Stress and deformation tensor

Author: Prof. Dr.-Ing. habil. Heinz Konietzky (TU Bergakademie Freiberg, Geotechnical In-
stitute)

1 Stress and deformation tensor ............................................................................ 2

Introduction .................................................................................................... 2
Example and definition of tensors.................................................................. 2
Representation of tensors and change of basis ............................................ 4
Pseudotensors............................................................................................... 5
Typical tensor operations .............................................................................. 6
Stress tensor ................................................................................................. 8
Deformation tensor ...................................................................................... 22
Compatibility condition................................................................................. 28
Equilibrium conditions.................................................................................. 29

Editor: Prof. Dr.-Ing. habil. Heinz Konietzky Layout: Gunther Lüttschwager

TU Bergakademie Freiberg, Institut für Geotechnik, Gustav-Zeuner-Straße 1, 09599 Freiberg  sekr_fm@ifgt.tu-freiberg.de
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

1 Stress and deformation tensor

Introduction
Geomechanical calculations have to consider the following 3 fundamental relations:
 Equilibrium conditions
 Compatibility conditions
 Constitutive laws
The coupling between the stresses and deformations is performed by the constitutive
laws (material laws) as indicated by figure 1. In order to describe these relations effec-
tively, the theory of tensors was developed at the end of the 19th century. During the 20th
century, the use of tensors has extended beyond continuum mechanics and now includes
among others the fields of special and general relativity, quantum mechanics, fluid me-
chanics and electromagnetism. In the context of geomechanics, we will use second-order
tensors to describe stresses and deformations and fourth-order tensors to describe the
constitutive laws. The scheme in figure 1 illustrates the interaction of the individual com-
ponents, which are explained in more detail within the next chapters.

inner + outer Forces Displacements

FI, FA ui

Equilibrium Compatibility
conditions conditions

Stresses Constitutive laws Deformations

ij ij

Fig.1: Geomechanical calculation scheme

Example and definition of tensors

Let’s examine the known vector product in 3
. The vector product of two vectors pro-
duces a third vector

z  w  x, z 3
1.1

Understood as a function that maps x  z( x ) , the vector product is linear so that

page 2 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

w   x    w  x  ,
1.2
w   x  y   w  x   w  y 

We will call such a linear function a tensor, in this specific case a second-order tensor.
Any linear function in 3 can be described through a multiplication with a matrix, so that
we can write

33
z  w  x  Wx, W 1.3

In the concrete case of the vector product, the matrix which describes the tensor takes
the following form
 0 w 3 w2 
 
W   w3 0 w1  1.4
 w 0 
 2 w1

Another example of a tensor is the rotation of a vector:

The rotation of a vector in 2

is a function which maps x  y ( x ) . This function is once
again linear
y  mx   my  x 
1.5
y  w  x   y w   y  x 

This function is therefore again called a (second-order) tensor and the rotation tensor
can be described by means of matrix multiplication.
y ( x )  Yx 1.6

With the rotation matrix

 cos   sin 
Y   1.7
 sin cos  

These two examples motivate the following definition: A multilinear function (i.e. a function
which is linear in all its arguments) that acts on a vector and generates another vector is
called a second-order tensor. Because vectors themselves can be used to represent lin-

page 3 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

ear functions, they can similarly be understood as tensors of a lower order, with our ten-
sors of second order acting on these lower-order tensors. This leads to the following in-
ductive, though highly abstract definition of tensors:

Tensors of the order n  r  s are the multilinear functions between the two tensor
spaces of the order r and s.

Representation of tensors and change of basis

In these two examples, we discussed that second-order tensors can be described by
matrices. Similarly, tensors of lower and higher order can be described by the generali-
zation of matrices in different dimensions. This leads to the representation of tensors up
to the fourth order in 3 as shown in tab. 1:

In addition to this index notation, different types of tensors can be can be expressed by
means of dashes above the symbols

a scalar = zeroth-order tensor

a vector = first-order tensor
a 3  3 matrix = second-order tensor
…

Because tensors are linear functions between vector spaces, every tensor can be ex-
pressed through components with respect to a basis of the vector spaces. Let’s now ex-
amine what happens when we change the basis of the vector space on which the tensor
operates.

Let’s assume that e   e1,..., en  and e   e1,..., en  are (ordered) bases of the n-dimen-
sional vector space V. Every vector, including every basis vector can be described as a
linear combination of the basis vectors.
n
ej   aij ei 1.8
i 1

This means that a change of basis is described through a series of coefficients 𝑎𝑖𝑗 . If Tij
are the components of the Tensor T with respect to the basis e, then, because of the
linearity of tensors, we can obtain the components of T with respect to e’ through

Tab. 1: Matrix and tensor definition

symbol matrix type tensor order no. of values phys. example

a scalar zeroth 1 density
ai vector first 3 displacement
aij 33 second 9 stress
aijk 333 third 27 --
aijkl 3333 fourth 81 stiffness matrix

page 4 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

n
 n 
Tkl     akj aliTij  1.9
i 1  j 1 

with k, l = 1,2…n, or using the shorter Summation Convention

Tkl  akj aliTij 1.10

Going forward, this summation will always be implied if an index appears twice in a mul-
tiplicative term. It is worth noting that there are different ways to define tensors. Occa-
sionally, the described transformation behavior of the describing matrices is used as an
equivalent definition to the one we used.

Pseudotensors
If tensors can be described through generalized matrices, one can ask the question why
we bothered with our original definition, which is certainly less intuitive. In short, not eve-
rything that can be described as a n-dimensional matrix behaves like a tensor. For exam-
ple, let’s examine the permutation symbol, also called the Levi-Civita-symbol. This symbol
is defined by the sign of a permutation of the numbers 1,2,…,n for an integer n. The
permutation symbol can be defined in any dimension greater than one. In two dimensions,
it is

1 if  i , j   1,2 

 ij  1 if  i , j    2,1 1.11

 0 if i  j

And arranged into a 2  2 antisymmetric matrix

 0 1
 ij    1.12
 1 0 

In three dimensions, it is,

1 if  i , j , k   1,2,3  or  2,3,1 or  3,1,2  (even permutations)


 ijk  1 if  i , j , k    3,2,1 or 1,3,2  or  2,1,3  (uneven permutations) 1.13

 0 if i  j , i  k, j  k

Arranged into a 3  3  3 matrix:

1.14

page 5 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

While the permutation symbol has a representation as a generalized matrix, it does not
follow the transformation rules of a tensor. Under certain orthogonal transformations, for
example a reflection in an odd number of dimensions, it should be multiplied by -1 if it
were a tensor. However, the permutation symbol does not change at all and is therefore
not a proper tensor.

Typical tensor operations

Composition
Let A, B be two second-order tensors. The Tensor C = AB is called composition of A and
B if

Cx  A  Bx  1.15

The composition of two tensors is not generally commutative so that AB  BA . The com-
position is called commutative if the opposite is true.

Powers, polynomials and functions of second-order tensors

On the basis of the composition we can define the powers of second-order tensors as
follows
Am  A  A  ...  A, A0  1 1.16

Powers of tensors are characterized by the following properties

A   A
l
  k Ak
k
Ak Al  Ak l , k
 Akl ,
1.17

Transposition
The transposed tensor AT is defined by

AT x  xA . 1.18

From this it follows that

A 
T
T
 A. 1.19

The transpose of a matrix is created by switching the row and column indices:

T
 a11 ar 1   a11 ar 1 
   
   . 1.20
a ars  a ars 
 r1  1s

The transposition of tensors is a linear operation and the transposition of the composi-
tion of second-order tensors is

page 6 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

 AB 
T
 B T AT 1.21

Inversion
Let y  Ax :

The tensor A is called invertible if there is a tensor A-1 which satisfies the condition

x  A1y 1.22

If the inverse exists it follows that

A1A  I 1.23

With the unit tensor I, also called Kronecker delta

 1 0 0
  0 if i  j
I   0 1 0    ij   1.24
 0 0 1 1 if i  j
 

Tensor product
The tensor product enables us to construct a tensor from two lower-order tensors. The
order of the resulting tensor is the sum of the orders of the original tensors

A  B v1,...,v n ,v n 1,...,v n m   A v1,...,v n  B v n 1,...,v n m  1.25

The resulting function is again linear in all its arguments and therefore a tensor.

page 7 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

Stress tensor
Load is generated by outer forces FA (area force) or inner force FI (volume forces) ac-

cording to
Fig. 1.1. For an arbitrary orientated cut a stress vector t is obtained, assumed that only
forces and no moments are transferred. A denotes the area, where the force vector is
considered.

 F 
t  lim  
A0 A
 
1.26

The stress state can be defined in a cartesian coordinate system as illustrated in fig. 1.2.

Along the three faces of the cube three stress vectors t1, t2 and t3 can be obtained:
Whereby {i1, i2, i3} represent the three stress components on the particular cube faces
(Fig. 1.2).

In detail the stress tensor can be described as follows:

 11  12  13   xx  xy  xz 
 
 ij  t1, t2 , t3    21  22  23     yx  yy  yz 
T

 31  32  33    zx  zy  zz 
1.27

Fig. 1.1: Solid body with volume and area forces

page 8 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

Fig. 1.2: 3-dimentional stress components at a cube

page 9 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

The first index specifies the normal of the particular face under consideration, the second
index the impact direction of the stress component. According to equation 1.23 the stress
tensor consists of 9 elements. However, assumed that the sum of the moments is zero,
pairwise identical shear stresses are obtained. This feature is also called ‘Boltzmann-
Axiom’ and explained in more detail in Fig. 1.3 and by equation 1.24 for the 2-dimensional
case (the extension to 3D is straightforward).

M xy  0   xy  l  4l 2   yx  l  4l 2   xy   yx

M xz  0   xz  l  4l 2   zx  l  4l 2   xz   zx 1.28

M yz  0   yz  l  4l   zy  l  4l
2 2
  yz   zy

From eq.1.28 it follows, that the stress tensor is symmetric, that means:
T
 ij   ji or 
1.29

Therefore, the number of stress values is reduced from 9 to 6 (three pairwise identical
shear stresses meaning no rotations). The relationship between stress vector and stress
tensor is obtained on the basis of the equilibrium conditions in direction of the coordinates
xi (Fig. 1.4):


ni  cos n, xi , 
1.30
dAi  ni dA ,
1.31
where ni is the unit normal vector.

yy

yx
xy
l
xx xx

xy
yx

yy
Fig. 1.3: Equilibrium considerations for a volume element (2D, x-y-plane)

page 10 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

Fig. 1.4: Orientation of stress tensor and stress vector

t1 dA   11 dA1   21 dA2   31 dA3

t2 dA   12 dA1   22 dA2   32 dA3
t3 dA   13 dA1   23 dA2   33 dA3
1.32
Using (1.30) and (1.31) equation 1.32 can be simplified as follows:

t1  11n1  21n2  31n3

t2  12n1  22n2  32n3 .
t3  13 n1  23 n2  33 n3
1.33
Equation 1.33 can be rewritten in tensor form as follows:

ti   ji n j  ij n j
T .
n n
1.34

Equation 1.34 documents the equality of pairwise shear stresses. The so defined second-
order stress tensor is called ‘Cauchy stress tensor‘ or ‘true’ stress tensor or ‘Euler stress
tensor’. The Cauchy stress tensor ij relates the current force vector to the current (de-
formed) area element.

dFi   ji dAj
1.35
Fi: current force vector
Aj: current area element with dAj  n j dA

Alternatively, the current force vector Fi can be related to the original area A° (that means
before any deformation!). Such a stress tensor is called ‚Nominal stress tensor‘, ‘La-
grange stress tensor’ or ‘First Piola-Kirchhoff tensor’ Tij:

page 11 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

dFi  T ji dAj
1.36

The stress tensor can be decomposed into normal and shear components (n: normal
vector; m: tangential vector) as illustrated by Fig. 1.5:

 n  ni ti  ni  ij  n j
1.37
or
n  mi ti  mi ij n j
1.38
In detail, equations 1.37 and 1.38 can also be written as:

 n  n1  11 n1  n1  12 n2  n1  13 n3
 n2  21 n1  n2  22 n2  n2  23 n3
 n3  31 n1  n3  32 n2  n3  33 n3
1.39
From equation 1.39 follows for instance:

 1
 
n  0   n   11 and
0
 
0
 
n  0   n   33 .
 1
 

For the shear stress follows:

 n  m1  11 n1  m1  12 n2  m1  13 n3
 m2  21 n1  m2  22 n2  m2  23 n3 .
 m3  31 n1  m3  32 n2  m3  33 n3
1.40

From equation 1.40 follows for instance:

 1 0
   
n  0 ; m   1   n   21
0 0
   
0 0
   
n  0 ; m   1   n   23
 1 0
   

page 12 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

n
n
m
m
.
Fig. 1.5: Decomposition of stress vector t into normal and shear stress component

If  n  mi  ji n j , than:
 1 0
   
n  0 ; m   1   n   12
0 0
   

Thereby, it always holds: ni ni = 1 and mi mi = 1

Now we consider specific directions, where only normal stresses  exist, but no shear
stress . For such a constellation it holds:

ti = ij  nj and ti =   ij  nj,

1.41
where nj characterizes the principal stress directions.

Equalization of both expressions in 1.41 yields:

ij nj =   ij  nj or (ij - ij  )  nj = 0.

1.42

Equation 1.42 describes an eigenvalue problem with eigenvalues  und nj. The non-trivial
solution is obtained if the coefficient determinant of equation 1.42 vanishes:

det  ij   ij   0 ,
1.43
or

 11    12  13
 12  22    23  0 .
 13  23  33  
1.44

page 13 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

The solution of equation 1.44 is a characteristic equation of third order:

 3  I1  2  I2   I3  0 ,
1.45
where the following holds:

I1   KK  11   22   33   ij  ij ,
1.46
  12  11  13  22  23
I2 
1
2
 ii  jj   ij  ji   11  
 21  22  31  33  32  33 ,
  11 22   22 33   11 33   12
2
  23
2
  31
2

1.47
11 
I3  det  ij  
3
  ii  jj  KK   ij  jK  Ki   ij  ji  KK 
3 2 2 .
  11 22 33   11 23
2
  22 13
2
  33 12
2
 2 12 23 31
1.48

The values I1, I2, I3 are called ‚main invariants‘ (I1: first main invariant, I2: second main
invariant, I3: third main invariant) of the stress tensor, that means that they are independ-
ent of the coordinate systems (independent of translations or rotations of the reference
system). Besides these main invariants there are the so called ‚basic invariants‘, which
can be considered as a special subset of the main invariants. They are defined as follows:

J1  kk  I1
1 1
J2  ij  ji  I12  I2 .
2 2
1 1
J3  ij  jk ki  I13  I1 I2  I3
3 3
1.49

Besides the cartesian representation it is also possible to find a formulation in form of the
principal stresses:

I1  1   2   3 ,
1.50
I2  1 2   2 3  1 3 ,
1.51
I3   1 2 3 .
1.52

An interesting decomposition of the stress tensor is possible, if a mean normal stress is

defined as follows:

page 14 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

1 1
 0   KK   11   22   33  .
3 3
1.53
0 is also called „hydrostatic stress state“ or ‘mean stress’ or ‘spherical stress’.
Based on these definitions the stress tensor can be written as:

 ij   0 ij  sij
1.54
In terms of matrix notation this means:

 11  12  13   0 0 0   11   0  12  13 
  0  0    
 21  22  23    0   21  22   0  23 
 31  32  33   0 0  0    31  32  33   0 
,
 0 0 0   s11 s12 s13 
 0  0   s s23 
 0  21 s22
 0 0  0  s31 s32 s33 
1.55
where sij is referred as deviatoric stress part.
For the spherical tensor as well as for the stress deviator invariants can be defined. The
main invariants for the spherical tensor are given as follows:
3
I1  3 0 I2   02 I3   03
2
1.56
The corresponding basic invariants are:

3 2
J1  3 0 J2  0 J3   03
2
1.57
For the deviatoric part the main invariants are:

I1D  skk  11   0    22   0    33   0   0

1.58

I2D 
1
2
sii s jj  sij s ji 
  11  0  22  0    22  0  33  0    11  0  33  0   12
2
  223   31
2

1.59
I  det  sij 
D
3

1 1 3 
  sii s jj skk  sij s jk ski  sij s ji skk 
32 2 
1.60
The basic invariants for the deviatoric part are:

page 15 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

J1D  skk  0

1.61
1 1
sij s ji   11  0    22  0    33  0   212 2 
2 2 2
J2D  2
 223
2
 231
2 2  
1
  11  22    22  33    33  11    12
2 2 2 2
 23
2
 31
2

6 
1
 1  2    2  3    3  1  
2 2 2

6 
1.62
1
J3D  sij s jk ski   1  0    2  0    3  0 
3
1.63

Quite often stress components are defined, which are related to the octahedral plane.
The octahedral plane is equally inclined to the principal stress directions (hydrostatic
axis). The principal stresses act along the x1, x2 and x3 direction:

 1 0 0
 
ij   0 2 0
0 0 3 


The stress vector tj is defined by the three principal stress components 1, 2 and 3.
Regarding the normal on the octahedral plane the stress vector tj has the following carte-
sian components:

1
tiN  ij n j nj  .
3
1.64
1
x1
 1 
  arc cos    54,7
nj
3  3
t j  1, 2 , 3 
tj
2 
 
x2
2
x 1
3 3
Fig. 1.6: Representation of octahedral stresses

page 16 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

The projection and summation of the components on the vektor nj (hydrostatic axis) pro-
vides the octahedral normal stress:

1  1    1
OCT    2  3    1  2  3   0 .
3 3 3 3 3
1.65

The octahedral normal stress is equivalent to mean stress (Equation 1.53). The subtrac-
tion of the octahedral normal stresses from the principal stresses leads to the deviatoric
stresses:
s1  1  0
s2   2   0
s3  3  0
1.66

These deviatoric stresses can also be referred to the octahedral plane and given as Car-
tesian components:

s1 s2 s3
t1s  t2s  t3s  .
3 3 3
1.67

The addition of vectors leads to the octahedral shear stresses:

t   t   t 
2 2 2
 OCT  1 2 3

s12 s22 s32

  
3 3 3
.

1 2
3

s1  s22  s32 
2 D 1
 J2  sij sij
3 3
1.68

Another very popular quantity is the so-called ‚von-Mises equivalent stress‘ F. This stress
value is based on a strength criterion, which relates the yield stress F to the stress devi-
ator:

0  3J2D  F2 .
1.69

page 17 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

This implies that:

3 1
 F  3J2D   1   2    2   3    1   3 
2 2 2
sij sij 
2 2
1.70
and

2 2 2
 OCT  F  F .
3 3
1.71

Principal stresses and principal stress directions:

The stress tensor as a symmetric linear operator has the characteristic, that it can be
diagonalised. That means, there are three orientations (directions) perpendicular to each
other in space, where the corresponding normal stresses reach extreme values (principal
stresses or principal normal stresses) and the shear stresses vanish. In this case, only
the trace of the tensors has non-vanishing values:

 1 0 0
 
ij   0 2 0 .
0 0 3 

1.72

The stress vectors on these specific surface areas coincide with the directions of the
normal vectors of these surface areas. Therefore, the stress vectors have only one non-
vanishing component. Thus, for the stress vector at the considered surface area it holds:

ti  ni 

and

t1  n1 1  l 1
t 2  n2 2  m 2 .
t3  n3 3  n 3
1.73

The normal vector ni  l , m, n describes the principal normal stress directions. For the
unit vector the following holds in general:

n
i 1
2
i  l 2  m2  n 2  1 ,

1.74

squaring equation 1. yields:

page 18 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

t12  l 2  12
t 22  m 2  22
t32  n 2  32
1.75
and
t12
l  2
2

1
t 22
m2  .
22
t32
n2 
32
1.76
The addition of the equations 1.76 under consideration of equation 1.74 gives:

t12 t22 t32

   1.
 12  22  32
1.77

Equation 1.77 describes an ellipsoid, that means the values 1, 2 and 3 represent the
half-axes of the ellipsoid (Fig. 1.7). The surface of the ellipsoid represents all possible
stress vectors. If two principal stresses are equal, a spheroid is coming up. If all principal
stresses are equal (isotropic stress state) a sphere is coming up.
In geomechanics, especially in soil mechanics, descriptions on the basis of the deviatoric
stress plane, see Fig. 1.8, are very common.

Fig. 1.7: Prinzipal stress ellipsoid

page 19 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019


2 3 const.
1
1
Deviatorebene

T (
1 2 3
s3che Achse
t
s 1 stati
ro
2
 3
arccos    Hyd

 3 
h
3

2
Fig. 1.8: Decomposition of the stress state into hydrostatic and deviatoric part, where the stress vector t
defines the stress point T

1'

T 

3'
2'
Fig. 1.9: Illustration of Lode angle θ in the -plane

3 3
h   1   2   3   I1
3 3
1.78
s  s s s 2
1
2
2
2
3 2J D
2

1.79

On the deviatoric plane it holds:

 1   2   3   const.
1.80

The deviatoric plane through the coordinate system is also called π-plane (fig. 1.9).

page 20 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

It holds:

3 3 J3D
cos  3    3
and
J 
2 D 2
2

1.81
 
1  3 3 J D3 .
  arccos
3  2 3 
 (J D2 ) 2

1.82
In geotechnical engineering the follwoing two modified invariants are often used:
Roscoe invariants p und q as well as Lode angle θ. Thereby, it holds:

1
p  1,
3
1.83
q 3J D
2 and
1.84
 
1  3 3 J D3 
  arccos .
3  2 3 
 (J 2 ) 
D 2

1.85

For the conventional triaxial test the following expressions (eq. 1.86-1.88) can be de-
duced:

Fig. 1.10: Illustration of Lode angle in the principal stress space

page 21 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

1
p ( 1  2  3 ) ,
3
1.86
q   1   3 and
1.87
1
 arccos (3 6 s1  s 32)  3 6 s1s2s3 .
3
1.88

Deformation tensor
For the coordinates of a point at the initial and final deformed state the following inverse
o  o o
 
relations exist: xi  xi  x j  and x i  x i  x j  .
   
The definition of the deformation tensor can be made in two systems:

1. In relation to the undeformed initial system

(= Lagrange approach), that means ui is a function of the initial coordinates
 
ui  ui  x j  .
 
1.89
1. In relation to the deformed final system
(= Euler approach), that means ui is a function of the final coordinates.
~
 
ui  ui  x j  .
 
1.90

x2
P
ui
„Lagrange“
P
x2
x1
x3
x1
x3
x2
P
ui
„Euler“
P x2
x1
x3
x1
x3

page 22 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

Fig. 1.11: Euler and Lagrange approaches in respect to deformations

The general definition of the deformation tensor reads as follows:

L xK xK
 ij  (Lagrange)
 xi  x j
1.91
and
 
E
 xK  xK
 ij  (Euler).
x i x j
1.92
ui ui
With the help of the gradient tensors (= displacement gradients) and , respec-
 xj x j
tively, the deformation tensor can be defined as follows:

„Lagrange“:

  xi u
xi  xi  ui  xi  with   ij  i and
   xj xj
1.91
L  ui   ui 
 jK    ij    ij  
   
  x j   xK 
,
uK u j ui ui
  jK   
 x j  xK  x j  xK
1.94

„Euler“:
 xi u
xi  xi  u  x j  with   ij  i and
x j x j
1.95
E u u ui ui
 jK   jK  i  K  .
xK x j x j uK
1.96

page 23 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

Illustration of the fundamental distinction between Euler and Lagrange approaches using
numerical meshing:

a) Lagrange Same nodes, but different „geographic“ coordinates

B (2, 4)

B (2, 4)
A (2, 2) A (2, 2)

Original Deformed

b) Euler new nodes, but old „geographic“ coordinates

B (2, 4)

A (2, 2) B (2, 2)
A (2, 1)

Original Deformed

For the Lagrangian approach the grid follows the deformations. For the Euler approach
the material ‚flows‘ through the stiff grid.

Besides the displacement gradient and the deformation tensor, the deformation gradient
Fij is of vitial importance:
xi  xj
FijL   Fij or FijE   Fij( 1) .
 xj xi

1.97

The deformation gradient is a second-rank tensor. He projects the line element vector
dsi (initial configuration) to line element vector ds (current configuration). Thereby, the
same material points are considered (Fig. 1.12).

page 24 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

Bahnlinien ds

d s°

x
Fig. 1.12: Illustration of deformation gradient

It holds:

d si  Fij  d s j

and

d si  Fij( 1)  d s j .
1.98

From the engineering point of view the deformation gradient can be defined according to
equation 1.94:

G
1 L

 ij    i K   jK 
2  
 
1  u j u ui ui 
  K 

2x 
 K  x j  x j  xK 
1.99

or according to equation 1.96:

A
1 E

 jK    jK   jK 
2 
.
1  u u ui ui 
  j  K  
2  uK x j x j xK 
1.100

page 25 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

Expression 1.99 is called ‘Green deformation tensor’, the expression 1.100 is called ‘Al-
mansi deformation tensor’. In the engineering praxis the Green deformation tensor is pre-
ferred. Moreover, most often the quadratic term is neglected under the assumption, that
u i
1. Thus, for small deformation, the distinction between Langrangian and Eulerian
 xj
approaches disappears and the simplified deformation tensor is given as:
 
1  ui u j .
 ij  
2   x  xi

 j 
1.101
The deformation tensor according to equation 1.101 can be extended to include rotations:

 ij 
1
2
 ui , j  u j , i    u i , j  u j , i 
1
2
.
 eij  w ij
Deformations Rotations

1.102

Fig. 1.13: Illustration of rotation and deformation (2D)

It holds:

 0 w12 w13  w12   w 21

 
w ij   w 21 0 w 23  with w13   w 31 and
w  w 23   w 32
 31 w 32 0 
1.103
 e11 e12 e13  e12  e21
 
eij   e21 e22 e23  with e13  e31 .
e  e23  e32
 31 e32 e33 

page 26 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

1.104
Thus, the deformation tensor can be written as:

 e11 e12  w12 e13  w13 

 
 ij   e21  w 21 e22 e23  w 23  ,
e  w e32  w 32 e33 
 31 31

1.105
with
eij 
1
2
 ij   ji  and w ij 
1
2
 ij   ji  for i  j .

1.106
eij is called deformation tensor, wij is called rotation tensor. It holds:

1
eij   ij for i  j ,
2
1.107
Where ij are shear strain components and e11, e22 and e33 are direct strain components
(elongations or shortenings).

The volumetric strain v is given by the following expression:

 dV
v   KK  11  22  33 .
dV
1.108

The mean direct strain (elongation or shortening)  0 is given by:

1 1
 0   KK   v .
3 3
1.109

In most cases rotations are neglected and it holds:

 e11 e12 e13  e12  e21

 
 ij   e21 e22 e23  with e23  e32 .
e e33  e13  e31
 31 e32
1.110

In complete analogy to the stress tensor invariants can be defined also for the deformation
tensor, e.g.:

page 27 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

I1  e11  e22  e33 ,

1.111
I2  e11 e22  e22 e33  e11 e33 and
1.112
I3  e11 e22 e33 .
1.113
Compatibility condition
From expression 1.114 the strain components can be obtained in a unique manner. Oth-
erwise, the displacements can not be obtained in a unique manner based on given strains
only. The compatibility conditions (= conditions of integrability) are necessary additional
requirements to deduce displacements on the basis of given strain components by inte-
gration. The consideration of the compatibility conditions guarantees that strains lead to
a ‘correct’ displacement field and the continuum is not disturbed. Starting point is the
deformation tensor:

 ij 
1
2
ui , j  u j ,i  .
1.114

Second derivatives of equation 1.114 with corresponding index permutations give the
following four expressions:

 ij , kl 
1
2
u i , jkl  u j , ikl 
 kl , ij 
1
2
u k , lij  ul , kij 
.
 ik , jl 
1
2
u i , kjl  uk , ijl 
 jl , ik 
1
2
u j , lik  ul , jik 
1.115

Due to the fact that the sequence of differentation is arbitrary, through addition and sub-
traction of the expressions 1.115 the following expression is obtained:

 ij , kl   kl , ij   ik , jl   jl , ik  0
1.116

From expression 1.116 the 6 compatibility conditions can be deduced under the condition
 ij   ji for i  j as follows:

page 28 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

11, 22   22, 11  212, 12  0

 22, 33   33, 22  2 23, 23  0
 33, 11  11, 33  213, 13  0
.
11, 23   23, 11  13, 21  12, 31  0
 22, 31   31, 22   21, 32   23, 12  0
 33, 12  12, 33   32, 13   31, 23  0
1.117

First equation in 1.117 can exemplary also be written as:

 2 xx   yy  2  xy
2

  2 .
y 2 x 2 x y
1.118

Under plain strain conditions all strain components and derivations in respect to the third
direction in space vanish, that means only equation 1.118 is left over. Equation 1.118
indicates, that the second derivations of the direct strains and the second derivations of
the angular distortions have to be in due proportion. The above used term strain is the so
called ‚technical strain‘ or ‘Cauchy strain’ in contrast to the so called ‘logarithmic strain’ or
‘Hencky strain’. Only for small deformations both expressions (Equations 1.119 and
1.120) provide nearly the same value:

l
Technical strain:  ,
l
1.119
l
Logarithmic strain:   ln .
l
1.120

Equilibrium conditions
For any volume element inside a body the forces and moments have to be in equilib-
rium. Usually it is assumed, that the solid body does not rotate and therefore the sum of
the moments is zero by default. According to Fig. 1.14 the following yields:

F x  0:
  x    
  x  d x  d y d z   x d y d z   yx  yx d y  d z d x
 x   y  ,
  zx 
  yx d z d x   zx  dz d x dy
 z 
  zx d x d y  Fx d x d y d z
1.121

page 29 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

F y  0:
  y    
  y  d y  d x d z   y d x d z   zy  zy d z  dx d y
 y   z 
and
  xy 
  xy d z d y   xy  d x dy dz
 x 
  zy d x d y  Fy d x d y d z
1.122

F z  0:
  z    
  z  d z  d x d y   z d x d y   zy  zy d y  d x d z
 zy  y
    .
  
  zy d x d z   xz  xz d x  d y d z
 x 
  xz d y d z  Fz d x d y d z
1.123

Fig. 1.14: Force equilibrium at volume element (Fi: volume forces)

page 30 of 31
 Stress and deformation tensor
Only for private and internal use! Updated: 23. Mai 2019

Equations 1.121 to 1.123 can be simplified in the following way:

 x  yx  zx
   Fx  0 ,
x y z
1.124
 xy  y  zy
   Fy  0 ,
x y z
1.125
 xz  yz  z
   Fz  0 .
x y z
1.126

page 31 of 31