EE750

Advanced Engineering Electromagnetics

Lecture 14

1
 Applications of MoM
• Example on static problems
• Example on 2D scattering problems
• Wire Antennas and scatterers
References
R.F. Harrington, “Field Computation by Moment Methods”
C.A. Balanis, “Advanced Engineering Electroamgnetics”
M. Sadiku,”Numerical Techniques in Electromagnetics”
S.M. Rao et al., “Electromagnetic scattering by surfaces of
arbitrary shape”

EE750, 2003, Dr. Mohamed Bakr 2

 A Charged Conducting Plate

• Find the charge distribution and capacitance of a metalic

plate of dimensions 2a×2a whose potential is Φ = Vo
z
Φ = Vo
y

2a
x

2a

EE750, 2003, Dr. Mohamed Bakr 3

A Charged Conducting Plate (Cont’d)

• The potential and charge satisfy for the unbounded medium

qev
∇ Φ=−
2

ε
• The well-known solution for this problem is
Φ (r ) = ∫∫∫ G (r , r ′) qev (r ′)dx′dy′dz ′
V′

qev (r ′)
Φ (r ) = ∫∫∫ dx′dy′dz ′, R = r - r ′
V ′ 4πεR
• As the plate is assumed to be in the xy plane we may also
write a a qes (r ′)
Φ ( x, y , z ) = ∫ ∫ ′ ′
dx dy , R = ( x − x′) + ( y − y′) + z 2
2 2

- a -a 4πεR

EE750, 2003, Dr. Mohamed Bakr 4

A Charged Conducting Plate (Cont’d)

• We divide the conducting plate into N square subsections

and define the subsectional basis function
 1 on ∆ S n , the n th subsection
fn=
 0, otherwise
• We then expand the unknown surface charge density in
terms of the subsectional basis functions
qes
(x − x ) + ( y − y )
a a
Vo = L(qes ) = ∫ ∫ , R= ′ 2
′ 2

− a- a 4πεR

a a ∑α n fn a a fn
Vo = −∫a -∫a
n
dx′dy′ = ∑ α n ∫ ∫ dx′dy′
4πεR n −a -a 4πεR
EE750, 2003, Dr. Mohamed Bakr 5
A Charged Conducting Plate (Cont’d)

• But as the nth basis function is nonzero only over the nth
subsection we may write
1
Vo = ∑ α n ∫∫ dx′dy′ (one equation in N unknowns)
n ∆S n 4πεR
• We utilize point matching by enforcing the above equation at
the centers of each subsection
1
Vo = ∑ α n ∫∫ ′ ′ ′
dx dy , R m = ( xm − x ) + y m
2
( − x ′)
2

n ∆ S n 4πε R m

m = 1, 2, L, N
• Alternatively, Vo = ∑ l mnα n , m = 1, 2, L , N
n
1
l mn = ∫∫ dx′dy′
∆ S n 4πε R m

EE750, 2003, Dr. Mohamed Bakr 6

A Charged Conducting Plate (Cont’d)

• It follows that the coefficients αn are obtained by solving

 l11 l12 L l1N   α 1  Vo 
    
 l 21 l 22 L l 2 N  α 2  Vo 
=
M M M M  M   M 
    
l N 1 l N 2 l NN  α N  Vo 

• Postprocessing: The capacitance of the conducting plate is

approximated by
N

qt ∑α n ∆ S n
C= = n =1

Vo Vo

EE750, 2003, Dr. Mohamed Bakr 7

A Charged Conducting Plate (Cont’d)

Harrington, Field
Computation by
Moment Methods

The charge distribution along the width of the plate

EE750, 2003, Dr. Mohamed Bakr 8

Scattering Problems
Es
n
Ei

conductor

• An incident wave generates surface currents that in turn

generate a scattered field such that
n × ( E i + E s ) = 0 (zero total tangential electric field)
• In a scattering problem it is required to determine the surface
currents. Es is obtained as a byproduct
EE750, 2003, Dr. Mohamed Bakr 9
Scattering by a Conducting Cylinder of a TM Wave
y
i
E z

ρ′ ρ

x
n

• Incident field has only z direction E = E zi a z

• Fields are dependent on x and y directions. It follows that we
can solve this problem as a 2D problem

EE750, 2003, Dr. Mohamed Bakr 10

Scattering by a Conducting Cylinder (Cont’d)

• Starting with Maxwell’s equations

(∇ × E ) = − jωµH , (∇ × H ) = J + jωεE
• For the case J=Jz we have
∇ E z + k E z = jωµ J z (We consider only the z component)
2 2

• The corresponding Green’s function is obtained by setting

J z = δ ( x − x′)δ ( y − y′) to obtain
− kη 2
G ( ρ, ρ′) = H o (k ρ − ρ′ )
4
• The scattered electric field is thus given by
kη
E z ( ρ) = − ∫ Jz ( ρ′) H o (k ρ − ρ′ )dC ′
s 2

4 C′

EE750, 2003, Dr. Mohamed Bakr 11

Scattering by a Conducting Cylinder (Cont’d)

• For the problem at hand we must have E = − s i

E z for all
z
points on the surface of the cylinder
• It follows that we have
kη
Ez
i
( ρ ) = ∫ Jz ( ρ′) H o (k ρ − ρ′ )dC ′, ∀ρ ∈ C ′
2

4 C′
The only unknown in this equation is Jz
• We expand Jz in terms of the subsectional basis functions

 1 on ∆C n , the n th subsection N
fn= J z = ∑ α n fn
 0, otherwise n =1

EE750, 2003, Dr. Mohamed Bakr 12

 Scattering by a Conducting Cylinder (Cont’d)
• It follows that we have
kη N
E z ( ρ) =
i
∑ α n ∫ f n H o2 (k ρ − ρ′ )dC ′, ∀ρ ∈ C ′
4 n =1 C ′
kη N
E ( ρ) =
i
z ∑ α n ∫ H o2 (k ρ − ρ′ )dC ′, ∀ρ ∈ C ′
4 n =1 ∆C n
(one equation in N unknowns)
• We utilize point matching to enforce the above equation at
the centers of the subsections ρ m = ( x m , y m), m = 1, 2, L , N
kη N
E z ( ρm ) =
i
∑ α n ∫ H o2 ( k ρm − ρ′ )dC ′, m = 1, 2, L , N
4 n =1 ∆C n
(N equation in N unknowns)
EE750, 2003, Dr. Mohamed Bakr 13
Scattering by a Conducting Cylinder (Cont’d)

Harrington, Field
Computation by
Moment Methods

For a uniform plane wave incident at an angle φi we have

= ijk ( xcos φ + ysin φ ) = jk.r
E ez i i e

EE750, 2003, Dr. Mohamed Bakr 14

Pocklington’s Integral Equation
z Ei

l/2
y

x l/2

2a
• The target is to determine the current distribution and
consequently the scattered field due to an incident field for a
finite-diameter wire
EE750, 2003, Dr. Mohamed Bakr 15
 Pocklington’s Integral Equation (Cont’d)
• The main relation for this scatterer is
E
i
(
z ρ = a ) = − E z ( ρ = a)
s

• The equations governing the scattered field are

E = − jωA − ( j / ωµε )(∇(∇. A))
• We need only the z component of the field
∂ ∂
2 2
- j A - j
E z (r ) =
s
( β Az +
2 z
) E z (r ) =
s
( β + 2 )Az
2

ωµε ∂z 2
ωµε ∂z
• The z component of the magnetic vector potential is
µ e
-jβR
µ l/ 2 2π e
-jβR
Az ( r ) = ∫∫ J z ( r ′) ds ′ = ∫ ∫ J z ( z ′, φ ′) adφ ′dz ′
4π S R 4π -l/ 2 0 R
R = r - r′

EE750, 2003, Dr. Mohamed Bakr 16

 Pocklington’s Integral Equation (Cont’d)
• If the wire is thin, Jz is not a function of φ
l/ 2  1 2π e-jβR  r
Az (r ) = µ ∫ 2πa J z ( z ′)  ∫ dφ ′dz ′ z − z′
-l/ 2  2π 0 4πR  ρ

I z (z ′) G (r , z ′)
a
R
• The distance R in cylindrical coordinate is
R = ρ 2 + a 2 − 2 ρacos(φ − φ ′) + ( z − z ′)2
• For observation points on the wire surface we have
R = 2 a − 2 a cos(φ − φ ) + ( z − z )
2 2
′ ′ 2

φ −φ′
R = 4 a sin (
2 2
) + ( z − z ′)2
2
EE750, 2003, Dr. Mohamed Bakr 17
Pocklington’s Integral Equation (Cont’d)
• But as Az has a φ symmetry, we may write
Az(ρ=a,z,φ) = Az(ρ=a,z,0)

l/2
φ′
Az ( a , z ) = µ ∫Iz ( z ′)G ( z , z ′) dz ′, R =
2
4 2
a sin ( ) + ′
(z−z )
2

−l / 2 2
• The scattered field at the wire surface is thus given by
∂ l/2
2
- j
E z ( a, z ) =
s
( β + 2 ) ∫ I z ( z ′)G ( z , z ′)dz ′
2

ωε ∂ z −l / 2
• But as E iz (a, z ) = − E sz (a, z ), we may write
∂
l/2 2
- jωεE z (a, z ) = ∫ I z ( z ′)( β + 2 )G ( z , z ′)dz ′
i 2

−l / 2 ∂z
Pocklington’s integral equation (only Iz is not known)
EE750, 2003, Dr. Mohamed Bakr 18
 Solution of Pocklington’s Integral equation
• Divide the wire into N non overlapping segments
• Expand the unknown current in terms of the
N
basis functions I z ( z ) = ∑ I n u n ( z )
n =1 ∆
• For pulse functions we have
1, z n-1/ 2 < z < z n +1/ 2
un = 
 0, otherwise

• For triangular functions we have

∆ − z − zn

 ∆ , z n-1 < z < z n +1
=
un 

0 otherwise

EE750, 2003, Dr. Mohamed Bakr 19

 Solution of Pocklington’s Equation (Cont’d)
• It follows that
∂
l/2 N 2
- jωεE z (a, z ) = ∫ ∑ I n u n ( z ′)( β + )G ( z , z ′)dz ′
i 2

−l / 2 n =1 ∂z 2

∂
N l/2 2
- jωεE z (a, z ) = ∑ I n ∫ u n ( z ′)( β + )G ( z , z ′)dz ′
i 2

n =1 −l / 2 ∂z 2

using a pulse function

∂
N 2
- jωεE z (a, z ) = ∑ I n ∫ ( β + 2 )G ( z , z ′)dz ′
i 2

n =1 l n ∂z
N
E ( z) = ∑ I n Gn ( z)
i
z
n =1
One equation in N unknowns
EE750, 2003, Dr. Mohamed Bakr 20
 Solution of Pocklington’s Equation (Cont’d)
• Enforcing this equation at the center of each segment, we get
N equations in N unknowns
N
E z ( z m) = ∑ I n G n ( z m), m = 1, 2, L, N
i
n =1

 G1 ( z1) G 2 ( z1) L G N ( z1)   I 1   E iz ( z1) 

 ( ) ( ) L ( )    i 
G
 1 2 z G 2 z 2 G N z 2 I (
 2  =  E z z2  )
 M M M M  M   M 
    i 
G1 ( z N ) G 2 ( z N ) G N ( z N )  I N   E z ( z N )

EE750, 2003, Dr. Mohamed Bakr 21