Cardiff School of Engineering

Autumn Semester 2020/21

EN2090 Engineering Mathematics 2

Divergence & Curl

Prof. Adrian Porch

Dr. Yue Zhou

[email protected]
 1
Aims of this week
❖ Divergence
• Understand the basic concept of divergence
• Understand divergence theorem and the physical interpretation
• Assess divergence of a vector field from its field lines
• Calculate the divergence of a vector field

❖ Curl

• Understand the basic concept of curl

• Understand Stokes’ theorem and the physical interpretation
• Assess curl of a vector field from its field lines
• Calculate the curl of a vector field
 2
2.5 Surface integrals involving vector fields
We often calculate the flow (usually called flux) of a vector field F
through a surface S. The flux depends on the orientation of F to the
surface, so we introduce a surface area vector S that is perpendicular
to that surface. We denote flux as the scalar  and evaluate it using

uniform F,  = F •S non-uniform F, S
 = F • dS

surface integral

The flux through a closed surface is S

 = F • dS

This must be linked to the generation or termination

of field lines of F within the volume V that encloses S
(i.e. the number of sources and sinks of F within V).
 3
2.6 The divergence of a vector field
We’ll start in Cartesian co-ordinates and consider a general vector
field F = (Fx , Fy , Fz ), where the vector components are themselves a
function of position ( x , y , z ) . The divergence of F is defined to be
   Fx Fy Fz
div F   • F =  , ,  • (Fx , Fy , Fz ) = + +
 x y z  x y z
Note that  • F is a scalar field, i.e. it is a scalar function of
position.
In cylindrical polar co-ordinates we write F = ( Fr , F , Fz ) and it
can be shown that at the point with position vector r = (r ,, z )
1 1 F Fz
•F = (rFr ) + +
r r r  z
At each point  • F is the same in any vector basis, i.e. is invariant.
Before calculating  • F we will try to develop its physical
meaning.
 4
2.7 The Divergence Theorem
We’ll assume this theorem without proof, although the proof itself is
straightforward. Consider some closed surface S which encloses a
volume V, within which is a vector field F is present.
The divergence theorem states that the volume
integral of  • F throughout V equals the
outward flux of F through the closed surface S
that bounds V,
i.e. V S
 • F dV = F • dS

The flux
S F • dS is simply the total number of field lines crossing S.

Some vector fields can be analysed in terms of field sources and

sinks. Field lines always diverge away from a source (where they
start) and converge towards a sink (where they stop). This must be
linked to  • F , as we will now demonstrate.
 5

S F • dS > 0 means a net outward flux of F through

S,
so there must be an excess of sources of field F in V.

S F • dS < 0 means a net inward flux of F through S,

so there must be an excess of sinks of field F in V.

S F • dS = 0 means no net flux of F ,

so this means either () equal

numbers of sources/sinks enclosed by
S, or (→) no sources or sinks enclosed.
Thus by letting the total volume enclosed V → 0, we can use the
scalar  • F to quantify the local density of field sources and sinks.
 6
2.8 Divergence theorem applied to the electric field

Applying the divergence theorem to the electric flux density D gives

V  • D dV = S D • dS
But Gauss’ law (itself derived from Coulomb’s law, the basis of all
electrostatics) states that the flux of D through any closed surface S
equals the free charge Q enclosed within the surface, i.e.

 D • dS = Q
S

If Q is distributed with a volume charge density  throughout V,

then
Q =   dV
V
→ S D • dS = V dV = V  • DdV → •D = 

 • D =  is the first of Maxwell’s equations (M1) and is known as

the differential form of Gauss’ law (integral form is shown above).
 7
2.9 The physical interpretation of Gauss’ law

 D • dS = 0 means that the closed surface S

S
+ encloses no free charge and the net outward
flux of D through the surface is zero (i.e. every
closed surface
line of D that enters the surface must leave it).

 D • dS  0 means that there is a finite free

S
+
charge density within S and so there must be a
finite net flux of D through the surface.

Hence, free charges are sources and sinks of D. A source of D is a

positive charge; at the position of the charge  • D =   0. A sink of
D is a negative charge, and at the position of charge  • D =   0 .

At a point in space at which there is no free charge,  • D = 0 .

 8
2.10 The divergence theorem applied to B

This states that   • B dV =  B • dS , and we’d expect  • B = m

V S
where  m is the “magnetic charge density”. However, up until now
nobody has been able to detect experimentally an isolated “magnetic
charge” (i.e. a magnetic monopole); people are still looking!

It is true that N poles act as sources of B and S poles act as sinks of B,

but N and S poles always occur in pairs. Therefore we conclude that
 m = 0 , with the consequence that
  • B dV =  B • dS = 0 →  • B = 0 always
V S
 • B = 0 is the second Maxwell equation (M2),
meaning that lines of B always form closed loops,
i.e. as many lines of B enter a closed surface as
leave it. If B started and stopped at points then
 • B  0 at these points, which never happens.
 9
2.11 Assessing divergence of a vector field from its field lines
Suppose that we are given the field line distribution of a vector field
A and we need to determine whether or not  • F = 0 at a point.

Clearly, we could look for the positions of sources (where  • F  0 )

and sinks (where  • F  0 ) of F; otherwise, the following “test
sphere” method can be used as a “divergence detector”.

We enclose the point of interest with a sphere of volume V, and we

apply the divergence theorem to the sphere in the limit when V → 0
1 

V

i.e.  • F dV = F • dS reduces to
S
 • F = lim 
V → 0 V

S
F • dS 



Hence, if our infinitessimally small “test sphere” has a non-zero flux

of F through its surface then  • F at the sphere’s centre is non-
zero; outward flux means that  • F  0 , inward flux means that
 • F  0.
 10
Worked example 2.4
Use the “test sphere” to find the points in the following vector fields
where the divergence is zero and also non-zero. (To be done in class).

1 1

0.5 0.5

0 0

0.5 0.5

1 1

1 0.5 0 0.5 1 1 0.5 0 0.5 1

M N M N
If the above plots represent D, at the point where  • D  0 there is a
free +charge. At all points where  • D = 0 there is no free charge.
 11
2.12 Calculating the divergence of a vector field
We always have the formula for  • F to rely on, illustrated below.
Worked example 2.5
(a) Calculate  • F at the point (2,−3, 1) when F = (x 2 y, y 3 + xz 2 , 2 xyz ).
(a) Derive a formula for  • F if F = − and  = 4 sin x cos 3z .
Fx Fy Fz
(a)  • F = + + = 2 xy + 3 y 2 + 2 xy = y (4 x + 3 y ) = 3
x y z
           2   2   2 
(b) Note  •  =  , ,  •  , ,  = 2 + 2 + 2   2 
 x y z   x y z  x y z
The scalar field  2  is called the Laplacian of ; for this example
 2  2  2
  = 2 + 2 + 2 = −4 sin x cos 3z − 36 sin x cos 3z = −40
2
x y z
The equation  2 = 0 is called Laplace’s equation and is very
important in Engineering; it is solved to find temperature
distributions, electric potential distributions, etc., etc.
 12
2.13 The curl of a vector field
In Cartesian co-ordinates, we define the curl of the vector field F as
the determinant
aˆ x aˆ y aˆ z
  
curl F    F =
x y z
Fx Fy Fz
 Fz Fy   Fx Fz   Fy Fx 
=  − aˆ x +  − aˆ y +  − aˆ z
 y z   z x   x y 

Fz Fy
i.e. x component of   F is − , etc. for y and z
y z
components
Thus, the curl of a vector field is also a vector field.

An important vector identity is  • (  F ) = 0 for any vector field F

(see worked example 2.7 for its proof).
In cylindrical polar co-ordinates we write F = (Fr , F , Fz ) and it
13

can be shown that at the point with position vector r = (r , , z )

aˆ r r aˆ  aˆ z
1   
curl F    F = 
r r  z
Fr rF Fz
 1 Fz F   Fr Fz   1  (rF ) 1 Fr 
= − aˆ r +  − aˆ  +  − aˆ z
 r  z   z r   r r r  
1 Fz F
i.e. the radial component of   F is − , etc.
r  z

As with  • F , the magnitude of   F at each point does not

depend on the vector basis in which it is represented (i.e.   F at
each point is invariant with respect to change of vector basis).
 14
2.14 Stokes’ Theorem
Suppose an unclosed surface S is bounded by a closed loop L.
Stokes’ theorem states that the
line integral of any vector field F
plane surface along the loop L bounding S
S, bounded equals the flux of the vector
by loop L
loop L   F through S, i.e.

L F • d = S (  F ) • dS
We’ll won’t prove Stokes’ law for this module, although the proof is
fairly straightforward.
The sense of circulation of the line integral
S around the loop L and the surface normal to S are
loop L related by the right hand rule, as shown opposite.
 15
2.15 Application of Stokes’ Theorem to the magnetic field H

We’ll now apply Stokes’ theorem to the vector field H, giving

 H • d =  (  H ) • dS
L S
But Ampère’s law (which itself is derived from the Biot-Savart law,
the basis of all magnetostatics) states that

 H • d = I   J • dS
L S
where J is the current density through S. Hence we conclude that

 (  H ) • dS =  J • dS →  H = J
S S

This is the differential form of Ampère’s law; it is half of Maxwell’s

fourth equation (“half”, since it is only valid for DC vector fields). At
all points where   H  0 there must be a finite current density J; if
  H = 0 at a point then we must have J = 0 at that point.
 16
2.16 Assessing the curl of a vector field from its field lines
The physical interpretation of   F is that   F tells us the amount
of circulation (or rotation) of the vector field F at a point. To find if
  F  0 at some point we place a tiny “test wheel” at the point.
Field causes c/w Field causes anti c/w
rotation,   F  0 rotation,   F  0
and points in (RHR) and points out (RHR)
More rigorously, we place an infinitessimally small “test loop” L of
surface area S to find   F at a point. Applying Stokes’ theorem
1 
L
S

F • d = (  F ) • dS → (  F )n = lim  F • d 
S → 0 S
 L


where (  F )n is the component of   F along the surface
normal to the area S, i.e. parallel to surface area vector S = Sa n .
The direction of   F is found by applying the right hand rule
(RHR) to the circulation of the line integral of F around the closed
loop L.
 17
Worked example 2.6
Find the points in the following vector fields where the curl is zero
and also non-zero. (This example will be done in class).
1 1

0.5 0.5

0 0

0.5 0.5

1 1

1 0.5 0 0.5 1 1 0.5 0 0.5 1

M N M N
If RH plot represents H, at the point where   H  0 there is a
current density J into the page; at points with   H = 0 then J = 0 .
 18
2.17 Calculating the curl of a vector field
Worked example 2.7
( )
Calculate the curl of F = 3xy, − 2 xz, 5 y 2 . Demonstrate that
 • (  F ) = 0 for this vector field and prove this result generally.
aˆ x aˆ y aˆ z
  
 F = = (10 y + 2 x, 0, − 2 z − 3 x )
x y z
3 xy − 2 xz 5 y 2

For this vector field  • (  F ) = 2 + 0 − 2 = 0 , but quite generally

aˆ x aˆ y aˆ z
    Fz Fy   Fx Fz   Fy Fx 
 F = =  − aˆ x +  − aˆ y +  − aˆ z
x y z  y z   z x   x y 
Fx Fy Fz
 2 Fz  Fy  2 Fx  2 Fz  Fy  2 Fx
2 2
  • (  F ) = − + − + − =0
xy xz yz yx zx zy
 19
2.18 Conservative vector fields have zero curl
In Section 2.4 we found that F is conservative if

L F • d = 0
Such conservative vector fields are generated from
the gradient of a scalar field, i.e. F =  f .
But using Stokes’ theorem L F • d = S (  F ) • dS →  F = 0

Hence, if   F = 0 for some vector field F then it must be

conservative. Equivalently, if F =  f then   F = 0 .
For the electric field E = −, we must have   E = 0 . This is
Maxwell’s 3rd equation, applying to static (i.e. DC) electric fields. For
example, at position ( x, y, z ) from a point charge q at the origin
kq kq
= →E= ( x, y , z ) →   E = 0
(x 2
+y +z
2
)
2 1/ 2
(x 2
+y +z
2
)
2 1/ 2
 20
straightforward, but tedious, to show!
2.19 Maxwell’s Equations
We’ve met some of these in part, but here they are in all their glory!
Differential Integral
form form

•D =   D • dS =   dV
S
•B = 0  B • dS = 0
S
B 
 E = −
t 
L
E • d = − 
t S
B • dS

D  D 
James Clerk Maxwell  H = J +
t
 
H • d =  J +

 • dS
t 
(1831 - 1879) L S
 21
Maxwell’s four equations form the basis for the whole of
electromagnetism, and present the most complete theoretical
description of any of the four fundamental forces of nature.
2.20 Laplace’s Equation
Let F =  f be some conservative vector field. Its divergence is
      f f f   2 f  2 f  2 f
 •  f =  , ,  •  , ,  = 2 + 2 + 2  2 f
 x y z   x y z  x y z
(see worked example 2.5)

The scalar field  2 f is called the Laplacian of f, and is shorthand

for the following Cartesian and cylindrical polar forms
 2
f  2
f  2
f 1   f  1  2
f  2
f
 f  2 + 2 + 2 
2
r  + 2 2 + 2
x y z r r  r  r  z
The equation  2 f = 0 is called Laplace’s equation and its solution
is needed in many Engineering problems involving flow (e.g. heat
flux, electrical current, fluid flow, etc.). We will illustrate this with
 22
the following two examples of heat flux q and temperature T;
however, replace these with E (or J) and  and we have solved the
same problem for electric field (or current) – such is the beauty of
maths!
Worked example 2.8
Two parallel metal plates are held at temperatures T1 and T0 , and
are separated by a material of thermal conductivity  and thickness d.
Assuming steady-state heat flux, derive an equation for temperature
as a function of position x between the plates and for the heat flux.
The heat flux vector is q = −  T . For steady-state heat flow it can
be shown that  • q = 0 and so temperature T satisfies  2T = 0 .

The heat flux between the plates is along the x-axis only, so T is a
function of x only and Laplace’s equation simplifies to
 23
T T T d T
2 2 2 2
 T = 2 + 2 + 2 = 2 =0
2
x y z dx

d 2T dT
Integrating 2
= 0 twice → = c1 → T ( x ) = c1 x + c2
dx dx
We find the constants of integration c1 , c2 by applying the boundary
conditions T (0 ) = T1 , T (d ) = T0 → c1 = (T0 − T1 ) / d , c2 = T1
x
Hence the temperature profile is T ( x ) = −(T1 − T0 ) + T1
d
dT T1 − T0
The heat flux is q = −  = − c1 =  (Fourier’s eqn. in 1D)
dx d
The 1D thermal map, lines of heat flux and thermal profile for this
example of 1D heat conduction are plotted below.
24
 25
Worked example 2.9
A hot cylindrical wire (radius a) is held at temperature T1 and is
surrounded by a cold cylindrical sheath (inner radius b) held at T0 ,
separated by a material of thermal conductivity . Assuming steady-
state heat flux, derive an equation for temperature as a function of
position r between the wire and sheath and the heat flux in this space.
The heat flux is radial for a  r  b , so T is a function of r only and
1   T  1  2
T  2
T 1 d  dT 
T=
2
r  + 2 2 + 2 = r =0
r r  r  r  z r dr  dr 
Integrating twice
d  dT  dT
r =0 → r = c1 → T (r ) = c1 ln r + c2
dr  dr  dr
We find the constants of integration c1 , c2 by applying the boundary
conditions T (a ) = T1 , T (b ) = T0 →
T1 = c1 ln a + c2 , T0 = c1 ln b + c2
 26

→ c1 = −(T1 − T0 ) / ln(b / a ), c2 = (T1 ln b − T0 ln a ) / ln(b / a )

T0 ln(r / a ) − T1 ln(r / b )
The temperature profile is therefore T (r ) =
ln(b / a )
dT c (T1 − T0 ) 1 1
The heat flux is q = −  = − 1 = , i.e. q 
dr r ln(b / a ) r r
2(T1 − T0 )
The (constant) thermal power (per m) is W = 2r q =
ln(b / a )

The temperature distribution This is the magnitude of the

T (r ) has radial lines of heat heat flux q .
 27
flux q = −  T .
These contour plots are shown as radial line graphs below, with
values computed for the wire parameters shown.

EEE students should compare these profiles to those obtained for the
electric potential  and electric potential E in a coaxial cable. This
should come as no great surprise since E = −  and  2 = 0 in
the space between the cable’s conductors.
 28
You will find out more in the EN2076 lectures and simulate the
electric field and electric potential in the associated COMSOL labs.