ISE Tensor Notation p218-20
ISE Tensor Notation p218-20
ISE Tensor Notation p218-20
The equations of this book, and in many others, are written in Cartesian tensor notation. The reason
is that most equations get a simpler outlook in this way. We start out with a three-dimensional
Cartesian coordinate system:
x3
x2
x1
A vector has three components, for instance a space vector and a velocity vector:
Vector form: x, component form: [x 1 , x 2 , x 3 ] or [x, y, z].
Vector form: u, component form: [u 1 , u 2 , u 3 ] or [u, v, w] or [u x , u y , u z ].
In Cartesian tensor form, a vector (1st order tensor) is denoted with one index, for instance
x i with i = 1, 2 or 3; x i = [x 1 , x 2 , x 3 ],
u i with i = 1, 2 or 3; u i = [u 1 , u 2 , u 3 ].
The stress tensor is a 2nd order tensor and get two indices:
τi j with i = 1, 2 or 3 and j = 1, 2 or 3.
The indices are chosen arbitrarily, so that u i is the same as u j or u n – that is, provided that the
indices i , j and n are not used elsewhere in the term in question. Similarly, τ i j is the same as τik ,
τ j k or τ pn . Each of the indices has the value 1, 2 or 3.
Example: In general, τ 13 = τ23 . These are two (of nine) individual components of the tensor.
Nevertheless, we have τ i j = τik = τ pn . In the latter instance, each of the three symbols represents
the entire tensor (all the 9 components), and it is one and the same tensor. If we in addition say that
(i, j ) = (1, 3) and ( p, n) = (2, 3), it is then obvious that τ i j = τ pn . In that case, we are back to
talking about individual components of the tensor.
When two indices in a tensor or a term are the same, we should sum over these indices from 1 to 3,
for instance
3
u i u i should be read as u i u i = u1u 1 + u 2u 2 + u 3u 3,
i=1
3
∂u i ∂u i ∂u 1 ∂u 2 ∂u 3
should be read as = + + ,
∂ xi ∂ xi ∂ x1 ∂ x2 ∂ x3
i=1
3
∂u i ∂u i ∂u i ∂u i ∂u i
uj should be read as uj = u1 + u2 + u3 .
∂x j ∂x j ∂ x1 ∂ x2 ∂ x3
j =1
The expressions to the right are unequal since one term has one index that is repeated, and hence
should be summed, whereas the other have two different indices and should not be summed. The for-
mer represent one single quantity each, which is the sum of three terms. Each of the latter represent
9 different combinations of the two indices, that is, 9 different quantities.
In a term of an equation it does not matter whether it reads ∂u i /∂ x i or ∂u n /∂ x n , that is, provided the
indices i and n are not used otherwise in the term.
Some equations contain terms with two or more pairs of repeated indices. Then you go on summing.
The product u i u i has 3 terms that should be summed, τ i j (∂u i /∂ x j ) has 32 =9 terms, ai j b j k cik has
220 Appendix A The fundamental equations – and something more
Two special tensors are the Kronecker delta δ i j and the permutation tensor i j k :
⎡ ⎤
0 1 0 0
1 when i = j
δi j = or δi j = ⎣ 0 1 0 ⎦ , (A.107)
0 when i = j
0 0 1
⎧
⎨ 1 when (i, j, k) is (1,2,3), (2,3,1) or (3,1,2) ,
i j k = −1 when (i, j, k) is (3,2,1), (2,1,3) or (1,3,2) , (A.108)
⎩ 0 when i = j and/or i = k and/or j = k.
For two-dimensional cases, the indices get the values 1 or 2, rather than 1, 2 or 3. For one-
dimensional cases all indices get the value 1.
This section is on Cartesian tensors. It is common to present the equations on Cartesian form, also
when using non-Cartesian coordinate systems.
Tensor analysis, both Cartesian and general, is further described in, for instance, Tyldesley (1975),
Aris (1962), Irgens (1982) and Sokolnikoff (1964). The former two are the easier to digest, and are
also more related to fluid mechanics, whereas the latter two have a more general aim.
Certain forms of the fundamental equations are presented in cylinder coordinates (polar coordinates)
and in spherical coordinates by Bird, Stewart and Lightfoot (1960:83,317) (also reproduced by Kuo,
1986:175–201).