You are on page 1of 13

Lecture 02: Coordinate

Transformation of Tensors
Jayadeep U. B.
Dept. of Mechanical Engg., NIT Calicut.
Introduction
Coordinate systems are chosen by people to study a given system,
hence they are completely artificial to the system.
Such coordinate systems may not correspond to the critical
directions. Further, the critical directions generally change from
point to point.
Additionally, different idealizations like beams, rods etc. require
local coordinate systems, which could be different for different
parts of the system. A global coordinate system is required to
combine the parameters from all such local coordinate systems.
Hence, we need to study the behavior of tensorial quantities under
transformation of coordinate systems.

2 ME6302 Metal Forming


Review of Index Notation
Change in coordinate system from x-y-z to x1-x2-x3.
Representing a tensor of any order by its component.
Summation Convention: Repeated indices (dummy indices) within
a term imply summation over the range of their values.
Explicitly state it, when summation is not implied for the repeated
index.
No index shall repeat more than two times in a term, except
when summation is not implied and explicitly stated so.
Other indices (called free indices) should match in all terms.
Use of Kronecker delta, especially as a replacement operator.

3 ME6302 Metal Forming


Coordinate Transformations
By coordinate transformations, only rotations are considered as
any translation will change the point of consideration, which can
have a totally different variable value in a field.
We assume that tensor components are known in a coordinate
system, which we shall call as the original coordinate system.
The purpose coordinate transformation relations is to obtain the
tensor components in a new coordinate system.
It is assumed that the angle between any two coordinate directions
in the original and the new coordinate systems is known.
Zeroth order tensors (scalars) are not influenced by coordinate
transformations.

4 ME6302 Metal Forming


Transformation of Vectors
Original coordinate system: {x1, x2, x3}
with corresponding unit vectors
{ei}={e1, e2, e3}.
New (rotated) coordinate system: {x1,
x2, x3} with corresponding unit vectors
{ei}={e1, e2, e3}.
Define: Qij = cos ( xi, x j ) = ei e j
Hence we get: e1 = Q11e1 + Q12 e2 + Q13e3
e2 = Q21e1 + Q22 e2 + Q23e3
e3 = Q31e1 + Q32 e2 + Q33e3
Using index notation: ei = Qij e j
5 ME6302 Metal Forming
Transformation of Vectors contd.
Inverting this relation we get: ei = Q ji ej
An arbitrary vector can be written as: v = v1e1 + v2e2 + v3e3 = vi ei
= v1e1 + v2 e2 + v3e3 = viei
Substituting we get: v = vi Q ji ej
Since v = vj ej , we get vj = Qjivi or vi = Qij vj
Similarly, we get vi = Q ji vj

H.W.: Rotation matrix (or tensor) Q is an orthogonal tensor with


the property QQT=I. Show it from the above expressions, and
hence prove that the determinant of Q is 1.

6 ME6302 Metal Forming


Transformation of Cartesian Tensors
The case of transformation of vector components under
coordinate transformations can be generalized to Cartesian
tensors of any order:
a = a, zeroth order (scalar)
ai = Qip a p , first order (vector)
aij = Qip Q jq a pq , second order (matrix variable)
= Qip Q jq Qkr a pqr , third order tensor
aijk

n = Qip Q jq Qkr Qnu a pqru , n th order tensor
aijk
The last one is of nth order, assuming that there are n indices.

7 ME6302 Metal Forming


Isotropic Tensors
The transformation equations are very useful in determining
displacement, stress and strain in different coordinate directions.
There exists some special tensors, whose components do not
change under coordinate transformations. Such tensors are called
isotropic tensors.
All scalars can be thought to be isotropic, though the concept is
not of much use in case of scalars.
There are no isotropic vectors.
Isotropic tensors of (any) higher order exist; Kronecker delta
(forming components of the identity tensor) is an example.
Concept of isotropy plays crucial roles in material symmetry and
many other cases.

8 ME6302 Metal Forming


Principal Values and Directions
Here we consider symmetric second order tensors, which occur
commonly in TOE.
For such a tensor, there exists some special directions called
principal directions, for which:
An = n or Aij n j = ni
Clearly, this is an eigenvalue problem, and such directions are the
eigen vectors of the tensor.
Corresponding values of are called the eigenvalues or principal
values.
To give an example, in case of a stress tensor, principal directions
give the planes with only normal stresses, while the principal
values are the corresponding normal stresses.
9 ME6302 Metal Forming
Principal Values and Directions contd.
We can rewrite the system as: ( Aij ij ) n j = 0
Solutions exist only if det Aij ij = 0
Expanding, we get a cubic polynomial,
det Aij ij = 3 + I A 2 II A + III A = 0
where I A = Aii = A11 + A22 + A33

II A = ( Aii Ajj Aij Aij ) / 2 =


A11 A12 A22 A23 A11 A13
+ +
A21 A22 A32 A33 A31 A33
III A = det Aij

These three coefficients are called the fundamental invariants of A.


10 ME6302 Metal Forming
Principal Values and Directions contd.
This characteristic equation can be solved to obtain the principal values
and directions.
Depending on the principal values, three possibilities arise:
All three principal values are distinct, three unique principal directions are
obtained.
Two of the principal values are same, but the third one is distinct: there is a
unique principal direction corresponding to the distinct principal value, and
any direction in a plane perpendicular to this direction is a principal
direction.
All three principal values are same: any direction in the space is a principal
direction.
In any of these cases, it is possible to choose three mutually orthogonal
principal directions, leading to a principal coordinate system.
11 ME6302 Metal Forming
Principal Values and Directions contd.
With respect to the principal coordinate system, the second order
tensor becomes diagonal, with the principal values being the
diagonal elements.
The largest and smallest possible diagonal elements of the tensor
are given by the largest and smallest principal values.
When, all the principal values are same, the tensor is diagonal
with all diagonal elements same in all coordinate systems.
Using the principal values, the fundamental invariants can be
written as: I A = 1 + 2 + 3
II A = 12 + 2 3 + 13
III A = 12 3
12 ME6302 Metal Forming
References
Sadd, M.H., Elasticity: Theory, Applications and Numerics,
Academic Press.

13 ME6302 Metal Forming

You might also like