EN0175

09 / 21 / 06

We continue on the mathematical background. Base tensors: ei e j (dyadic form)

1 0 0 0 1 0 0 0 1 v v v v v v e1 e1 = 0 0 0 , e1 e2 = 0 0 0 , e1 e3 = 0 0 0 0 0 0 0 0 0 0 0 0
Note that: ei e j e j ei , ei e j e j ei

(v

Rules of operation:

v v v (e e )e
i j

v v v v = ei (e j ek ) = jk ei

v v v v v v v ek (ei e j ) = (ek ei ) e j = ki e j
Tensor product of vectors (also called tensor dyad):

a1b1 a1b2 v v v v v v a b = (ai ei ) (b j e j ) = ai b j (ei e j ) = a2b1 a2b2 a3b1 a3b2 b1 a3 ] b2 = a1b1 + a2b2 + a3b3 b3 a1b1 b3 ] = a2b1 a3b1
v

a1b3 a2b3 a3b3

We can compare the matrix forms of different vector products:

v v a b = [a1 a2

a1 v v a b = a2 [b1 b2 a3
v

a1b2 a2b2 a3b2

a1b3 a2b3 a3b3

In general, T = Tij ei e j

0 0 1 0 1 0 1 0 0 T11 T12 T13 T T 21 22 T23 = T11 0 0 0 + T12 0 0 0 + T13 0 0 0 + L 0 0 0 0 0 0 0 0 0 T31 T32 T33
v v Tij = ei T e j
For example,

EN0175

09 / 21 / 06

T11 T12 T13 1 T11 = [1 0 0] T21 T22 T23 0 0 T31 T32 T33
One can also verify this using index notation:

v v v v v v v v Tij = ei (T pq e p eq )e j = T pq (ei e p )(eq e j ) = T pq ip qj = Tij

This definition is very useful in practice in deriving the components of a tensor from its physical descriptions.

3
Example 1: Find the tensor R that reflects any vectors with respect

2 1

to the 2-3 plane.

Solution: According to the definition of reflection, we have

v v v v v v R e1 = e1 , R e2 = e2 , R e3 = e3
Therefore:

1 0 0 v v Rij = ei R e j = 0 1 0 0 0 1

3
Example 2: Find the tensor Q that rotates any vector about the x3

2
axis by 90 o clockwise.

1
v v v v v v Q e1 = e2 , Q e2 = e1 , Q e3 = e3
Therefore:

0 1 0 v v Qij = ei Q e j = 1 0 0 0 0 1
Symmetric tensors:

EN0175

09 / 21 / 06

T11 T12 T13 T T = T12 T22 T23 = T T13 T23 T33

Antisymmetric tensors:

0 A= A12 A13

A12 0 A23

A13 T A23 = A 0

An arbitrary tensor P can be formed into a sum of a symmetric tensor and an antisymmetric tensor:

P=

P+P PP S A + =P +P 2 2
T T

Eigenvalues of a tensor:

v v T u =u
where u is called the eigenvectors of T .

is the eigenvalue.

det(T I ) = 0
For a second rank tensor in 3 dimension, there are 3 eigenvalues and 3 eigenvectors. Tensor transformation under change of coordinates:

v e3

v' e2
v e1
In different coordinates,

v' e1
v e2

v' e3

EN0175

09 / 21 / 06

T11 T12 T13 v v T = Tij ei e j = T21 T22 T23 T31 T32 T33
' ' ' T11 T12 T13 v' ' ' v' ' ' T = T pq e p eq = T21 T22 T23 ' ' ' T31 T32 T33

How are Tij and Tij related?

'

v v v ' v' T = Tij ei e j = Tpq e p eq' v' v v' v v v v' v v' v ' Tpq = ep T eq' = e p (Tij ei e j ) eq' = Tij e p ei eq e j
Example: 2D Cartesian coordinate and Polar coordinate.

)(

v e2
r

v e

v er

v e1

v v v v v v v v T = T11e1 e1 + T12 e1 e2 + T21e2 e1 + T22 e2 e2 v v v v v v v v = Trr er er + Tr er e + T r e er + T e e

v v v v v v v v v v v v T11 = e1 T e1 = e1 (Trr er er + Tr er e + T r e er + T e e ) e1 = Trr cos 2 Tr cos sin T r cos sin + T sin 2

T12 = Trr cos sin + Tr cos 2 T r sin 2 T cos sin

Tensor equations:

EN0175

09 / 21 / 06

Example: Generalized Hookes Law

ij =

1 + ij kk ij E E

Inverting this equation by letting j i ,

ii =
i.e.

1 + 1 2 ii kk ii = ii E E E

kk =

E kk 1 2

Substituting

kk into the original equation,

ij =
1 + E ij kk ij E E 1 2

1 + ij = ij + kk ij E 1 2

ij =

E E ij + (1 + )(1 2 ) kk ij 1 + = 2 ij + kk ij

E , : Youngs modulus and Poissons ratio

E E , = are called Lam constants. (1 + )(1 2 ) 2(1 + )

Divergence theorem:

v ( v )dv =
S

v v v n dS

Generalization:

EN0175

09 / 21 / 06

()

,i

dv =

() ni dS

()

can be any vector or tensor field.