Stress and Strain Tensors

Stress at a point.
Imagine an arbitrary solid body oriented in a cartesian coordinate system. A number
of forces are acting on this body in different directions but the net force (the vector
sum of the forces) on the body is 0. Conceptually slice the body on a plane normal to
the x-direction (parallel to the yz-plane). Take a small area on this plane and call it

A x 
A x x̂ . Calculate the resolved force acting on this small area and call it
F .

F 
F x 
F y 
F z 
F x x̂ 
F y ŷ 
F z ẑ .

Notice that since
F is the total force acting only on
A x , the magnitude of
F will

change as
A x changes.

We can define three scalar quantities,

F
F
F
 xx  lim ----------x- ,  xy   xy  lim ----------y- , and  xz   xz  lim ----------z- .

A x → 0
A x
A x → 0
A x
A x → 0
A x

The first subscript refers to the plane and the second refers to the force direction. If
we do the same conceptual experiment at the same location but in the y and z-direc-
tions, we obtain

F
F
F
 yy  lim ----------y- ,  yx   yx  lim ----------x- ,  yz   yz  lim ----------z- ,

A y → 0
A y
A y → 0
A y
A y → 0
A y

F
F
F
 zz  lim ----------z- ,  zx   zx  lim ----------x- , and  zy   zy  lim ----------y- .

A z → 0
A z
A z → 0
A z
A z → 0
A z

For static equilibrium  xy   yx ,  yz   zy , and  xz   zx , resulting in six inde-

pendent scalar quantities. These six scalars arranged in an ordered 3  3 matrix

forms the stress tensor,

 xx  xy  xz

  ij   xy  yy  yz .
 xz  yz  zz

E106 Stress and Strain Tensor Summary Page 1

For example, the stress tensor for a cylinder with cross-sectional area A 0 in uniaxial

tension from force F is

F
------- 0 0
A0

  ij  if the cylinder axis and F are both parallel to the x-axis,
0 00
0 00

0 0 0

  ij F
 0 ------
- 0 if the cylinder axis and F are both parallel to the y-axis,
A0
0 0 0

00 0
and
  ij  0 0 0 if the cylinder axis and F are both parallel to the z-axis.
F
0 0 -------
A0

The sign convention for the stress elements is that a positive force on a positive face
or a negative force on a negative face is positive. All others are negative. As a final
example, a cube oriented so that its faces are perpendicular to the coordinate axes,

with an area per face of A 0 has the following forces applied to it: force F 1 applied to

the positive x face in the positive x direction, force F 2 applied to the positive y face in

E106 Stress and Strain Tensor Summary Page 2

the positive y direction, and force F 3 applied to the positive x face in the negative y

direction. The necessary forces to keep the cube form moving are applied to the other
faces. The resultant stress tensor is

F F
------1- -----------3- 0
A0 A0

  ij   F F .
-----------3- ------2- 0
A0 A0
0 0 0

Stress on a plane.
It is often necessary to calculate the stress on an arbitrarily-oriented plane with nor-

mal n̂ inside a solid. A force balance on the tetrahedron formed by the intersection of

the plane with the coordinate axes provides the needed results. We define a stress

vector, s , defined as limit of the net force acting on the plane, F , per unit area as the

area shrinks to zero. This vector can be decomposed into the normal stress on the

plane (the force per unit area in the direction normal to the plane), s n , and the shear

stress on the plane (the force per unit area in a direction lying in the plane), s s . It

can also be decomposed into components in the three coordinate-axis directions, s x ,

s y , and s z .

In other words, s  s x  s y  s z  s x x̂  s y ŷ  s z ẑ . If we define the direction

cosines of n̂ as

k  n̂ • x̂  cos  , l  n̂ • ŷ  cos , and m  n̂ • ẑ  cos

,

then

s x   xx k   xy l   xz m , s y   xy k   yy l   yz m , and

s z   xz k   yz l   zz m ,

E106 Stress and Strain Tensor Summary Page 3

or

sx  xx  xy  xz k
sy   ij n̂   xy  yy  yz l .
sz  xz  yz  zz m

We can then find the normal component of the stress, s n , by

k
s n  s • n̂  s x s y s z l .
m

The shear component can be determined but requires a little more work.

E106 Stress and Strain Tensor Summary Page 4

Coordinate transformations and stress invariants.

It is often useful to know the stress tensor in a coordinate system that has been
rotated and/or translated with respect to the original coordinate system. We can
T T
transform the coordinates, x y z into the coordinates, x y z by use of a

1 T
transformation matrix, Q , where Q has the property Q  Q . In particular, a

rotation about the z -axis through an angle  is given by

x cos  sin  0 x
y   sin  cos  0 y .
z 0 0 1 z

To get all of the elements of the stress tensor in the new coordinate system,
T

  Q
Q .

The above relationship is often used to define a tensor of rank 2. Several properties
of the stress tensor remain unchanged by a change in coordinates. These properties
are called invariants. These invariants are closely related to important quantities.

The first invariant, I 1 , is the trace of the matrix,

I 1   xx   yy   zz .

The hydrostatic component of  ij (the part due to uniform pressure on all exterior

surfaces of the solid) is equal to I 1  3 .

The second invariant, I 2 is given by

I 2   xy
2  2  2   
yz zx xx yy   yy  zz   zz  xx .

The third invariant is

I 3   xx  yy  zz  2  xy  yz  zx   xx  2yz   yy  zx
2   2 .
zz xy

A common question in stress problem is: Is there a coordinate system for which all of
the shear stresses disappear, and the remaining stresses are purely tensile or com-

E106 Stress and Strain Tensor Summary Page 5

pressive? It turns out that there is. The resultant stresses are called the principal
stresses, the planes on which they occur are the principal planes, and the directions
of the resultant force components are the principal directions or principal axes. If we

call the principal stresses  1 ,  2 , and  3 , then the problem appears as: Are there

values of  for which

 xx    xy  xz
Det  xy  yy    yz  0?
 xz  yz  zz  

The principal stresses are the eigenvalues and the principal directions are the eigen-
vectors. The eigenvalue problem can be rewritten in terms of the three invariants as

3  I 12  I 2  I 3  0 .

For any stress tensor, three real (but possibly not distinct) roots will result.
The Von Mises yielding criterion.
In a complex stress field it is not easy to determine if the stress has exceeded the
yield stress in the body. Von Mises proposed the following criterion: Yielding occurs
when the second invariant of the stress deviator exceeds some critical value.
The stress deviator is the stress tensor with the hydrostatic component removed, i.e.,

 xx  h  xy  xz
 ijD   ij  hI 
 xx   yy   zz .
where h  -----------------------------------------
 xy  yy  h  yz -
3
 xz  yz  zz  h

The second invariant of  ijD is

(  xx   yy ) 2  (  yy   zz ) 2  (  zz   xx ) 2  6 (  xy 2  2  2 )
I 2  ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
yz xz ,
-
6

which must be greater than some constant k 2 for yielding to occur. In terms of the

yield stress,  y , the criterion is

E106 Stress and Strain Tensor Summary Page 6

 (  xx   yy ) 2  (  yy   zz ) 2  (  zz   xx ) 2  6 (  xy 2  2  2 )
yz xz
1/2
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
-  y
2

for yielding to occur.

The Strain Tensor.
Normal strain is the change in length in a given direction divided by the initial
length in that direction. Shear strain is the complement of the angle between two
initially perpendicular line segments. If you apply a force to a solid object you may
end up simultaneously translating, rotating and deforming the object. The vector
function which describes the difference between the initial position and the final
position of each point in the object is

u( x, y, z)
u( x, y, z)  v( x, y, z) .
w( x, y, z)

If we take the gradient of u we end up with a tensor

u u u
x  y z
u  v v v .
x  y z
w w w
x  y z

T
For small strains we find that u  u  2 where

 xx  xy  xz
   kl   yx  yy  yz .
 zx  zy  zz

The diagonal terms are the normal strains in the x, y, and z directions respectively.
The off-diagonal terms are equal to one-half of the engineering shear strain, e.g.,

 xy 
xy  2 . In terms of u ,

E106 Stress and Strain Tensor Summary Page 7

 1 1 1
 xy  --- u  v ,  xz  --- u  w , and  yz  --- v  w .
2  y x 2 z x 2 z  y

As was the case with stress  yx   xy ,  yz   zy , and  xz   zx . We can determine

the strains in a rotated coordinate system in the same way as for stresses. We can
T T
transform the coordinates, x y z into the coordinates, x y z by use of a

1 T
transformation matrix, Q , where Q has the property Q  Q . To get all of the

T
elements of the strain tensor in the new coordinate system, [  ]  Q [  ]Q .

Relationship between stress and strain.

Every member of  kl will cause a corresponding stress in  ij . The relationship can

be written as  ij  C ijkl  kl . Writing out the first term explicitly should suffice to

explain the notation.

 xx  C xxxx  xx  C xxxy  xy  C xxxz  xz

 C xxyx  yx  C xxyy  yy  C xxyz  yz
 C xxzx  zx  C xxzy  zy  C xxzz  zz
.

Fortunately only 21 of the 81 C ijkl -terms are unique. To simplify the notation, the

stress and strain tensors are rewritten as vectors. The simplified notation is known
as contracted notation. First the off-diagonal strain terms are converted to engineer-
ing shear strains.

 xx 2  xy 2  xz  xx
xy
xz
2  yx  yy 2  yz 
yx  yy
yz .
2  zx 2  zy  zz
zx
zy  zz

The resulting matrix is no longer a tensor because it doesn’t follow the coordinate-
transformation rules. Then the elements are renumbered.

E106 Stress and Strain Tensor Summary Page 8

  xx  xy  xz 1 6 5  xx
xy
xz 1 6 5
 xy  yy  yz   6  2  4 ,
yx  yy
yz   6  2  4 .
 xz  yz  zz 5 4 3
zx
zy  zz 5 4 3

Then the matrices are written as vectors,

1  xx 1  xx
2  yy 2  yy
1 6 5 1 6 5
3  zz 3  zz
6 2 4 ⇒  , 6 2 4 ⇒  .
4  yz 4
yz
5 4 3 5 4 3
5  xz 5
xz
6  xy 6
xy

Finally the relationships between the stress vector and the strain vector is
expressed.

1 Q 11 Q 12 Q 13 Q 14 Q 15 Q 16  1
2 Q 12 Q 22 Q 23 Q 24 Q 25 Q 26  2
3 Q 13 Q 23 Q 33 Q 34 Q 35 Q 36  3
 .
4 Q 14 Q 24 Q 34 Q 44 Q 45 Q 46  4
5 Q 15 Q 25 Q 35 Q 45 Q 55 Q 56  5
6 Q 16 Q 26 Q 36 Q 46 Q 56 Q 66  6

The materials-property matrix with all of the Q’s is known as the stiffness matrix.

Unfortunately Q is used for both the stiffness matrix and the coordinate transfor-

mation matrix. Don’t get them confused. The stiffness matrix is used when all of the
strains are known and the values of the stresses are to be determined. In the more
common case of the stresses being known and the strains to be determined, the

inverse of the stiffness matrix, called the compliance matrix, S , must be used.

The relationship between Q and S is that S  Q 1 . There are a number of simpli-

fied cases for the stiffness and compliance matrices.

E106 Stress and Strain Tensor Summary Page 9

 1 S 11 S 12 S 13 S 14 S 15 S 16  1
2 S 12 S 22 S 23 S 24 S 25 S 26  2
3 S 13 S 23 S 33 S 34 S 35 S 36  3
 .
4 S 14 S 24 S 34 S 44 S 45 S 46  4
5 S 15 S 25 S 35 S 45 S 55 S 56  5
6 S 16 S 26 S 36 S 46 S 56 S 66  6

Linear elastic isotropic materials:

The simplest materials are ones in which the properties do not vary with direction,
or linear elastic isotropic materials. In a linear elastic isotropic material character-

ized by Young’s modulus, E , Poisson’s ratio,  , and the shear modulus,

E
G  ---------------------- , the following relationships between the stress tensor and the strain
2(1  )

tensor hold:

1
 xx  ---- [  xx   (  yy   zz ) ] ,
E

1
 yy  ---- [  yy   (  xx   zz ) ] ,
E

and

1
 zz  ---- [  zz   (  xx   yy ) ] ,
E

also

xy  2  xy   xy ,
--------
G

 xz ,

xz  2  xz  -------
-
G

and

 yz

yz  2  yz  --------
G

E106 Stress and Strain Tensor Summary Page 10

It is left to the reader (usually on an exam) to wade through the alphabet soup and
determine the stiffness and compliance matrices for a linear elastic isotropic mate-
rial.
Linear elastic orthotropic materials:
For an orthotropic material (one in which the properties in the y- and z-directions are
the same but different in the x-direction, such as a composite material with the fibers
all oriented in the x-direction) the stiffness matrix has the form

Q 11 Q 12 Q 13 0 0 0
Q 12 Q 22 Q 23 0 0 0
Q 13 Q 23 Q 33 0 0 0
Q  ,
0 0 0 Q 44 0 0
0 0 0 0 Q 55 0
0 0 0 0 0 Q 66

and the compliance matrix has the form

S 11 S 12 S 13 0 0 0
S 12 S 22 S 23 0 0 0
S 13 S 23 S 33 0 0 0
S  .
0 0 0 S 44 0 0
0 0 0 0 S 55 0
0 0 0 0 0 S 66

Since the properties in the y and z directions are equal, the two-dimensional x,y-case
is often considered when determining materials properties. In this case the stiffness-
matrix relationship becomes

  xx  Q 11 Q 12 0   xx 
   
  yy   Q 12 Q 22 0   yy 
   
  xy  0 0 Q 66 
xy 

and the compliance-matrix relationship becomes

E106 Stress and Strain Tensor Summary Page 11

  xx  S 11 S 12 0   xx 
   
  yy   S 12 S 22 0   yy .
   

xy  0 0 S 66   xy 

First a uniaxial tensile test is performed in the x-direction.  xx is applied and  xx ,

and  yy are measured. As always in a uniaxial tensile test,  xx  E cl  xx , and

 yy
 lt   -------
- , where E cl is the elastic modulus in the longitudinal or x direction, and
 xx

 lt is Poisson’s ratio for stress in the longitudinal direction and strain in the trans-

verse direction. By comparison with the compliance matrix, it is seen that

1 , and S    lt .
S 11  -------
- 12 --------
E cl E cl

Next a uniaxial tensile test is performed in the y-direction.  yy is applied and  xx ,

and  yy are measured. As before,

 xx , where E is the elastic modulus in the transverse

 yy  E ct  yy , and  tl   -------
- ct
 yy

or y direction, and  tl is Poisson’s ratio for stress in the transverse direction and

strain in the longitudinal direction. By comparison with the compliance matrix, it is

seen that

 tl .
1 , and S   -------
S 22  -------
- 12 -
E ct E ct

A final test with pure shear gives the relationship

1.
S 66  ---
-
G

The interested reader should be able to determine the relationship between  lt and

E106 Stress and Strain Tensor Summary Page 12

  lt and S    tl .
 tl from the relationships S 12   -------
- 12 --------
E cl E ct

The structure of Q is not nearly so simply related to E cl , E ct ,  lt , and  tl as is S ,

but, as before, it can be determined from SQ  I , or S  Q 1 .

E106 Stress and Strain Tensor Summary Page 13