomogeneities
The second class of defects which instead of eigenstrains in a
homogeneous material are characterized by inhomogeneous, i.e., spatially
varying, material properties are called inhomogeneities. We proceed in
that we first describe these defects by an equivalent eigenstrain in some
homogeneous comparison material in order to then apply again Eshelbys
result.
3.2.1 Ellipsoidal Inhomogeneities
As an important special case which allows applying the Eshelbys result
we consider an ellipsoidal material inhomogeneity
(r)
in an unbounded
matrix. Now the elastic properties are piecewise constant and given by the
elasticity tensors L
(r)
inside
(r)
for r =1, 2, 3... (inhomogeneities) and L
(0)
in
the surrounding matrix.
The total strain inside the inhomogeneity
(r)
as a function of the
external load
0
(or equal macroscopic strain E) is then
( ) ( ) ( ) ( )
( ) [ ] A L L M S 1 : : : :
1 0 0
+
r r
(3.09)
where E
(r)
is the strain in the r
th
phase, 1 is the fourth-order unit tensor,
M
(0)
is the compliance tensor of the matrix, L
(r)
is the stiffness tensor of the
r
th
phase while L
(0)
is the stiffness tensor of the matrix, E is the
macroscopic strain (equal to external load), and finally the fourth-order
tensor A is one of the influence tensors (the second one is denoted B),
namely so called strain concentration factor describing the relation
between the strain inside the inhomogeneity and the external load. The
previous equation can be written using the index notation as
( ) ( )
kl
r
ijkl
r
ij
A
(3.10)
The total stress inside the inhomogeneity
(r)
as a function of the
external load
0
(or equal macroscopic stress ) is
( ) ( ) ( ) ( )
( ) [ ] B L L M S 1 : : : :
0 0
+
r r
(3.11)
which can be written in the index notation as
( ) ( )
kl
r
ijkl
r
ij
B
HYDROSTATIC AND DISTORTIONAL COMPONENTS 30
(3.12)
3.3Effective Elastic Properties
A macroscopically homogeneous material may have a heterogeneous
microstructure on the microscopic level. Under certain conditions, the
material can be described on the macroscopic level as homogeneous with
spatially constant effective properties. This means that the microstructure
is averaged; this micro-to-macro transition is called homogenization.
The suitable volume for homogenization is called representative
volume element (RVE). The representative volume element must be big
enough to include enough non-homogeneities of materials (statistically
homogeneous distribution of the defects / heterogeneities), but small
enough to have the stresses and strains within RVE uniform (size with
respect to the analyzed detail of a structure).
3.3.1 Averaging
The macro-stresses and macro-strains which characterize the
mechanical state of the macroscopic material point are defined as the
volumetric averages of the microscopic fields:
( )

d
ij ij ij
x
1
(3.13)
for effective (average) stress in the volume . Employing divergence
theorem (also known as Gauss theorem, Green theorem or per-partes
integration in more dimensions) the macroscopic stress can also be
expressed by integrals over the boundary (curve integrals) of the
averaging domain :

dA x t
j i ij
1
(3.14)
The macro-strains are calculated as
( )

d
ij ij ij
x
1
(3.15)
which can be also expressed as
HYDROSTATIC AND DISTORTIONAL COMPONENTS 31
( ) ( )


+

dA n u n u d u u
i j j i i j j i ij
2
1
2
1
, ,
(3.16)
Often a volume of a heterogeneous material consists of n
subdomains and a matrix with volume fractions
( ) n r c
r
r
,..., 0
) (
) (

(3.17)
where r = 0 is used for the matrix itself. Obviously then
( )

n
r
r
c
0
1
(3.18)
where the elastic properties L
(r)
are piecewise constant. In case of such a
microstructure consisting of discrete phases we have
( ) ) (
0
r
ij
n
r
r
ij
c

(3.19)
since
( ) ( ) ( )
( ) ) (
0
0
) (
) (
) (
0
) (
) ( ) (
1 1 1
r
ij
n
r
r
n
r
r
ij
r
r
n
r
r
ij ij ij
c
d d d
r r

,
_

,
_

x x x
(3.20)
and analogously for the strains it holds that
( ) ) (
0
r
ij
n
r
r
ij
c

(3.21)
HYDROSTATIC AND DISTORTIONAL COMPONENTS 32
which means that the total stress / strain is the sum of phase stress /
strains with the weight c
(r)
.
3.3.2 Effective Elastic Constants
Analogous to the elasticity law on the microscopic level the effective
stiffness tensor is defined by the linear relation between the macro-
stresses and macro-strains:
kl
eff
ijkl ij
L
(3.19)
The interpretation of the effective stiffness tensor as a material property is
subjected to several conditions. It is, for instance, the equality of the
average strain energy density <U> in the volume when expressed by
means of the microscopic and macroscopic quantities:
kl
eff
ijkl ij
r
kl
eff
ijkl
r
ij
r
kl ijkl
r
ij
E L E E L E E L E U
2
1
2
1
2
1
) ( ) ( ) ( ) (

(3.20)
This requirement, known as the Hill condition (Hill, 1963) can also be
written in the form
ij ij ij ij

(3.21)
The relation between the applied (macroscopic) strains and the strains
within the individual phases can be expressed as
( ) ( ) ( )
mn
eff
ijmn mn
r
klmn
r
ijkl
n
r
r r
kl
r
ijkl
n
r
r r
ij
n
r
r
ij
L A L c L c c


) ( ) ( ) (
0
) ( ) (
0
) (
0
(3.22)
Therefore the effective stiffness matrix (in engineering notation) is
calculated as follows:
( ) ) ( ) (
0
r r
n
r
r eff
c A L L

(3.23)
Analogously, the compliance matrix can be calculated as
HYDROSTATIC AND DISTORTIONAL COMPONENTS 33
( ) ) ( ) (
0
r r
n
r
r eff
c B M M

(3.24)
It can be proved that
( ) ( )
I B I A


) (
0
) (
0
r
n
r
r r
n
r
r
c c and
(3.25)
where I is the identity matrix, A and B are strain and stress concentration
tensors, respectively. For the matrix with only one type of inhomogeneity
the combination of the previous equations results in the following
expression:
( ) ( ) ( ) ( )
( )
( ) ( )
( )
( )
( )
) 0 ( ) 1 ( ) 0 ( 0 ) 1 ( ) 0 ( 0 ) 1 ( ) 0 ( ) 0 ( 0
) 1 ( ) 1 ( 1 ) 0 ( ) 0 ( 0 ) 1 ( ) 1 ( 1 ) 0 ( ) 0 ( 0
A L L L A I L A L
A I L A L A L A L L
+ +
+ +
c c c
c c c c
eff
(3.26)
where the quantities with the superscript (0) represent matrix and the
quantities with the superscript (1) represent an inhomogeneity (also called
reinforcement or defect). Using the same reasoning for the different
boundary conditions (loading by the external load
0
) and for one type of
inhomogeneity, the stiffness tensor can be derived as
( )
( ) [ ]
1
) 0 ( ) 1 ( ) 0 ( 0 ) 1 (
:

+ B M M M L c
eff
(3.27)
3.3.3 Voigt and Reuss Approximation
In a homogeneous material the boundary conditions lead to
homogeneous (spatially constant) stress and strain fields. The first
possibility are prescribed linear displacements u
i
=
ij
0
x
j
and therefore a
constant strain
0
= const = E. The second possibility is loading by uniform
tractions t
i
=
ij
0
n
j
where
0
= const = .
In case of a heterogeneous material the simplest approximation is to
assume the micro-fields to be constant, in accordance with the boundary
conditions.
These approximations are exact only in one-dimensional special cases
of different materials arranged in parallel (Voigh) or in series (Reuss).
Despite obvious deficiencies the simple approximations by Voigt and
Reuss bear the advantage that they yield exact bounds for the true
HYDROSTATIC AND DISTORTIONAL COMPONENTS 34
effective elastic constants of a heterogeneous material. It can be shown
that
eff eff eff
K K K
Voigt Reuss

(3.28)
and
eff eff eff
G G G
Voigt Reuss

(3.29)
3.3.3.1 Voigt Approximation
If according to Voigt (1889) the strains inside a volume of a
heterogeneous body are taken to be constant (
(r)
= E = const). It is
obvious that the influence tensor is for Voigt approximation is A = 1. The
effective stiffness matrix (or tensor) is then approximated as follows:

( ) ) (
0
r
n
r
r eff
c L L L

Voigt
(3.30)
In the special case of discrete phases of an isotropic material the above
approximation leads to the effective bulk modulus
( ) ) (
0
r
n
r
r eff
K c K

Voigt
(3.31)
and shear modulus
( ) ) (
0
r
n
r
r eff
G c G

Voigt
(3.32)
If one of the phases is rigid (e.g., L
(1)

) one obtains L
eff
from the Voigt
approximation.
3.3.3.2 Reuss Approximation
Analogously, in the approximation according to Reuss (1929) a constant
stress field is assumed (
(r)
= = const) which corresponds to B = 1. The
effective compliance matrix (or tensor) is then approximated as follows:
HYDROSTATIC AND DISTORTIONAL COMPONENTS 35

( ) ) (
0
Reuss
r
n
r
r eff
c M M M

(3.33)
In the special case of discrete phases of an isotropic material the above
approximation leads to the effective bulk modulus
( )

n
r
r
r
eff
K
c
K
0
) (
Reuss
1
(3.31)
and shear modulus
( )

n
r
r
r
eff
G
c
G
0
) (
Voigt
1
(3.32)
In case of a matrix containing cavities or cracks the vanishing stiffness
(e.g., L
(1)

0), leads to L
eff
0.
3.3.4 Dilute (Non-Interacting) Defect Distribution
The simplest situation for modelling is when the inhomogeneities or
defects are so dilutely distributed in the homogeneous matrix that their
interaction among each other and with the boundary of the RVE can be
neglected (dilute distribution).
Fig. 3.3: Model of dilute phase distribution
As illustrated in the previous figure, each defect can be considered in
an unbounded domain subjected to a uniform far-field loading
0
= < >
HYDROSTATIC AND DISTORTIONAL COMPONENTS 36
= E or

0
= < > = . The characteristic dimension of the defects therefore has
to be small compared to their distance or to the distance from the
boundary of the RVE. Although the solutions obtained under these
idealizations are valid only for very small volume fractions (c
(r)
<< 1) they
form the basis for important generalization.
In case of an ellipsoidal inhomogeneity the strain inside the
inhomogeneity
(r)
is constant and given by the influence tensor A
(r)
, which
is calculated for each phase separately as
( ) ( ) ( ) ( )
( ) [ ]
1 0 0 ) 0 (
dil
: :

+ L L M S 1 A
r r
(3.33)
where the Eshelby tensor (or matrix in engineering notation) depends on
the matrix material.
Only in case of spherical isotropic inhomogeneities in an isotropic
matrix the macroscopic (effective) behaviour isotropic and the previous
equation can be split into the volumetric and deviatoric parts.
3.3.5 Mori-Tanaka Model
The approximation of a dilute distribution of non-interacting defects is
equivalent to the assumption that in a sufficient distance from each defect
the constant strain field
0
or stress field
0
of the external loading
prevails. This assumption is the starting point for a refinement of the
model to account for an interaction of inhomogeneities (defects).
In Mori-Tanaka model (1973) the strain or stress field in the matrix is,
in a sufficient distance from an inhomogeneity, approximated by the
constant filed
< >
M
or the average stress < >
M
as illustrated in the following figure.
Fig. 3.4: Interaction of inhomogeneities in the Mori-Tanaka model
HYDROSTATIC AND DISTORTIONAL COMPONENTS 37
The loading of each phase then depends on the existence of other
defects via the average matrix strain < >
M
= E
0
or the average matrix
stress < >
M
=
0
. Fluctuations of the local fields, however, are neglected
in this approximation of inhomogeneities interactions.
In the view of the idealized consideration of a single inhomogeneity in
an unbounded matrix, subjected to effective loading < >
M
or < >
M
, the
Mori-Tanaka model formally equals that of a dilute distribution and allows
the application of the already known concentration tensors from the dilute
distribution model, A
dil
(r)
, to represent the average strain in inhomogeneity:
( ) ( ) ( ) 0
dil
A
r r

(3.34)
In order to determine the effective material properties the average defect
strain needs to be represented as a function of the macroscopic quantities

0
= < > =E or

0
= < > = .
The total macroscopic strain is then
( ) ( ) ( ) ( ) ( ) ( )
( ) ( ) ) 0 ( ) (
1
0
) 0 ( ) (
1
0 0 ) (
1
0 0
A I
A

,
_

+
+ +

r
n
r
r
r
n
r
r r
n
r
r
c c
c c c c
(3.35)
For multi-phase materials consisting of a matrix (superscript 0) into
which n inhomogeneity phases are embedded the Mori-Tanaka models are
based on the following relations
( ) ( )
E A E
0
MT
0

(3.36)
where
( ) ( ) ( ) ( ) 1
dil
1
0 0
MT

,
_

+

r
n
r
r
c c A I A
(3.37)
is the Mori-Tanaka strain concentration factor for the matrix. The strain in
the individual inhomogeneities is
( ) ( ) ( ) ( ) ( )
A A A
0
MT dil
0
dil
r r r

HYDROSTATIC AND DISTORTIONAL COMPONENTS 38
(3.38)
Analogously the stress concentration factor for Mori-Tanaka method takes
the form
( ) ( ) ( ) ( ) 1
dil
1
0 0
MT

,
_

+

r
n
r
r
c c B I B
(3.39)
and the stress in the individual inhomogeneities is
( ) ( ) ( ) ( ) ( )
B B B
0
MT dil
0
dil
r r r

(3.40)
The effective macroscopic elastic tensor (matrix) is obtained as
( ) ( ) ( ) ( ) ( ) ( )
( ) ( ) ( ) ( ) 0
MT dil
1
) ( ) 0 ( 0
0
MT dil
1
) ( 0
MT
) 0 ( 0 ) ( ) (
0
A A L L
A A L A L A L L

,
_

+
+

r
n
r
r r
r
n
r
r r r r
n
r
r eff
c c
c c c
(3.41)
and the compliance elastic tensor (matrix) as
( ) ( ) ( ) ( ) ( ) ( )
( ) ( ) ( ) ( ) 0
MT dil
1
) ( ) 0 ( 0
0
MT dil
1
) ( 0
MT
) 0 ( 0 ) ( ) (
0
B B M M
B B M B M B M M

,
_

+
+

r
n
r
r r
r
n
r
r r r r
n
r
r eff
c c
c c c
(3.42)
The Mori-Tanaka model, in contrast to the model of a dilute
distribution, correctly covers the extreme cases of c
(r)
= 0 and c
(r)
= 1
(homogeneous material) and therefore can formally be applied for
arbitrary volume fractions c
(r)
.
In the special case of an isotropic matrix which contains isotropic
spherical inhomogeneities the Mori-Tanaka model yields, irrespective of
the spatial arrangement of the phases, an isotropic overall behaviour with
effective elastic constants:
HYDROSTATIC AND DISTORTIONAL COMPONENTS 39
( )
( )
1
) 0 (
) (
0
1
1
) 0 (
) (
0
) (
1
1 1
1 1

1
]
1

,
_

+
1
]
1

,
_

K
K
c
K
K
K c
K
r n
r
r
r
r
n
r
r
eff

(3.43)
and
( )
( )
1
) 0 (
) (
0
1
1
) 0 (
) (
0
) (
1
1 1
1 1

1
]
1

,
_

+
1
]
1

,
_

G
G
c
G
G
G c
G
r n
r
r
r
r
n
r
r
eff

(3.44)
where
( )
( )
( ) ( )
( )
( ) ( )
( )
( ) ( )
( )
0 0
0 0
0
0 0
0
0
4 3 5
2 6
4 3
3
G K
G K
G K
K
+
+

+
and
(3.45)
3.4Strength homogenization