Upload
gopi-saiteja
View
66
Download
0
Tags:
Embed Size (px)
Citation preview
FDTD: An Introduction
Dr. Rakhesh Singh Kshetrimayum
2/25/2015FDTD by R. S. Kshetrimayum1
Dr. Rakhesh Singh Kshetrimayum
FDTD: An Introduction
� Background on commercial softwares and Who’s Who’s in CEM
(a) EMSS software FEKO� Originated from Dr. Ulrich Jacobus.� Dr. Francs Meyer (a student of Prof. D. B. Davidson, a
2/25/2015FDTD by R. S. Kshetrimayum2
� Dr. Francs Meyer (a student of Prof. D. B. Davidson, a Professor at University of Stellenbosch, South Africa) is one of the main person behind this company
FDTD: An Introduction
(b) IE3D by Zeland Software Inc.� Founded by Dr. Jian-Xiong Zheng� Did PhD under Prof. D. C. Chang (President Emeritus of
NYU Poly)(c) REMCOM XFDTD is a product of Penn. State University
2/25/2015FDTD by R. S. Kshetrimayum3
(c) REMCOM XFDTD is a product of Penn. State University� Founders are H. Scott Langdon, Dr. Raymond Lubbers and
Dr. Christopher Penny
FDTD: An Introduction
(d) James C. Rautio is founder of SONNET� He did his PhD from Syracuse university, USA under Prof. R.
F. Harrington (considered as originator of MoM in EM area)(e) Dr. WenhuaYu who was a member of Prof. Raj Mittra group
at Penn. State University, USA founded GEMS software
2/25/2015FDTD by R. S. Kshetrimayum4
at Penn. State University, USA founded GEMS software� And list goes on
FDTD: An Introduction
� FDTD is one of the very popular techniques � in Computational electromagnetics (CEM)
� What is CEM?� In CEM
� Try to solve complex EM problems
2/25/2015FDTD by R. S. Kshetrimayum5
� Try to solve complex EM problems� which do not have tractable analytical solutions � Using numerical methods employing computers
FDTD: An Introduction
Why CEM?� Possible to simulate a device/experiment/phenomenon any
number of times � Try to achieve the best or optimal result before actually
doing the experiments (preserving resources)
2/25/2015FDTD by R. S. Kshetrimayum6
doing the experiments (preserving resources)� Some experiments are
� dangerous to perform like lightning strike on aircraft or � not possible to conduct because of limited facilities/resources
FDTD: An Introduction
� How to learn CEM (including FDTD)?� Can be learnt by doing it� Once you learn the basic concepts� You should write or code simple programs in any language� Always start with simple problems to gain confidence
2/25/2015FDTD by R. S. Kshetrimayum7
� Always start with simple problems to gain confidence� and try for more difficult problems later
FDTD: An Introduction
� Many CEM methods are available� MoM� FEM� FDTD
� FDTD (our choice)
2/25/2015FDTD by R. S. Kshetrimayum8
� FDTD (our choice)� Easy to implement� Can see fields and waves in real time (time domain techniques)
� Expected outcome of this talk � Understand fundamentals of FDTD� Will explain and give 1-D, 2-D and 3-D FDTD programs� Can start your research works right way
FDTD: An Introduction
� Let us start with 4 Maxwell’s equations
2. 0B∇ • =r
ε
ρvE =•∇r
.1
2/25/2015FDTD by R. S. Kshetrimayum9
3.D
H Jt
∂∇× = +
∂
rr r
4.B
Et
∂∇× = −
∂
urur
Maxwell curl equations (why?)
FDTD: An Introduction� What is FDTD?� Finite Difference Time Domain� Basically, we replace spatial & time derivatives in the two
Maxwell’s curl equations � by central finite difference approximation
2/25/2015FDTD by R. S. Kshetrimayum10
� What is central finite difference approximation of derivatives?� Consider a function f(x), its derivative at x0 from central finite
difference approximation is given by
0 0
'0
0
( ) 2 2( )
x xf x f x
df xf x
dx x
∆ ∆ + − −
= ≅∆
FDTD: An Introduction
� The time-dependent and source free ( ) Maxwell’s curl equations in a medium with ε=ε0εr and µ = µ0 µr are
0J =r
0 0
1 1;
r r
E HH E
t tε ε µ µ
∂ ∂= ∇× = − ∇×
∂ ∂
r rr r
3 equations 3 equations
2/25/2015FDTD by R. S. Kshetrimayum11
� For Cartesian coordinate systems, � expanding the curl equations, � equating the vector components,
� we have 6 equations
FDTD: An Introduction
� Expanding the first vector curl equation
� we get 3 equations
0
1
r
EH
t ε ε
∂= ∇×
∂
rr
2/25/2015FDTD by R. S. Kshetrimayum12
0
0
0
1
1
1
yx z
r
y x z
r
y xz
r
HE H
t y z
E H H
t z x
H HE
t x y
ε ε
ε ε
ε ε
∂ ∂ ∂= −
∂ ∂ ∂
∂ ∂ ∂ = −
∂ ∂ ∂
∂ ∂∂= −
∂ ∂ ∂
FDTD: An Introduction
� Expanding the second vector curl equation
� We get 3 equations
0
1
r
HE
t µ µ
∂= − ∇×
∂
rr
1 EH E∂ ∂ ∂
2/25/2015FDTD by R. S. Kshetrimayum13
0
0
0
1
1
1
yx z
r
y xz
r
yxz
r
EH E
t z y
H EE
t x z
EEH
t y x
µ µ
µ µ
µ µ
∂ ∂ ∂= −
∂ ∂ ∂
∂ ∂∂ = −
∂ ∂ ∂
∂ ∂∂= −
∂ ∂ ∂
FDTD: An Introduction
As we know that FDTD is a time-domain solverThe question is how do we solve those 6 equations above?1. 1-D FDTD update equations
� For 1-D case (a major simplification), we can consider (a) Linearly polarized wave along x-axis
2/25/2015FDTD by R. S. Kshetrimayum14
� (a) Linearly polarized wave along x-axis � exciting an electric field which has Ex only (Ey = Ez = 0)
� (b) propagation along z-axis� no variation in the x-y plane, i.e. ∂/∂x = 0 and ∂/∂y = 0
FDTD: An Introduction
� Then, the above 6 equations� reduce to 2 equations for 1-D FDTD
0
1 yx
r
HE
t zε ε
∂ ∂= −
∂ ∂
2/25/2015FDTD by R. S. Kshetrimayum15
� We want to solve these equations at different locations and time in solution space
� Hence we need to discretize both in time and space
0
1y x
r
H E
t zµ µ
∂ ∂ = −
∂ ∂
FDTD: An Introduction
2 3 4 5
∆z
1
space discretization size
running index k
2/25/2015FDTD by R. S. Kshetrimayum16
Fig. 1(a) Discretizations in space
Fig. 1(b) Discretization in time
∆t
1 2 3 4 5
time discretization size running index n
FDTD: An Introduction
� Notations:� n is the time index and � k is the spatial index,
� which indexes � Locate positions z = k∆z
2/25/2015FDTD by R. S. Kshetrimayum17
� Locate positions z = k∆z� Usually ∆z is taken as 10-15 per wavelength� gives accurate
results � times t = n∆t and� ∆t (dependent on ∆z) chosen from Courant stability criterion � gives stable FDTD solution (will be discussed later)
FDTD: An Introduction
� for the first equation:
� the central difference approximations for both the temporal and spatial derivatives are obtained at
0
1 yx
r
HE
t zε ε
∂ ∂= −
∂ ∂
2/25/2015FDTD by R. S. Kshetrimayum18
and spatial derivatives are obtained at � z = k∆z, (space discretization size and index)� t = n∆t (time discretization size and index)
0
1 1 1 1, , , ,
12 2 2 2x x y y
r
E k n E k n H k n H k n
t zε ε
+ − − + − −
= −∆ ∆
FDTD: An Introduction
� for the second equation:
� the central difference approximations for both the temporal and spatial derivatives are obtained at
� at (z + ∆z/2, t + ∆t/2)
0
1y x
r
H E
t zµ µ
∂ ∂ = −
∂ ∂
2/25/2015FDTD by R. S. Kshetrimayum19
� at (z + ∆z/2, t + ∆t/2) � (increment all the time and space steps by 1/2):
0
1 1 1 1, 1 , 1, ,
12 2 2 2y y x x
r
H k n H k n E k n E k n
t zµ µ
+ + − + + + − +
= −∆ ∆
FDTD: An Introduction
� The above equations can be rearranged as a pair of ‘computer update equations’, � which can be repeatedly updated in loop, � to obtain the next time values of Ex (k, n+1/2) and Hy (k + ½.
n+1)
2/25/2015FDTD by R. S. Kshetrimayum20
n+1) � from the previous time values as follows
0
1 1 1 1 1, , , ,
2 2 2 2x x y y
r
tE k n E k n H k n H k n
zε ε
∆ + = − − + − − ∆
0
1 1 1 1 1, 1 , 1, ,
2 2 2 2y y x x
r
tH k n H k n E k n E k n
zµ µ
∆ + + = + − + + − + ∆
FDTD: An Introduction
k-2 k-1 k k+1 k+22/1−n
xE
nH
2/25/2015FDTD by R. S. Kshetrimayum21
Fig. 2 Interleaving of E and H fields
k-2 k-1 k k+1 k+2
k-5/2 k-3/2 K-1/2 K+1/2K+3/2K+5/22/1+n
xE
n
yH
FDTD: An Introduction
� Interleaving of the E and H fields in space and time in the FDTD formulation
� to calculate Ex(k), the neighbouring values of Hy at k-1/2 and k+1/2 of the previous time instant are needed
� Similarly, to calculate Hy(k+1/2), for instance, the
2/25/2015FDTD by R. S. Kshetrimayum22
� Similarly, to calculate Hy(k+1/2), for instance, the neighbouring values of Ex at k and k+1 of the previous time instant are needed
FDTD: An Introduction
� In the above equations, � ε0 and µ0 differ by several orders of magnitude, � Ex and Hy will differ by several orders of magnitude
� Numerical error is minimized by making the following change of variables as
2/25/2015FDTD by R. S. Kshetrimayum23
change of variables as
xx EEµ
ε='
FDTD: An Introduction
� Hence,
' '1 1 1 1 1 1 1, , , ,
2 2 2 2x x y y
tE k n E k n H k n H k n
zεε ε
µ µ
∆ + = − − + − − ∆
2/25/2015FDTD by R. S. Kshetrimayum24
' '
' '
1 1 1 1 1, , , ,
2 2 2 2
1 1 1 1, ,
2 2
x x y y
x x y
tE k n E k n H k n H k n
z
tE k n E k n H k
z
ε
µ ε
µε
∆ ⇒ + = − − + − − ∆
∆ ⇒ + = − − +
∆
' '
1, ,
2 2
1 1 1 1 1, , , ,
2 2 2 2
y
x x y y
n H k n
tE k n E k n H k n H k n
zµε
− −
∆ ⇒ + = − + − − + ∆
FDTD: An Introduction
� Similarly,
' '1 1 1 1 1 1, 1 , 1, ,
2 2 2 2
1 1 1 1 1
y y x x
tH k n H k n E k n E k n
z
t
µε
µ
∆ + + = + − + + − + ∆
∆
2/25/2015FDTD by R. S. Kshetrimayum25
' '1 1 1 1 1, 1 , 1, ,
2 2 2 2
1 1 1, 1 ,
2 2
y y x x
y y
tH k n H k n E k n E k n
z
tH k n H k n E
z
µε
µε
∆ ⇒ + + = + − + + − + ∆
∆ ⇒ + + = + +
∆
' '1 1, 1,
2 2x xk n E k n
+ − + +
FDTD: An Introduction
Courant stability criteria:
� Courant stability criteria dictates the � relationship between the time increment ∆t with respect to
space increment ∆z� in order to have a stable FDTD solution of the electromagnetic
2/25/2015FDTD by R. S. Kshetrimayum26
� in order to have a stable FDTD solution of the electromagnetic problems
� In isotropic media, an electromagnetic wave propagates a distance of one cell in time ∆t = ∆z/vp, � where vp is the phase velocity
FDTD: An Introduction
� This limits the maximum time step� This equation implies that an EM wave cannot be allowed
� to move more than a space cell during a time step� Otherwise, FDTD will start diverging� If we choose ∆t > ∆z/v ,
2/25/2015FDTD by R. S. Kshetrimayum27
� If we choose ∆t > ∆z/vp, � the distance moved by the EM wave over the time interval ∆t
will be more than ∆z, � the EM wave will leave out the next node/cell and � FDTD cells are not causally interconnected and
� hence, it leads to instability
FDTD: An Introduction
cell
2/25/2015FDTD by R. S. Kshetrimayum28
y∆
x∆Fig. 3 (a) 2-D discretizations
x
y
FDTD: An Introduction
∆
2
∆ Propagation at θ=450
2/25/2015FDTD by R. S. Kshetrimayum29
Plane wave front
∆
Propagation at θ=00
Fig. 3 (b) Plane wave front propagation along θ=00 andθ=450
FDTD: An Introduction
� Propagation at θ=450
� Wavefront jumps from one row of nodes to the next row, � the spacing between the consecutive rows of nodes is
2
∆
2/25/2015FDTD by R. S. Kshetrimayum30
� Similarly for 3-D case2
3
∆
FDTD: An Introduction
� In the case of a 2-D simulation, � we have to allow for the propagation in the diagonal
direction, � which brings the time requirement to ∆t = ∆z/√2vp
� Obviously, three-dimensional simulation requires ∆t =
2/25/2015FDTD by R. S. Kshetrimayum31
� Obviously, three-dimensional simulation requires ∆t = ∆z/√3vp
� We will use in all our simulations a time step ∆t of
2p
zt
v
∆∆ =
⋅
FDTD: An Introduction
� where vp is the phase velocity, � which satisfies the requirements in 1-D, 2-D and 3-D for all
media (√2≈1.414<√3≈1.7321<2)� In 3D case, it may be more appropriate to modify the above
stability criteria as
2/25/2015FDTD by R. S. Kshetrimayum32
stability criteria as
� The above factor now can be simplified as
( )min , ,
2 p
x y zt
v
∆ ∆ ∆∆ =
⋅
1 1
2 2
p
p
p
v zt tv
z z v zεµ
⋅ ∆∆ ∆= ⋅ = =
∆ ∆ ⋅ ⋅ ∆
FDTD: An Introduction� Making use of this in the above two equations, � we obtain the following equations
' '1 1 1 1 1, , , ,
2 2 2 2 2x x y yE k n E k n H k n H k n
+ = − + − − +
' '1 1 1 1 1, 1 , , 1,H k n H k n E k n E k n
+ + = + + + − + +
2/25/2015FDTD by R. S. Kshetrimayum33
� Hence the computer update equations are� ex[k] = ex[k] + (0.5/er(k))*( hy[k-1] - hy[k] )� hy[k] = hy[k] + 0.5*( ex[k] - ex[k+1] )
' '1 1 1 1 1, 1 , , 1,
2 2 2 2 2y y x xH k n H k n E k n E k n
+ + = + + + − + +
FDTD: An Introduction� k+1/2 and k-1/2 are replaced by k or k-1� Note that the n or n + 1/2 or n −1/2 in the superscripts do not
appear� FDTD Simulation of Gaussian pulse propagation in free space (fdtd_1d_1.m)
2/25/2015FDTD by R. S. Kshetrimayum34
(fdtd_1d_1.m)� FDTD Simulation of Gaussian pulse hitting a dielectric medium (fdtd1_die.m)
� From theoretical analysis of EM wave hitting a dielectric surface
21
21
12
12
εε
εε
ηη
ηη
+
−=
+
−=Γ
21
1
12
222
εε
ε
ηη
ητ
+=
+=
FDTD: An Introduction
� For ε1=1.0 and ε2=4.0, we have � FDTD simulation of Absorbing Boundary Condition (fdtd_1d_3.m)
� In calculating the E field, � we need to know the surrounding H values;
3
2;
3
1=−=Γ τ
2/25/2015FDTD by R. S. Kshetrimayum35
� we need to know the surrounding H values; � this is the fundamental assumption in FDTD
� At the edge of the problem space, � we will not have the value to one side, � but we know there are no sources outside the problem space
FDTD: An Introduction
� The wave travels ∆z/2 (=c0 · ∆ t) distance in one time step, � so it takes two time steps for a wave front to cross one cell
� Suppose we are looking for a boundary condition at the end where k=1
� Now if we write the E field at k=1 as
2/25/2015FDTD by R. S. Kshetrimayum36
� Now if we write the E field at k=1 as � Ex (1,n) = Ex (2, n-2), � then the fields at the edge will not reflect
� This condition must be applied at both ends
FDTD: An Introduction
2. FDTD Solution to Maxwell’s equations in 2-D Space
� In deriving 2-D FDTD formulation, we choose between one of two groups of three vectors each:
� (a) Transverse magnetic (TMz) mode, which is composed of Ez, Hx, and Hy or
2/25/2015FDTD by R. S. Kshetrimayum37
Ez, Hx, and Hy or � (b) Transverse electric (TEz) mode, which is composed of Ex,
Ey, and Hz
FDTD: An Introduction
� Unlike time-harmonic guided waves, � none of the fields vary with z so that � there is no propagation in the z-direction
� But in general propagation along x- or y- directions or both is possible
2/25/2015FDTD by R. S. Kshetrimayum38
� FDTD simulation of TM mode wave propagation (fdtd_2d_TM_1.m )
� Expanding the Maxwell’s curl equations with � Ex = 0, Ey = 0, Hz = 0 and ∂/∂z = 0,
� we obtain, � 3 equations from the 6 equations
FDTD: An Introduction
� 3 equations as follows
1 y xzH HE
t x yε
∂ ∂∂= −
∂ ∂ ∂
1H E ∂ ∂
2/25/2015FDTD by R. S. Kshetrimayum39
1
1
x z
y z
H E
t y
H E
t x
µ
µ
∂ ∂=
∂ ∂
∂ ∂ = −
∂ ∂
FDTD: An Introduction
� Central finite difference approximations of the above 3 equations are as follows
1 1, , , ,
2 2
1 1 1 1
z zE i j n E i j n
t
+ − −
∆
1 y xzH HE
t x yε
∂ ∂∂= −
∂ ∂ ∂
2/25/2015FDTD by R. S. Kshetrimayum40
� n+1/2�n+1 & j�j+1/2
0
1 1 1 1, , , , , , , ,
1 2 2 2 2y y x x
r
H i j n H i j n H i j n H i j n
x yε ε
+ − − + − −
= −∆ ∆
0
1 1 1 1, , 1 , , , 1, , ,
12 2 2 2x x z z
r
H i j n H i j n E i j n E i j n
t yµ µ
+ + − + + + − +
= −∆ ∆
FDTD: An Introduction
n+1/2�n+1 & i�i+1/2
� Rearranging the above equations, we can write the final 3 update equations’ expressions as
0
1 1 1 1, , 1 , , 1, , , ,
12 2 2 2y x z z
r
H i j n H i j n E i j n E i j n
t xµ µ
+ + − + + + − +
=∆ ∆
2/25/2015FDTD by R. S. Kshetrimayum41
update equations’ expressions as
0
1, ,
2
1 1 1 1, , , , , , , ,
1 2 2 2 2, ,
2
z
y y x x
z
r
E i j n
H i j n H i j n H i j n H i j nt
E i j nx yε ε
+
+ − − + − − ∆ = − + −
∆ ∆
FDTD: An Introduction
0
1 1, 1, , ,
1 1 2 2, , 1 , ,
2 2
z z
x x
r
E i j n E i j nt
H i j n H i j nyµ µ
+ + − + ∆ + + = + −
∆
2/25/2015FDTD by R. S. Kshetrimayum42
0
1 11, , , ,
1 1 2 2, , 1 , ,
2 2
z z
y x
r
E i j n E i j nt
H i j n H i j nxµ µ
+ + − + ∆ + + = + +
∆
FDTD: An Introduction
� Computer update equations are (k+1/2 and k-1/2 are 0
1, ,
2
1 1 1 1, , , , , , , ,
1 2 2 2 2, ,
2
z
y y x x
z
r
E i j n
H i j n H i j n H i j n H i j nt
E i j nx yε ε
+
+ − − + − − ∆ = − + −
∆ ∆
2/25/2015FDTD by R. S. Kshetrimayum43
� Computer update equations are (k+1/2 and k-1/2 are replaced by k or k-1)
� ez(i,j)=ez(i,j)+{dt/(eps_0*eps_r)}*[{Hy(i,j)-Hy(i-1,j)}/dx-{Hx(i,j)-Hx(i,j-1)}/dy]
� Hx(i,j) = Hx(i,j)-{dt/(mu_0*dy)}*{Ez(i,j+1)-Ez(i,j)}� Hy(i,j) = Hy(i,j)+{dt/(mu_0*dx)}*{Ez(i+1,j)-Ez(i,j)}
FDTD: An Introduction
� Some new MATLAB commands:� image: IMAGE(C) displays matrix C as an image� imagesc: data is scaled to use the full color map� Colorbar: appends a vertical color scale to the current axis
Colormap: a matrix may have any number of rows but it
2/25/2015FDTD by R. S. Kshetrimayum44
� Colormap: a matrix may have any number of rows but it must have exactly 3 columns
� Each row is interpreted as a color, with the first element specifying the intensity of red, the second gree and the third blue
� Color intensity is specified on the interval of 0.0 to 1.0
FDTD: An Introduction
� For example, [0 0 0] is black� [1 1 1] is white� [1 0 0] is red� [0.5 0.5 0.5] is gray
[127/255 1 212/255] is aquamarine
2/25/2015FDTD by R. S. Kshetrimayum45
� [127/255 1 212/255] is aquamarine
FDTD: An Introduction
FDTD simulation of TM wave hitting a dielectric surface (right side), source at the center (2-D) (fdtd_2d_TM_2.m)
� Modify the above program as follows:� Specify the position of the dielectric surface (id=..,jd=..)� Choose a value of eps_r=10.0..
2/25/2015FDTD by R. S. Kshetrimayum46
� Choose a value of eps_r=10.0..� Modify the constant value C1 accordingly� Run the program� In free space wave is propagating freely and � on the other side reflection and transmission of the wave at
the interface
FDTD: An Introduction
� Wave propagates more slowly on the medium on the right and � the pulse length is shorter
� FDTD simulation of TM wave propagation due to multiple sources (fdtd_2d_TM_multi_source.m)
2/25/2015FDTD by R. S. Kshetrimayum47
(fdtd_2d_TM_multi_source.m)3. 3-D FDTD
� From two Maxwell’s curl equations, � we need to consider all 6 equations for 3-D case
0 0
1 1;
r r
E HH E
t tε ε µ µ
∂ ∂= ∇× = − ∇×
∂ ∂
r rr r
FDTD: An Introduction
Yee’s FDTD grid
� Electric fields are calculated at “integer” time-steps � and magnetic field at “half-integer” time-steps
� Electric field components are placed at mid-points of the corresponding edges
2/25/2015FDTD by R. S. Kshetrimayum48
corresponding edges� E.g. � Ex is placed at midpoints of edges oriented along x-direction� Ex is on half-grid in x and n the integer grids in y & z
FDTD: An Introduction
� The magnetic field components are � placed at the centers of the faces of the cubes and � oriented normal to the faces
� E.g.� Hx components are placed at the centers of the faces in the
2/25/2015FDTD by R. S. Kshetrimayum49
� Hx components are placed at the centers of the faces in the yz-plane
� Hence Hx is � on the integer grid in x and � on the half-grid in y & z
FDTD: An Introduction
2/25/2015FDTD by R. S. Kshetrimayum50Fig. 4 Yee’s FDTD grid
FDTD: An Introduction
( ) ( )
( ) ( )
( ) ( )
∆
−+−++−
∆
−+−++
=∆
+−+
++
++
+
z
rqpHrqpH
y
rqpHrqpH
t
rqpErqpE
n
y
n
y
n
z
n
z
r
n
x
n
x
2/1,,2/12/1,,2/11
,2/1,2/1,2/1,2/11
,,2/1,,2/1
2/12/1
2/12/1
0
1
εε
εε
∂
∂−
∂
∂=
∂
∂
z
H
y
H
t
E yz
r
x
0
1
εε
2/25/2015FDTD by R. S. Kshetrimayum51
∆zr 0εε
( ) ( )
( ) ( )
( ) ( )
∆
+−−++−
∆
−+−++
=∆
+−+
++
++
+
z
rqpHrqpH
z
rqpHrqpH
t
rqpErqpE
n
z
n
z
r
n
x
n
x
r
n
y
n
y
,2/1,2/1,2/1,2/11
2/1,2/1,2/1,2/1,1
,2/1,,2/1,
2/12/1
0
2/12/1
0
1
εε
εε
∂
∂−
∂
∂=
∂
∂
x
H
z
H
t
Ezx
r
y
0
1
εε
FDTD: An Introduction
( ) ( )
( ) ( )
( ) ( )
∆
+−−++−
∆
+−−++
=∆
+−+
++
++
+
y
rqpHrqpH
x
rqpHrqpH
t
rqpErqpE
n
x
n
x
n
y
n
y
r
n
z
n
z
2/1,2/1,2/1,2/1,1
2/1,,2/12/1,,2/11
2/1,,2/1,,
2/12/1
2/12/1
0
1
εε
εε
∂
∂−
∂
∂=
∂
∂
y
H
x
H
t
E xy
r
z
0
1
εε
2/25/2015FDTD by R. S. Kshetrimayum52
∆
−yr 0
εε
( ) ( )
( ) ( )
( ) ( )
∆
+−++−
∆
+−++
=∆
++−++ −+
y
rqpErqpE
z
rqpErqpE
t
rqpHrqpH
n
z
n
z
n
y
n
y
n
x
n
x
2/1,,2/1,1,1
,2/1,1,2/1,1
2/1,2/1,2/1,2/1,
0
0
2/12/1
µ
µ
∂
∂−
∂
∂−=
∂
∂
z
E
y
E
t
H yzx
0
1
µ
FDTD: An Introduction
( ) ( )
( ) ( )
( ) ( )
∆
+−++−
∆
+−++
=∆
++−++ −+
z
rqpErqpE
x
rqpErqpE
t
rqpHrqpH
n
x
n
x
n
z
n
z
n
y
n
y
,,2/11,,2/11
2/1,,2/1,,11
2/1,,2/12/1,,2/1
0
2/12/1
µ
µ
∂
∂−
∂
∂−=
∂
∂
x
E
z
E
t
Hzxy
0
1
µ
2/25/2015FDTD by R. S. Kshetrimayum53
∆z
0µ
( ) ( )
( ) ( )
( ) ( )
∆
+−++−
∆
+−++
=∆
++−++ −+
x
rqpErqpE
y
rqpErqpE
t
rqpHrqpH
n
y
n
y
n
x
n
x
n
z
n
z
,2/1,,2/1,11
,,2/1,2/1,2/11
,2/1,2/1,2/1,2/1
0
0
2/12/1
µ
µ
∂
∂−
∂
∂−=
∂
∂
y
E
x
E
t
H xyz
0
1
µ
FDTD: An Introduction
� FDTD simulation of cubical cavity (fdtd_3D_demo.m)
� We will use FDTD to find � the resonant frequencies of an air-filled cubical cavity with
metal walls� Peaks in the frequency response indicate
2/25/2015FDTD by R. S. Kshetrimayum54
� Peaks in the frequency response indicate � the presence of resonant modes in the cavity
� An initial Hz field excites only � the TE modes
� The first four eigenfrequencies� TE101, TE011/TE201, TE111 and TE102 modes
FDTD: An Introduction
4. Perfectly Matched Layers
� Berenger’s PML is an artificial material � that is theoretically designed to create no reflections
regardless of the � frequency,
2/25/2015FDTD by R. S. Kshetrimayum55
� frequency, � polarization and � angle of incidence of a plane wave upon its interface
FDTD: An Introduction
� PML could be analyzed in stretched coordinate systems� Source-free Maxwell’s equations
( ) ( )HE
EjHHjE
ss
ss
=•∇=•∇
=×∇−=×∇
;0;0
;;rr
rrrr
µε
ωεωµ
2/25/2015FDTD by R. S. Kshetrimayum56
� Consider plane waves
( ) ( )
( ) ( ) ( ) zzsz
yysy
xxsx
zyx
s
ss
∂
∂+
∂
∂+
∂
∂=∇
1ˆ
1ˆ
1ˆ
zkykxkrk
eHHeEE
zyx
rkjrkj
++=•
== •−•−
rr
rrrr rrrr
00 ;
FDTD: An Introduction
� Substituting these into Maxwell’s equations, we have,
( ) ( )
ˆˆˆ
;0;0
;;
++=
=•=•
−=×=×
zyxs
ss
ss
s
kz
s
ky
s
kxk
HkEk
EHkHEk
r
rrrr
rrrrrr
µε
εωµ
2/25/2015FDTD by R. S. Kshetrimayum57
� Solution to the above equation is given by
222
+
+
=•
y
z
y
y
x
xss
zyx
s
s
k
s
k
s
kkk
sss
rr
θφθφθ cos,sinsin,cossin zzyyxx kskkskksk ===
FDTD: An Introduction
� If sx is a complex number with a negative imaginary part, � the wave will be attenuated in the x direction
� Wave impedance is independent of coordinate stretching
( )0,1,
'''cossincossin '''
>≥== −−−−sseee
xjssjkxjksxjk xx φθφθ
2/25/2015FDTD by R. S. Kshetrimayum58
� Wave impedance is independent of coordinate stretching
ε
µωµη ===
kH
Err
r
FDTD: An Introduction'''
,1 jsssss yzx −===
==
'''
,1:4
jssss
scorners
yx
z
−==
=
y
2/25/2015FDTD by R. S. Kshetrimayum59
Object
Fig. 5 Computational domain truncated using the conductor backed PMLs
''',1 jsssss yzx −===
'''
,1
jsss
ss
x
zy
−=
=='''
,1
jsss
ss
x
zy
−=
==x
FDTD: An Introduction
( ) ( ) ( )
( ) ( ) ωµωµ
ωµ
syy
sxx
zyxs
HjEyys
HjExxs
HjEzzs
Eyys
Exxs
E
−=×∂
∂−=×
∂
∂
−=×∂
∂+×
∂
∂+×
∂
∂=×∇
;ˆ1
;ˆ1
;ˆ1
ˆ1
ˆ1
rrrr
rrrrr
2/25/2015FDTD by R. S. Kshetrimayum60
( )
ωε
σ
ωε
σ
ωε
σ
ωµ
zz
y
yx
x
szsysxszz
yx
jsjsjsgChoo
HHHHHjEzzs
ysxs
−=−=−=
++=−=×∂
∂
∂∂
1,1,1sin
;;ˆ1 rrrrrr
FDTD: An Introduction
( )
( )
( ) szzsz
szz
sy
ysy
syy
sxxsx
sxx
Ht
HHsjEz
z
Ht
HHsjEy
y
Ht
HHjjEx
x
rr
rr
rr
rr
rr
rr
ωε
µσµωµ
ωε
µσµωµ
ωε
µσµ
ωε
σωµ
−∂
∂−=−=×
∂
∂
−∂
∂−=−=×
∂
∂
−∂
∂−=
−−=×
∂
∂
ˆ
ˆ
1ˆ
2/25/2015FDTD by R. S. Kshetrimayum61
( )
( )
( )
( ) szzsz
syy
sy
sxxsx
szszz
Et
EHz
z
Et
EHy
y
Et
EHx
x
tz
rr
r
rr
r
rr
r
σε
σε
σε
ωε
+∂
∂=×
∂
∂
+∂
∂=×
∂
∂
+∂
∂=×
∂
∂
∂∂
ˆ
ˆ
ˆ
FDTD: An Introduction
5. References1. S. D. Gedney, Introduction to the Finite-Difference Time-
Domain Method for Electromagnetics, Morgan & Claypool, 2011
2. R. Garg, Analytical and Computational Methods in Electromagnetics, Artech House, 2008
2/25/2015FDTD by R. S. Kshetrimayum62
Electromagnetics, Artech House, 20083. A. Taflove and S. C. Hagness, Computational
electrodynamics the Finite-Difference Time-Domain Method, Artech House, 2005
4. T. Rylander, P. Ingelstrom and A. Bondeson, Computational Electromagnetics, Springer, 2013
FDTD: An Introduction
5. J. M. Jin, Theory and Computation of Electromagnetic Fields, IEEE Press, 2010
6. A. Elsherbeni and V. Demir, The Finite-Difference Time-Domain Method for Electromagnetics with MATLAB Simulations, Scitech Publishng, 2009
2/25/2015FDTD by R. S. Kshetrimayum63
Simulations, Scitech Publishng, 20097. D. B. Davidson, Computational Electromagnetics for RF
and Microwave Engineering, Cambridge University Press, 2011
8. U. S. Inan and R. A. Marshall, Numerical ElectromagneticsThe FDTD Method, Cambridge University Press, 2011