Lecture2 (Vectors and Tensors)
Lecture2 (Vectors and Tensors)
The physical quantities encountered in fluid mechanics can be classified into three
classes:
(a) Scalar - a quantity having magnitude but no direction, such as pressure, density,
viscosity, temperature, length, mass, volume and time;
(b) Vector - (1st rank tensor) a quantity having magnitude and deriction, such as
velocity, acceleration, displacement, linear momentum and force, and
(c) Tensor- (2nd rank tensor) a quantity having magnitude and two directions (e.g.
momentum flux, stress, rate of strain and vorticity).
Vector Addition
Given two arbitrary vectors a and b, by a+b we mean the vector formed by
connecting the tail of a to the head of b when b is moved such that its tail coincides with
head of a . A brief review of vector addition and multiplication can be found in
Calculus, Thomas.
b
a
a
a+ b
b
1. Vector Multiplication
Given two arbitrary vectors a and b, there are three types of vector products are defined:
Notation Result Definition
Dot product a .b Scalar abcos
Cross product a×b Vector ab|sin|n
Dyadic product ab Tensor -
where n is a unit vector which is normal to both a and b.
33
The sense of n is determined from the "right-hand rule". In the above definition, we
denote the magnitude (or length) of vector a by the scalar a. Using the following
notation:
( ) = scalar = lightface Italic such us a
[ ] = vector = boldface Roman such as a
{ } = tensor = boldface Greek such as
Definition of Dyadic product
The word "dyad" comes from Greek: "dy" means two while "ad" means adjacent. Thus
the name dyad refers to the way in which this product is denoted: the two vectors are
written adjacent to one another with no space or other operator in between.
There is no geometrical picture that I can draw which will explain what a dyadic
product. It's best to think of the dyadic product as a purely mathematical abstraction
having some very useful properties:
Dyadic Product ab - that mathematical entity which satisfies the following properties
(where a, b, v, and w are any four vectors)
1. ab.v=a(b.v) [which has the direction of a; note that ba.v=b(a.v) which has the
dircetion of b]. Thus ab ≠ ba.
2. v.ab=(v.a)b [Thus v.ab ≠ ab.v]
3. ab × v=a(b×v) which is another dyad
4. v×ab=(v×a)b
5. ab:vw=(a.w)(b.v) which is sometimes known as the inner-outer product or the
double-out product.
6. a(v+w)=av+aw (distribution for addition)
7. (v+w)a=va+wa
8. (s+t)ab=sab+tab (distribution for scalar multiplication)
9. sab = (sa)b=a(sb)
34
2. Decomposition into Scalar Components (System of Coordinates)
A coordinate system in the three-dimensional space is defined by choosing a set of three
linearly independent vectors, B={e1, e2, e3}, if none can be expressed as a linear
combination of the other two (e.g. i, j, and k). The set B is a basis of the three-
dimensional space, i.e., each vector v of this space is uniquely written as a linear
combination of this basis:
v v 1e1 v 2e 2 v 3e 3
where the vi are called the scalar components of v. In most cases, the vectors e1, e2 and
e3 are unit vectors. In the three coordinate systems, i.e., Cartesian, cylindrical and
spherical coordinates, the three vectors are, in addition, orthogonal. Hence, in all these
systems, the basis B={e1, e2, e3} is orthonormal:
1 if i j
e i .e j ij
0 if i j
where ij is called the Kronecker delta.
35
Figure (1) Cartesian coordinate (x,y,z)
The cylindrical and spherical polar coordinates are the two most important
orthogonal curvilinear coordinate systems. The cylindrical polar coordinates
(r, θ, z),with r 0, 0 2 and z are shown in fig (2)
36
The spherical polar coordinates (r,,) with r 0, 0 2 and 0 2
together with Cartesian coordinate with the same origin, are in figure (2).
Example. Show that the basis B={er, eθ, ez} of the cylindrical system is orthonormal.
Sol. since i.i j. j k.k 1 and i. j j.k k.i 0 , we obtain;
er .e r cos i sin j . cos i sin j cos2 sin 2 1
e .e
e z .e z k.k 1
er .e cos i sin j . sin i cos j 0
er .e z
e .e z
37
H.W. The position vector r defines the position of a point in space, with respect to
coordinate system. In Cartesian coordinate, r = x i + y j + z k, show that in
cylindrical coordinates, the position vector is given by r = r er + z ez, and in
spherical coordinate, r = r er.
In the following subsections, we will make use of the vector differential operator nabla
(or del), ∇ . In Cartesian coordinates, ∇ is defined by
i j k
x y z
The gradient of a scalar field f (x, y, z) is a vector field defined by
f f f
f i j k
x y z
The divergence of a vector field v(x, y, z) is a scalar field defined by
v x v y v z
.v
x y z
3. Tensors
In the previous sections, two kinds of products that can be formed with any two
unit basis vectors were defined, i.e. the dot product, ei · ej , and the cross product, ei × ej.
A third kind of product is the dyadic product, eiej , also referred to as a unit dyad. The
unit dyad eiej represents an ordered pair of coordinate directions. The nine possible unit
dyads, {e1e1, e1e2, e1e3, e2e1, e2e2, e2e3, e3e1, e3e2, e3e3}
A second-order tensor, τ , can thus be written as a linear combination of the unit dyads:
3 3
e e
i 1 j1
ij i j
where the scalars τij are referred to as the components of the tensor τ . Similarly, a third-
order tensor can be defined as the linear combination of all possible unit triads eiejek, etc.
Scalars can be viewed as zero-order tensors, and vectors as first-order tensors.
A tensor, τ , can be represented by means of a square matrix as
38
τ 11 12 13 e1
τ e1 , e 2 , e 3 21 22 23 e 2
31 32 33 e 3
and often simply by the matrix of its components,
τ 11 12 13
τ 21 22 23
31 32 33
Note that the equality sign “=” is loosely used, since τ is a tensor and not a matrix. For a
complete description of a tensor by means of matrix, the basis {e1, e2, e3} should be
provided.
The unit (or identity) tensor, I, is defined by
3 3
I e e
i 1 j1
ij i j e1e1 e 2e 2 e 3e 3
Each diagonal component of the matrix form of I is unity and the non-diagonal
components are zero:
1 0 0
I 0 1 0
0 0 1
The sum of two tensors, σ and τ , is the tensor whose components are the sums of the
corresponding components of the two tensors:
3 3 3 3 3 3
σ τ e e e e
i 1 j1
ij i j
i 1 j1
ij i j
i 1 j1
ij ij e ie j
The product of a tensor, τ , and a scalar, m, is the tensor whose components are equal to
the components of τ multiplied by m:
3 3 3 3
mτ m
ije ie j
m e e ij i j
i 1 j1 i 1 j1
T
The transpose, τ , of a tensor τ is defined by
39
3 3
τ
T
i 1 j1
ee
ji i j
The matrix form of τT is obtained by interchanging the rows and columns of the matrix
form of τ :
τ 11 21 31
τ T 12 22 32
13 23 33
If τT=τ , i.e., if τ is equal to its transpose, the tensor τ is said to be symmetric. If τT=−τ ,
the tensor τ is said to be antisymmetric (or skew symmetric).
The dyadic product of two vectors a and b can easily be constructed as follows:
3
3 3 3
ab
aie i
b je j
a b e e i j i j
i 1 j1 i 1 j1
0 0 0
e 2e 3 0 0 1
0 0 0
The most important operations involving unit dyads are the following:
40
(i) The single-dot product (or tensor product) of two unit dyads is a tensor defined by
e e e e e e
i j k l i j e k e l jk e ie l
This operation is not commutative
(ii) The double-dot product (or scalar product or inner product) of two unit dyads is a
scalar defined by;
e e : e e e e e
i j k l i l j e k il jk
It is easily seen that this operation is commutative.
(iii) The dot product of a unit dyad and a unit vector is a vector defined by
e e e
i j k e i e j e k ik e i
or
e i e je k e i e j e k ije k
Obviously, this operation is not commutative.
Operations involving tensors are easily performed by expanding the tensors into
components with respect to a given basis and using the elementary unit dyad operations.
The most important operations involving tensors are summarized below.
1. The single-dot product of two tensors (Tensor product)
If σ and τ are tensors, then
3 3
3 3
3 3 3 3
σ τ
i 1 j1
ije i e j
k 1
1
k e k e
e e e e
i 1 j1 k 1 1
ij k i j k
3 3 3 3 3 3 3
i 1 j1 k 1 1
ij k jk e i e
i 1 j1 1
ij j i ee
3 3
3
σ τ
i 1 1
j1
ij j e i e
The operation is not commutative. It is easily verified that σ I I σ σ
2. The double-dot product of two tensors (Scalar product)
41
3 3
σ :τ e e
i 1 j1
ij ij i j
3 3
3 3 3 3
σ a
i 1 j1
ije i e j ak e k
k 1
a e e e
i 1 j1 k 1
ij k i j k
3 3 3 3 3
a
i 1 j1 k 1
e
ij k ik k a e
i 1 j1
ij j ij i
3
3
σ a
i 1 j1
ijaj e i
Similarly, we find that
3
3
a σ
i 1
j1
jiaj e i
The vectors σ · a and a · σ are not, in general, equal.
u e
i 1
i i and e e
i 1 j1
ij i j
τ 11 12 13
τ 21 22 23 for τ .
31 32 33
42
Very frequently, in the literature, use is made of the index notation and the so-
called Einstein’s summation convention, in order to simplify expressions involving
vector and tensor operations by omitting the summation symbols.
In index notation, a vector v is represented as
3
vi v e
i 1
i i v
A tensor τ is represented as
3 3
ij e e
i 1 j1
ij i j τ
x i
i 1
x i
ei
x
i
y
j
z
k
where xi is the general Cartesian coordinate taking on the values of x, y and z. The
unit tensor I is represented by Kronecker’s delta:
3 3
ij e e
i 1 j1
ij i j I
It is evident that an explicit statement must be made when the tensor τij is to be
distinguished from its (i, j) element.
With Einstein’s summation convention, if an index appears twice in an expression,
then summation is implied with respect to the repeated index, over the range of that
index. The number of the free indices, i.e., the indices that appear only once, is the
number of directions associated with an expression; it thus determines whether an
expression is a scalar, a vector or a tensor. In the following expressions, there are no free
indices, and thus these are scalars:
3
u iv i u vi 1
i i uv
3 3
ii
i 1 j1
ii tr τ trace for
43
u i u i u x u y u z
3
x i
i 1
x i x
y
z
u
2f 2f 2f 2f 2f 2f
3
x i x i
or 2
x i i 1
2 2 2 2f
x i x
2
y z
2
where ∇ is the Laplacian operator to be discussed in more detail in Section (5). In the
following expression, there are two sets of double indices, and summation must be
performed over both sets:
3 3
ij ji
i 1 j1
ij ii σ:τ
3 3
3
ijku iv j
k 1 i 1 j1
ijku iv j e k u v
f f f f f
3
x i i 1 x i
ei
x
i
y
j
z
k f
3 3
jiv j
i 1
iiv j e i τ v
j1
Finally, the following quantities, having two free indices, are tensors:
3 3
u iv j u v e e
i 1 j1
i j i j uv
3
3 3
ik kj
i 1 j1 k 1
ik kj e i e j σ τ
u j 3 3
u j
x i i 1 j1 x i
e ie j u
44
5. Differential Operators
The nabla operator ∇ , already encountered in previous sections, is a differential
operator. In a Cartesian system of coordinates (x1, x2, x3), defined by the orthonormal
basis (e1, e2, e3),
3
x 1
e1
x 2
e2
x 3
e3 i 1
x i
ei
3
3
f f f f
f
i 1
ei f
x i
i 1
x i
ei
x 1
e1
x 2
e2
x 3
e3
u j
3 3 3 3
u
ei
x i u je j
x e ie j
i 1 j1 i 1 i 1 i
u i u u u
3 3 3
u
ei
x i u je j
x i
ij 1 2 3
x 1 x 2 x 3
i 1 j1 i 1
45
∇ · u measures changes in magnitude, or flux through a point. If u is the velocity,
then ∇ · u measures the rate of volume expansion per unit volume; hence, it is zero for
incompressible fluids.
The curl or rotation of a differentiable vector field u, denoted by ∇ ×u or curl u or
rot u, is a vector field:
e1 e2 e3
3
3
u
ei
x i u je j
x 1 x 2 x 3
i 1 j1
u1 u2 u3
or
u u u u u u
u 3 2 e1 1 3 e 2 2 1 e 3
x 2 x 3 x 3 x 1 x 1 x 2
The field ∇ ×u is often called the vorticity (or chirality) of u.
The divergence of a differentiable tensor field τ is a vector field:
3 3 3 3 3
ij
τ ek
k 1 x k
ije ie j
x ej
i 1 j1 i 1 j1 i
i j k
r r 0
x y z
x y z
46
2
Other useful operators involving the nabla operator are the Laplace operator ∇
and the operator u · ∇ , where u is a vector field. The Laplacian of a scalar f with
continuous second partial derivatives is defined as the divergence of the gradient:
2f 2f 2f
f f 2 2 2
2
x 1 x 2 x 3
2 2 2
i.e., 2 2 2
2
x 1 x 2 x 3
u u1e1 u 2e 2 u 3e 3 e1 e2 e3
x 1 x 2 x 3
u u1 u2 u3
x 1 x 2 x 3
The above expressions are valid only for Cartesian coordinate systems. In
curvilinear coordinate systems, the basis vectors are not constant and the forms of ∇ are
quite different. Notice that gradient always raises the order by one (the gradient of a
scalar is a vector, the gradient of a vector is a tensor and so on), while divergence
reduces the order of a quantity by one.
For any scalar function f with continuous second partial derivatives, the curl of the
gradient is zero,
f 0 (1)
For any vector function u with continuous second partial derivatives, the divergence
of the curl is zero,
u 0 (2)
Equations (1) and (2) are valid independently of the coordinate system.
In fluid mechanics, the vorticity ω of the velocity vector u is defined as the curl
of u, u
47
Example.
(a) Express the nabla operator
i j k in cylindrical polar coordinates.
x y z
(b) Determine ∇ c and ∇ · u, where c is a scalar and u is a vector.
(c) Derive the operator u · ∇ and the dyadic product ∇ u in cylindrical polar coordinates.
Sol.
(a) From Table (1), we have:
i = cos θ er − sin θ eθ , j = sin θ er + cos θ eθ and k = ez
Therefore, we just need to convert the derivatives with respect to x, y and z into
derivatives with respect to r, θ and z. Using the chain rule, we get:
r sin
cos
x x r x r r
cos
sin
y r r
z z
Substituting now into nabla operator, gives
sin
cos cos e r sin e
r r
cos
sin sin e r cos e ez
r r z
2 2
After some simplifications and using the trigonometric identity sin θ+sin θ =1, we get
1
er e ez
r r z
(b) The gradient of the scalar c is given by;
c 1 c c
c er e ez
r r z
48
For the divergence of the vector u, we have;
1
u er e e z u r e r u e u ze z
r r z
Noting that the only nonzero spatial derivatives of the unit vectors are
e r e
e and e r we obtain;
u r 1 er u e u
u u r e u e z
r r z
u 1 u 1 u
r u r e u e r e z
r r r z
u 1 u u r u z
r
r r r z
1 1 u u z
u ru r
r r r z
1
(c) u u r er u e u ze z er e ez
r r z
u
u ur uz
r r z
Finally, for the dyadic product ∇ u, we have;
1
u e r e e z u r e r u e u ze z
r r z
u u u
r e r e r e r e z e r e z
r r r
1 u r 1 e 1 u 1 e 1 u z
e e r u r r e e e u e e e z
r r r r r
u u u
r e z e r e z e z e z e z
z z z
49
u r u u
u e r e r e r e z e r e z
r r r
1 u 1 u 1 u z
r u e er u r e e e e z
r r r
u u u
r e z e r e z e z e z e z
z z z
50