Indonesi
an
Journa
l
of El
ect
ri
cal Engineer
ing
an
d
Comp
ut
er
Scie
nce
Vo
l.
1
3
,
No.
3
,
Ma
rch
201
9
, p
p.
8
53
~
860
IS
S
N: 25
02
-
4752, DO
I:
10
.11
591/ijeecs
.v1
3
.i
3
.pp
8
53
-
8
60
853
Journ
al h
om
e
page
:
http:
//
ia
es
core.c
om/j
ourn
als/i
ndex.
ph
p/ij
eecs
Optimi
satio
n
of
zinc o
xide surg
e arr
este
r design usi
ng
gravitati
onal se
arch al
gorithm an
d im
perialist
comp
etitive
algorith
m
Syahira
h
Ab
d
Ha
li
m
1
,
Az
ah M
ohamed
2
,
Nor
Az
w
an
M
ohamed
K
am
ar
i
3
, A
fid
a A
yob
4
, Ab H
alim
Ab
u
Bakar
5
,
H
az
le
e A
z
il
Ill
ias
6
1,2,3,4
Cent
re
for
I
nte
gra
te
d
Engi
n
e
eri
ng
S
y
stems
a
nd
Advanc
ed
T
e
chnol
ogi
es,
Uni
ver
siti
Keba
ngsa
an
Mal
a
y
s
ia,
Sela
ngor,
Mal
a
ysia
5
UM
Pow
er
Energ
y
Dedicat
ed
A
dvanc
ed
Centre (U
MP
EDAC),
W
i
sm
a
R&D
UM
,
Univer
sit
i
M
al
a
y
a
,
Kuala L
u
m
pur,
Malay
s
ia
6
Depa
rtment of
El
e
ct
ri
ca
l
Eng
in
ee
ring
,
Fa
cul
t
y
o
f
Engi
n
ee
ring
,
Univer
siti
Mal
a
y
a
,
Kuala L
um
pur,
Malay
s
ia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
S
e
p
30
, 201
8
Re
vised
N
ov
2
7
, 2
018
Accepte
d
Dec
15
, 201
8
Reduc
ing
e
lectr
i
c
fie
ld
str
ess
near
the
ene
rg
ised
e
nd
of
surge
arr
este
rs
is
ve
r
y
important
be
ca
u
se
it
m
a
y
inc
r
eas
e
the
l
ife
t
ime
of
the
high
l
y
st
ressed
ZnO
col
um
n
in
vic
in
ity
of
th
e
high
volt
ag
e
el
e
ct
rod
e.
Mos
t
of
pre
vious
works
were
base
d
on
m
anuf
ac
turers
’
proc
edur
es
and
tri
al
and
err
or
m
et
hod
to
improve
the
sur
ge
arr
este
r
desig
ns.
In
th
is
work,
opti
m
isat
ion
of
ZnO
surge
arr
este
r
design
m
odel
s
using
Gravi
t
at
ion
al
Se
a
rch
Algorit
hm
(
GS
A)
and
Im
per
ia
li
st
Com
pet
itive
Algori
t
hm
(ICA)
is
proposed.
The
su
rge
arr
est
er
m
odel
s
were
de
vel
oped
using
f
ini
te
el
emen
t
a
naly
s
is
(FEA)
and
used
to
det
ermine
the
elec
tr
ic
fi
el
d
d
istribution.
The
op
tim
isat
ion
m
et
hod
s
were
used
to
determ
ine
the
arr
este
r
d
esign
p
ara
m
et
ers
whi
ch
y
i
el
d
th
e
m
ini
m
um
el
ec
tr
ic
fie
ld
stress
surrounding
the
ene
r
giz
ed
end
of
th
e
surge
arr
este
rs.
GS
A
is
le
ss
complex
since
it
req
uire
s
onl
y
t
wo
par
amete
rs
to
be
adj
usted
i.e.
m
ass
and
vel
oc
ity
whil
e
I
CA
demons
tra
tes
faste
r
conv
erg
enc
e
and
b
et
t
er
ac
hi
eve
m
ent
of
global
opt
imum
.
Th
e
per
for
m
anc
e
of
the
p
roposed
m
et
hod
s
was
the
n
compare
d
with
the
m
anuf
ac
tu
rer
’s
te
st
d
ata
and
pre
v
iousl
y
d
eve
lop
ed
m
et
hods.
Ke
yw
or
d
s
:
Arrester
desi
gn
Ele
ct
ric fiel
d o
pti
m
iz
at
ion
Finit
e ele
m
ent
analy
sis
Copyright
©
201
9
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
Syahira
h Abd
Halim
,
Ce
ntre fo
r In
te
gr
at
e
d
E
ng
i
nee
rin
g
Sys
te
m
s an
d A
dvance
d Tec
hnol
og
ie
s
,
Un
i
ver
sit
i Ke
ba
ngsaan
Mal
ay
sia
, S
el
an
gor,
Ma
la
ysi
a
.
Em
a
il
:
sya
hirah
_h
@ukm
.ed
u.m
y
1.
INTROD
U
CTION
Zinc
oxi
de
(Z
nO)
s
urge
a
rr
e
ste
rs
ha
ve
bee
n
widely
us
e
d
nowa
days
in
ov
e
r
vo
lt
age
pr
otect
ion
of
transm
issi
on
li
ne
syst
em
s
[1
]
.
The
m
os
t
rec
ently
dev
el
op
e
d
Z
nO
surge
a
rr
est
er
s
do
no
t
require
s
erial
gaps
ow
i
ng
to
t
heir
excell
ent
nonl
inear
prop
e
rtie
s
[
2].
Ne
ver
t
heless,
t
he
re
m
ov
al
of
the
serial
ga
p
ca
use
s
Zn
O
disk
s
bein
g
co
ntinuo
us
ly
stressed
by
AC
powe
r
f
reque
nc
y
vo
lt
age
[
3].
To
incr
ease
th
e
un
if
or
m
it
y
o
f
the
el
e
ct
ric p
ote
ntial
d
ist
rib
utio
n,
op
ti
m
isa
t
ion
of the elec
tric
fiel
d
inten
sit
y aro
und
t
he
s
urge a
rr
est
er
is
need
e
d.
Nu
m
erous
res
earches
relat
ed
to
dete
rm
in
at
ion
of
el
ect
ric
fiel
d
in
a
su
r
ge
ar
reste
r
ha
ve
bee
n
publishe
d.
Han
et
al
.
descr
ibe
the
fo
rm
ulati
on
of
s
em
i
-
analy
ti
c
finite
e
leme
nt
m
e
thod
to
determ
ine
pote
ntial
distrib
ution
of
high
vo
lt
age
m
e
ta
l
ox
ide
arr
est
er
[
4].
Me
anwhil
e,
H
add
a
d
&
Nayl
or
c
om
pu
te
d
el
ect
ric
po
te
ntial
and
f
ie
ld
distrib
utio
n
of
a
ty
pical
po
ly
m
er
zi
nc
ox
i
de
s
urge
a
r
rester
by
m
ea
ns
of
finite
el
e
m
ent
m
et
ho
d
[
5].
E
konom
ou
et
al.
have
form
ulate
d
an
arti
fici
a
l
neu
ral
netw
orks
(AN
Ns)
ba
sed
m
et
ho
d
wh
ic
h
cou
l
d
be
us
ef
ul
in
determ
ining
el
ect
ric
fiel
d
distri
bu
ti
on
of
m
edium
vo
lt
age
s
urge
arr
est
er
s
f
or
de
si
gn
processes
and
diag
nosti
c test
s [6]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
1
3
, N
o.
3
,
Ma
rc
h
201
9
:
8
5
3
–
8
6
0
854
Howe
ver,
m
os
t
of
t
he
stu
dies
are
j
us
t
li
m
it
e
d
f
or
el
ect
ric
f
ie
ld
m
agn
it
ud
e
cal
culat
ion
.
O
nly
seve
ral
stud
ie
s
had
pre
sented
m
et
ho
ds
to
m
ini
m
iz
e
el
ect
ric
fiel
d
in
te
ns
it
y
of
s
urg
e
arr
est
er
s
duri
ng
norm
al
op
e
rati
ng
conditi
ons.
He
et
al
.
discuss
e
d
their
fin
ding
s
on
t
he
el
ect
r
ic
po
te
ntial
dis
tribu
ti
on
of
a
n
AC
1000
kV
ultra
high
volt
age
(
UHV)
m
et
al
ox
ide
sur
ge
arr
e
ste
r
[3
]
.
The
st
ud
y
ha
d
f
ocu
s
ed
on
analy
sin
g
an
d
i
m
pr
oving
the
el
ect
ric
po
te
ntial
distrib
ution
of
t
he
U
H
V
su
r
ge
a
rr
est
e
r
by
the
a
pp
li
ca
ti
on
of
th
ree
di
ff
ere
nt
ap
proa
ches;
fiel
d
-
ci
rc
uit co
m
bin
at
ion
, thre
e
-
dim
ension
al
(3D) fi
nite el
em
ent
m
et
ho
d (
FEM) a
nd circ
uit analy
sis.
Ku
m
ar
&
M
ogaveera
ha
d
at
te
m
pted
to
i
m
prov
e
the
volt
ag
e
distrib
utio
n
unif
or
m
ity
of
a
220
kV
a
nd
a
400
kV
s
urg
e
arr
est
ers
unde
r
cl
ean
an
d
poll
uted
e
nv
i
ron
m
ent
by
m
eans
of
n
um
erical
fiel
d
com
pu
ta
ti
on
[7
]
.
In
the
stu
dy,
de
sign
opti
m
isa
ti
on
of
the
sur
ge
a
rr
est
e
rs’
ge
om
et
ries
wh
i
ch
i
nclu
de
gr
a
ding
rin
g
a
nd
sp
acer
s
was
co
nducte
d
in
order
t
o
inv
e
sti
gate
their
infl
uen
ce
on
the
vo
lt
age
di
stribu
ti
on
un
de
r
norm
al
op
erati
ng
conditi
ons.
In
t
he
past
w
orks,
t
he
el
ect
ric
fiel
d
distrib
ution
of
a
s
urge
a
rr
est
er
great
ly
dep
e
nds
on
the
ge
om
et
ry
desig
n
a
nd
el
e
ct
rical
pr
ope
rtie
s
of
t
he
go
vernin
g
m
at
erial
.
The
m
ini
m
isa
tio
n
of
t
he
el
ect
r
ic
fiel
d
m
agn
itu
de
of
the
ar
re
ste
r
w
as
achieve
d
by
proposing
s
ever
al
desig
n
m
et
ho
ds.
H
ow
ever,
w
orks
on
us
i
ng
opti
m
isa
ti
on
m
et
ho
ds t
o
m
i
nim
ise
the elect
ric fiel
d
m
agni
tud
e a
re less li
kely
to be
fou
nd in
li
te
ratur
e
.
Ther
e
f
or
e,
i
n
this
w
ork,
m
inim
isi
ng
the
el
ect
ric
fiel
d
m
ag
nitud
e
i
n
tw
o
Zn
O
surge
a
rrest
er
desi
gn
m
od
el
s
is
pr
opos
e
d
by
us
in
g
Gr
a
vitat
ion
al
Searc
h
Algorit
hm
(G
SA
)
a
nd
Im
per
ia
li
st
Com
pet
it
ive
Algorithm
(I
CA
).
T
he
surge
ar
rester
m
od
el
s
we
re
de
velo
ped
us
in
g
finite
el
e
m
ent
analy
sis
(F
EA)
to
determ
ine
the
el
ect
ric
fiel
d
distrib
ut
ion.
T
o
validat
e
the
per
f
orm
ance
of
the
pr
opose
d
m
et
ho
d,
c
om
par
ison
with
the
m
anu
fact
ur
e
r’
s
te
st
data
and
oth
e
r
opti
m
isa
t
ion
m
et
ho
ds
w
ere
m
ade.
Using
opti
m
isa
ti
on
m
et
ho
ds,
it
is
m
or
e
conve
nient
to
perform
design
opti
m
isa
ti
on
on
t
he
s
urge
arr
est
er
ge
om
e
try
instea
d
of
changin
g
t
he
m
at
erial
pro
per
ti
es
by tr
ia
l and
e
rro
r
to
obtai
n
the
desired el
ect
ric fiel
d dist
rib
ution
.
2.
MO
DELL
IN
G OF
S
URGE A
RR
EST
E
R IN FE
A
Series
of
te
sts
wer
e
car
ried
out
on
tra
nsm
issi
on
li
ne
a
rr
est
er
ha
ving
rated
volt
age
of
12
0
kV.
Table
1
s
p
eci
f
ie
s
the
c
har
act
erist
ic
s
of
the
su
r
ge
a
rrest
er
in
this
w
ork,
as
sp
eci
fied
in
the
m
anufacture
r’s
datasheet.
Fig
ur
e
1
s
hows
t
he
tw
o
-
dim
ension
al
(
2D)
a
xi
al
sy
m
m
et
ric
m
od
el
geo
m
etr
y
of
the
Z
nO
arr
est
e
r
that
has
been
de
velo
ped
in
th
e
FEA
s
oft
ware.
The
‘Elect
r
ost
at
ic
’
interfac
e
of
t
he
AC/
D
C
m
od
ule
a
vaila
ble
in
the
softwa
re
w
as
us
e
d
to
s
olve
the
m
od
el
.
A
la
ye
r
of
ai
r
s
urr
oundin
g
the
arr
est
er
was
al
so
dra
wn
t
o
ob
serve
the
el
ect
ric
field
distrib
utio
n
on
the
s
urface
of
the
s
heds.
Ma
te
rial
pr
ope
rtie
s
as
sh
own
in
Ta
ble
2
we
r
e
then
assigne
d
t
o
eac
h do
m
ai
n.
Table
1.
Elec
tr
ic
al
an
d
I
ns
ulat
ion
Data
for A
r
rester
Para
m
eter
Valu
e
Rated
vo
ltag
e
1
2
0
kV
Maxi
m
u
m
r
esid
u
al
vo
ltag
e (
1
0
kA 8
/2
0
µs
curre
n
t)
3
1
6
kV
Heig
h
t
7
5
0
m
m
Ins
u
latio
n
m
a
terial
Silico
n
e r
u
b
b
er
Creepag
e
d
istan
ce
1
6
4
0
m
m
Figure
1. 2D a
xial sy
m
m
et
ric
al
o
f
ZnO
surg
e arr
e
ste
r
m
odel
Table
2.
Mat
er
ia
l Proper
ti
es
of the
S
urge
Arr
est
er C
om
pone
nts
Do
m
ain
Mater
i
al
Relativ
e per
m
ittiv
ity
,
ε
r
Co
n
d
u
ctiv
ity
,
σ
(S
m
-
1
)
1
Alu
m
in
iu
m
1
3
.77
4
x
1
0
4
2
Zinc o
x
id
e
2250
No
n
-
lin
ear
3
Silico
n
e r
u
b
b
er
3
.6
1x10
-
18
4
Fib
reglass
4
.2
1x10
-
18
5
Air
1
0
1
5
4
1
2
3
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Op
ti
mis
atio
n o
f zi
nc oxide s
urge
ar
reste
r
de
s
ign
us
i
ng g
r
avi
tati
onal se
ar
ch
… (
Syahira
h
A
bd Hali
m
)
855
Af
te
r
as
sig
ning
the
m
at
erial
pro
per
ti
es,
rele
van
t
inte
rf
ace
conditi
ons
we
r
e
app
li
ed
at
ea
ch
bo
undar
y
in
the
ge
om
etr
y.
The
gro
un
d
bo
unda
ry
conditi
on
was
app
li
ed
at
the
bo
tt
om
par
t
of
th
e
ar
rester
,
w
hich
sp
eci
fies
zer
o
el
ect
ric
po
te
ntial
.
The
uppe
r
pa
rt
of
the
arre
ste
r
was
sp
eci
fied
as
a
te
r
m
inal
so
urce
of
sta
ti
on
a
r
y
el
ect
ric
po
te
ntial
.
Since
t
he
ar
rester
was
e
ncl
os
e
d
in
a
sphe
r
ic
al
ai
r
do
m
ai
n,
it
s
oute
rm
os
t
bounda
ry
co
nd
it
ion
was
assi
gned
a
s
el
ect
ric
ins
ulati
on
. Th
e
inte
rf
ace
c
onditi
on
of
the
rem
ai
ni
ng
bo
unda
ries w
as
set
as
c
on
t
inu
it
y.
The
nex
t
ste
p
was
to
m
esh
the
ge
om
et
ry
i
nto
sm
al
le
r
ele
m
ents
for
ef
f
ect
ive
com
pu
ta
ti
on
.
In
t
his
m
od
el
,
extrem
el
y
fine
m
esh
el
e
m
ent
s
wer
e
a
ppli
ed
on
the
area
s
urrou
nd
i
ng
the
ZnO
col
um
n.
This
was
done
to
increase
the
preci
sion
of
the
el
ect
ric
fiel
d
distribu
ti
on
as
the
Zn
O
col
umn
determ
ines
the
be
hav
i
our
of
the
arr
est
er
.
3.
DESIG
N OPT
IMISATIO
N USI
NG IC
A AND
GS
A
The
s
urge
ar
re
ste
r
de
sig
n op
ti
m
isa
ti
on
p
r
oce
dures use
d
i
n
t
his wo
rk are
as
foll
ow
s:
Step
1: T
he
a
rrest
er w
a
s m
odel
le
d
in F
EA
s
of
t
war
e t
o
cal
c
ulate
the elect
r
ic
f
ie
ld
distrib
ut
ion
.
Step
2: T
he
m
od
el
w
a
s inter
fa
ced
with
GSA
and IC
A
al
go
rithm
s f
or
desig
n op
ti
m
isa
t
ion
.
Step
3: T
he op
t
i
m
isa
t
ion
was r
epeate
d u
ntil
the c
onve
rg
e
nc
e crit
eria are
r
e
ached.
The
op
ti
m
ise
d
var
ia
bles
m
us
t
resu
lt
in
m
i
nim
u
m
el
ec
tric
fiel
d
m
agn
it
ud
e
at
re
gions
wh
ic
h
a
re
consi
der
e
d
crit
ic
al
on
the
surge
ar
rester
.
I
n
this
case,
the
m
axi
m
u
m
electric
fiel
d
on
the
sil
ic
on
e
ru
bb
e
r
su
r
face
m
us
t
be
equ
al
or
le
ss
than
the
el
ect
ric
streng
th
of
ai
r,
i.e.
3
kVm
m
-
1
,
to
avo
id
a
ny
su
r
face
disc
harge
occurre
nce.
T
hus,
t
he o
bj
ect
iv
e f
un
ct
io
n
is
de
fine
d
as:
J
=
{
−
3
−
1
}
(1)
wh
e
re
Em
ax
is
the
m
axi
m
u
m
el
ect
ric
fiel
d
(in
kVm
m
-
1
)
po
int
al
ong
the
ar
rester
axis,
w
hi
ch
is
locat
ed
on
the
ti
p
of
th
e
ins
ul
at
ion
hous
i
ng
sh
e
d
near
to
t
he
hi
gh
volt
age
el
ect
ro
de
.
T
he
const
raints
de
f
ined
co
ns
ide
ri
ng
the
lim
it
s
relat
ive
to
arr
est
e
r’
s
di
m
ension
a
nd
m
at
erial
pr
op
e
rtie
s
are
sho
w
n
in
Ta
ble
3.
Th
ere
a
re
no
a
vaila
ble
sp
eci
fic
sta
nd
a
rd
s
relat
ed
to
the
s
urge
a
rr
es
te
r’
s
desig
n,
t
hu
s
the
va
riat
ion
lim
it
s
of
th
e
desi
gn
pa
ra
m
et
ers
wer
e
chose
n b
ased
on the
sim
ula
ti
on
r
e
su
lt
s and t
he geom
et
ric feasibil
it
y.
Table
3.
Lim
it
s
Rel
at
ive to Ar
rester
Dim
ensi
on a
nd Mat
eria
l P
r
op
e
rtie
s
Para
m
eter
1
2
0
kV arr
ester
Lower li
m
it
Up
p
er
li
m
it
Intial v
alu
e
Rad
iu
s o
f
FRP rod
0
.1 c
m
2
.0 c
m
1
.0 c
m
Rad
iu
s o
f
ins
u
latio
n
ho
u
sing
2
.0 c
m
4
.0 c
m
3
.0 c
m
Relativ
e per
m
ittiv
ity
of
silico
n
e r
u
b
b
er
5
.0
7
.0
6
.0
Relativ
e per
m
ittiv
ity
o
f
FRP rod
3
.7
1
0
.0
4
.2
3
.
1
.
Gr
avit
ati
on
al
Searc
h
Algori
th
m
(G
SA
)
Gr
a
vitat
ion
al
Searc
h
Algorit
hm
inv
ol
ves
a
group
of
sear
cher
a
gen
ts,
w
hich
interact
be
tween
them
thr
ough
the
gravit
y
force.
T
he
age
nts,
al
so
known
as
obj
e
ct
s,
are
m
easur
ed
i
n
te
rm
of
pe
rfor
m
ance
ac
cordi
ng
to their m
asses. Th
is
gravity
for
ce
cause
s m
ov
em
ent o
f
t
he object
s t
ow
a
rds o
the
r o
bj
ect
s
with
heav
ie
r
m
asses
.
The flo
wch
a
rt
of the
GSA m
et
hod
is
de
picte
d
in
Fig
ure
2
[
8,
9].
3
.
2
.
Im
peri
al
ist Compe
titi
ve
A
lg
orit
hm
(I
CA)
In
this
al
go
ri
th
m
,
the
popula
ti
on
i
nd
i
viduals
are
ref
e
rred
as
country.
In
it
ia
ll
y,
so
m
e
of
the
countries
with
bette
r
ob
j
ect
ive
functi
on
are
assig
ne
d
as
im
per
ia
li
st
s
w
hile
oth
e
rs
are
assi
gn
e
d
as
colo
nies
t
ha
t
al
l
tog
et
he
r
f
orm
s
om
e e
m
pires.
D
uri
ng the c
om
pet
it
ion
, th
e
pow
e
r
of the
dom
inant e
m
pire
will
incr
ease,
an
d
t
he
powe
r
of
the
weak
em
pire
will
decr
ease.
The
su
it
able
e
m
pires
will
ta
ke
po
sses
sio
n
of
their
colo
nies
.
At
th
e
end
of
the
com
petit
ion
,
the
re
will
be
on
ly
one
i
m
per
ia
li
st
a
nd
oth
er
s
will
be
it
s
colon
ie
s.
The
flo
wch
a
rt
of
the
ICA
m
et
ho
d
is
dep
ic
te
d i
n Fi
gure
3 [
10,
11
]
.
4.
EL
ECTRIC F
IEL
D
D
ISTR
I
BUTION
I
N
S
URGE A
R
RE
STE
R
The
el
ect
ric
fiel
d
distrib
uti
ons
in
the
rated
120
kV
surge
arr
est
er
m
od
el
geo
m
et
ry
wh
en
el
ect
ric
po
te
ntial
of
1
32
kV w
as
ap
plied
on
the
e
ne
r
gized
e
nd
a
re
i
ll
us
trat
ed
in
Fi
gure
4
a
nd
Fig
ur
e 5
.
T
he
c
ut
li
ne
2D
plo
t
of the elec
tric
f
ie
ld al
ong t
ip of th
e
ho
us
ing sh
eds
is il
lu
strat
ed
in
F
ig
ur
e 6
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
1
3
, N
o.
3
,
Ma
rc
h
201
9
:
8
5
3
–
8
6
0
856
Figure
2. Flo
w
char
t
of Gra
vitat
ion
al
Sea
rch
Algorithm
Figure
3. Flo
w
char
t
of I
m
per
i
al
ist
Co
m
petit
i
ve Alg
or
it
hm
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Op
ti
mis
atio
n o
f zi
nc oxide s
urge
ar
reste
r
de
s
ign
us
i
ng g
r
avi
tati
onal se
ar
ch
… (
Syahira
h
A
bd Hali
m
)
857
Figure
4. Ele
ct
ric fiel
d dist
ri
buti
on of
120 k
V
a
rr
est
er
Figure
5.
Ele
ct
ric fiel
d dist
ri
buti
on on t
he
ti
p o
f
s
hed
s
It
is
obvious
t
ha
t
the
el
ect
ric
fiel
d
distrib
utio
n
al
ong
t
he
s
he
ds
is
not
unif
orm
wh
ere h
ig
h
m
agn
it
ud
e
s
wer
e
obse
rv
e
d
m
a
inly
on
the
ti
p
of
t
he
s
he
ds
wh
ic
h
a
re
ne
arer
t
o
e
nergi
zed
e
nd
te
rm
inal.
This
is
due
to
the
diff
e
re
nt
m
at
erial
s
in
the
su
rge
arr
est
er
’s
str
uctu
re,
w
hich
hav
e
diff
e
re
nt
per
m
it
t
ivit
y
an
d
boun
dar
y
co
nd
it
io
n.
Re
fer
ri
ng
t
o
F
igure
6,
t
he
el
ect
ric
fiel
d
m
a
gn
it
u
de
al
so
s
hows
a
s
udde
n
increa
se
w
he
n
there
is
a
r
egio
n
transiti
on fro
m
air to
sil
ic
on
e
rub
ber
.
At the inter
fac
e b
et
wee
n
ai
r
a
nd
t
he
ho
us
in
g sheds, the
re is a d
isc
onti
nu
it
y of
the elect
ri
c fiel
d
due to
the
pe
rm
itti
vit
y
diff
e
re
nce
be
tween
both
m
at
erial
s.
The
r
el
at
ive
per
m
it
t
ivit
y,
ε,
descr
i
bes
t
he
a
bili
ty
of
th
e
m
at
erial
to
al
i
gn
t
he
el
ect
ric
fiel
d.
A
m
at
eri
al
with
highe
r
per
m
it
t
ivit
y
has
higher
a
bili
ty
to
al
ign
the
el
ect
ric
fiel
d,
th
us
resul
ti
ng
in
lo
wer
el
ect
ric
fiel
d
con
ce
ntrati
on
in
the
m
a
te
rial
.
In
this
case,
t
he
relat
ive
pe
rm
i
tt
ivit
y
of
sil
ic
one
r
ub
ber
is
norm
al
l
y
hig
he
r
tha
n
the
ai
r
re
gion.
Ther
e
f
or
e,
t
he
el
ect
ric
fiel
d
betwee
n
ti
ps
of
t
he
hous
i
ng
s
heds
and
t
he
ai
r
re
gi
on
is
m
or
e
intense
tha
n
in
th
e
inn
er
st
ru
ct
ure
of
t
he
housi
ng
s
he
ds
.
It
is
al
so
ob
s
er
ved
that t
he
el
ect
ric fiel
d
intensit
y dec
reases at the locat
ion
fur
t
her
than
the
hous
i
ng sh
e
ds
, m
ai
nly i
n
the
reg
i
on of air
.
Table
4
prese
nts
the
or
i
gin
a
l
and
opti
m
ised
pa
ram
et
ers
fo
r
FRP
r
od
an
d
sil
ic
on
e
r
ub
ber
hous
in
g
m
at
erial
s
of
th
e
120
kV
surg
e
arr
e
st
er,
cal
culat
ed
by
the
GSA
a
nd
IC
A
m
e
tho
ds.
All
op
ti
m
iz
ed
par
a
m
et
ers
wer
e
determ
ined
based
on
t
he
e
valuati
on
of
t
he
m
ini
m
u
m
value
of
t
he
obj
ect
ive
f
unct
ion.
T
he
optim
ise
d
dim
ension
pa
r
a
m
et
er
values
for
the
G
SA
a
lgorit
hm
on
ly
sh
ow
sli
ght
de
viati
on
s
from
t
he
opti
m
ise
d
va
lues
cal
culat
ed by t
he ICA al
gorithm
.
Table
5
pr
ese
nts
the
m
axi
m
u
m
e
le
ct
ric
fiel
d
al
ong
the
ar
rester
a
xis
ob
ta
ine
d
with
or
i
gin
a
l
dim
ension
s
a
nd
op
ti
m
ise
d
pa
ram
et
ers
of
t
he
120
kV
sur
ge
arr
est
er
.
It
is
obvi
ous
that
t
he
us
e
of
t
he
op
ti
m
ise
d
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
1
3
, N
o.
3
,
Ma
rc
h
201
9
:
8
5
3
–
8
6
0
858
dim
en
sion
an
d
relat
ive
per
m
it
ti
vity
values
of
the
FRP
r
od
an
d
the
sil
i
con
e
r
ubbe
r
hous
i
ng
sign
ific
antly
reduces
t
he
m
axim
u
m
el
ect
r
ic
fiel
d
on
t
he
ti
p
of
the
ho
us
in
g
s
he
d,
w
hich
is
nea
rer
to
the
e
ne
rg
iz
ed
e
nd
el
ect
ro
de
.
Im
ple
m
enting
the
optim
ise
d
value
s
fr
om
the
ICA
al
gorithm
resu
lt
s
in
sli
gh
tl
y
lower
el
ect
ric
fiel
d
intensit
y com
par
ed
to
t
he GS
A op
ti
m
ise
d
va
lues.
Figure
6.
2D c
ut li
ne of
elec
tr
ic
f
ie
ld alo
ng t
he
ti
p o
f
s
heds
Table
4.
In
it
ia
l
A
nd Opti
m
ise
d Desig
n Para
m
et
ers
of t
he Ar
rester
Para
m
eter
1
2
0
kV
Initial v
alu
e
GSA valu
e
ICA valu
e
Rad
iu
s o
f
FRP rod
(c
m
)
1
.0
1
.34
1
.13
Rad
iu
s o
f
ins
u
latio
n
ho
u
sing (c
m
)
3
.0
2
.07
2
.37
Per
m
i
ttiv
ity
of
FR
P r
o
d
4
.2
9
.52
7
.96
Per
m
i
ttiv
ity
of
sili
co
n
e r
u
b
b
er
6
.0
6
.69
6
.71
Table
5.
Ma
xi
m
u
m
Ele
c
tric
Fiel
d
Al
ong
t
he
A
rr
est
er
Axis
Mod
el
Maxi
m
u
m
elect
ric
f
ield
(
k
V
m
m
-
1
)
Initial v
alu
e
GSA Opti
m
ised
Red
u
ctio
n
us
in
g
GSA
ICA Opti
m
ised
Red
u
ctio
n
us
in
g
I
CA
1
2
0
kV
0
.98
3
8
0
.24
6
1
7
4
.98
%
0
.24
5
2
7
5
.08
%
5.
COMP
AR
I
S
ON WIT
H
M
ANUF
AC
T
U
RER
’S
D
ATA
SHEE
T
Fu
rt
her
validat
ion
on
the
acc
uracy
of
the
s
ur
ge
ar
rester
m
odel
was
m
ade
by
com
par
in
g
t
he
est
im
at
e
d
peak
resid
ual
vo
lt
age
s
with
t
he
m
anu
factu
r
er’
s
data.
T
he
m
easur
ed
data
for
the
re
sid
ua
l
vo
lt
ages
of
t
he
12
0
kV
s
urge
ar
rest
er
is
316
kV.
F
or
a
10
kA n
om
inal
discharg
e
cur
r
ent
of
8/
20
µs
,
the
resi
du
al
volt
ages obtai
ned
us
in
g
t
he
or
i
gin
al
dim
ension
par
am
et
ers
an
d
the
opti
m
ised
par
am
et
ers
s
how
only
sli
ght
dev
ia
ti
ons
from
the
m
anu
fact
ur
e
r’
s
d
at
a. Rel
at
ive error
s
of
t
he
di
scharge
vo
lt
ag
es f
or the
120 kV
a
rr
est
e
r
we
re th
en dete
rm
i
ned
t
o
eval
uate t
he pr
eci
sion
of the
a
rr
est
er
’s
c
ha
rac
te
risti
cs, calcula
te
d
usi
ng:
=
−
×
100%
(2)
wh
e
re
Vr
estimat
ed
is
the
est
i
m
a
te
d
discha
r
ge
vo
lt
age
a
nd
Vr
manufacture
r
is
t
he
volt
age
f
rom
the
m
anu
fa
ct
ur
er
’s
data.
As
ca
n
be
obs
erv
e
d
from
Ta
ble
6,
the
us
e
of
t
he
op
ti
m
is
ed
par
am
et
er
va
lues
in
both
s
urge
a
rr
est
e
r
m
od
el
s
red
uce
the
relat
ive
error
s
betwee
n
the
est
i
m
at
ed
and
m
anu
fact
urer
’s
data
of
the
resid
ual
vo
lt
ages.
Fu
rt
her
m
or
e,
both
opti
m
ise
d
par
am
et
ers
ob
t
ai
ned
f
ro
m
GS
A
an
d
ICA
it
erati
on
m
et
ho
d
al
so
repr
oduce
the
resid
ual
volt
ag
es
of
t
he
12
0
kV
a
rr
est
e
r
wit
h
lo
w
relat
ive
error.
The
good
ag
reem
ent
between
the
e
sti
m
at
ed
and
act
ual
m
ea
su
rem
ent
data
has
dem
on
strat
ed
that
desi
gn
op
ti
m
isa
t
ion
b
y
FEA
m
od
el
i
s
a
valid
opti
on
f
or
fu
t
ur
e a
rr
e
ste
r devel
op
m
ent.
Figure
7
s
how
s
the
el
ect
ric
fiel
d
distri
bu
ti
on
al
ong
t
he
120
kV
ar
rester
s
he
ds
for
or
i
gin
al
dim
ensio
n
par
am
et
ers,
G
SA
pa
ram
et
ers
and
IC
A
par
a
m
et
ers
resp
ect
ively
.
A
m
ino
r
reducti
on
in
th
e
m
axi
m
u
m
electric
fiel
d
m
agn
it
ude
was
obser
ve
d
from
GS
A
and
ICA
due
to
the
op
ti
m
ise
d
thickness
a
nd
relat
ive
per
m
i
tt
ivit
y
values
of
t
he
F
RP
la
ye
r
and
si
li
con
e
r
ubbe
r
hous
i
ng.
Ba
sed
on
t
hese
fi
gure
s,
disc
on
ti
nuit
y
of
the
el
ect
ric fiel
d
0
20
40
60
80
100
120
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
x
1
0
5
A
r
c
l
e
n
g
t
h
E
l
e
c
t
r
i
c
f
i
e
l
d
n
o
r
m
(
V
/
m
)
L
i
n
e
G
r
a
p
h
:
E
l
e
c
t
r
i
c
f
i
e
l
d
n
o
r
m
(
V
/
m
)
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Op
ti
mis
atio
n o
f zi
nc oxide s
urge
ar
reste
r
de
s
ign
us
i
ng g
r
avi
tati
onal se
ar
ch
… (
Syahira
h
A
bd Hali
m
)
859
betwee
n
the
ho
us
in
g
s
hed
a
nd
the
ai
r
dom
ain
wa
s
obse
rv
e
d
due
t
o
the
diff
e
ren
ce
i
n
the
per
m
it
t
ivit
y
value
of
sil
ic
on
e
r
ubbe
r
and
ai
r
.
Com
par
iso
n
bet
ween
the
pro
po
se
d
m
et
ho
d
a
nd
a
m
et
a
-
heu
risti
c
op
ti
m
isa
t
ion
m
et
hod
wh
ic
h
is
widel
y
us
ed,
t
he
pa
r
ti
cl
e
swar
m
op
tim
isa
ti
on
(P
S
O)
was
al
so
m
ad
e
[
12
-
15
]
.
F
r
om
Table
7,
it
can
be
seen
th
at
the
optim
isa
ti
on
m
e
thods
us
in
g
IC
A
a
nd
GSA
yi
el
d
lo
wer
m
inim
u
m
e
le
ct
ric
fiel
d
m
agn
it
ud
e
than
PSO.
This
s
ho
ws
that
IC
A
a
nd
GSA
are
a
bette
r
opti
m
isa
ti
on
m
et
ho
d
co
m
par
ed
to
PS
O.
T
he
feat
ur
e
s
wh
ic
h
m
ake
ICA
a
nd
GSA
ar
e
bette
r
tha
n
P
SO
a
re
good
c
onve
rgence
rate
,
hi
gh
qual
it
y
so
luti
on
a
nd
ca
pa
bili
ti
es
in
fin
ding local
optim
u
m
.
(a)
(b)
(c)
Figure
7. Ele
ct
ric fiel
d
m
agn
i
tud
e al
ong
t
he she
d of
120 k
V
arr
e
ste
r usin
g;
(
a)
origi
nal d
i
m
ension p
a
ram
et
ers,
(b) GS
A
par
a
m
et
ers
and (c
) ICA
pa
ram
et
er
s
Table
6.
Re
la
ti
ve
E
rrors Be
tw
een th
e
Esti
m
a
te
d
a
nd Ma
nuf
act
ur
er
’s Data
Mod
el
Res
id
u
al vo
ltag
e (
k
V)
Relativ
e erro
r
to
m
an
u
f
actu
rer
’s data
(%)
Initial
GSA
ICA
Inital
GSA
ICA
1
2
0
kV
3
1
7
.35
3
1
7
.28
3
1
7
.32
0
.42
0
.40
0
.41
0
20
40
60
80
100
120
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
x
1
0
5
A
r
c
l
e
n
g
t
h
E
l
e
c
t
r
i
c
f
i
e
l
d
n
o
r
m
(
V
/
m
)
L
i
n
e
G
r
a
p
h
:
E
l
e
c
t
r
i
c
f
i
e
l
d
n
o
r
m
(
V
/
m
)
0
20
40
60
80
100
120
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
x
1
0
5
L
i
n
e
G
r
a
p
h
:
E
l
e
c
t
r
i
c
f
i
e
l
d
n
o
r
m
(
V
/
m
)
A
r
c
l
e
n
g
t
h
E
l
e
c
t
r
i
c
f
i
e
l
d
n
o
r
m
(
V
/
m
)
0
20
40
60
80
100
120
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
x
1
0
5
L
i
n
e
G
r
a
p
h
:
E
l
e
c
t
r
i
c
f
i
e
l
d
n
o
r
m
(
V
/
m
)
A
r
c
l
e
n
g
t
h
E
l
e
c
t
r
i
c
f
i
e
l
d
n
o
r
m
(
V
/
m
)
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
1
3
, N
o.
3
,
Ma
rc
h
201
9
:
8
5
3
–
8
6
0
860
Table
7.
Mi
ni
m
u
m
Ele
c
tric
Fiel
d
O
btaine
d from
I
niti
al
a
nd
Op
ti
m
iz
ed
Param
et
ers
Mod
el
Mini
m
u
m
elect
ric
f
ield
(
k
V
m
m
-
1
)
Initial v
alu
e
GSA op
ti
m
ised
ICA op
ti
m
ised
PSO op
ti
m
ised
1
2
0
kV
0
.98
3
8
2
5
0
.24
6
1
3
0
0
.24
5
2
0
2
0
.25
6
5
6
6.
CONCL
US
I
O
N
In
this
wor
k,
t
he
opti
m
isa
ti
on
of
a
12
0
kV
Zn
O
surge
a
rr
e
ste
r
desig
n
m
od
el
us
in
g
GSA
and
ICA
has
been
s
ucces
sf
ully
propose
d.
The
s
urge
ar
rester
m
od
el
s
wer
e
s
uccess
f
ully
dev
el
ope
d
us
in
g
fi
nite
el
e
m
ent
analy
sis
an
d
use
d
t
o
determ
i
ne
t
he
el
ect
ric
fiel
d
distri
bu
t
ion
.
By
op
ti
m
iz
ing
t
he
relat
ive
per
m
it
ti
vit
y
an
d
rad
i
us
of
the
F
RP
rod
a
nd
th
e
sil
ic
on
e
r
ubbe
r
housi
ng,
the
el
ect
ric
fiel
d
intensit
y
su
rro
unding
the
ar
rest
er
ha
s
been
s
uccessfu
ll
y
red
uce
d
w
hi
le
m
ai
ntaining
it
s
ph
ysi
cal
prop
e
rtie
s.
W
it
h
op
ti
m
ise
d
arr
e
ste
r
dim
ensions,
the
lowest
c
on
ce
nt
rati
on
of
the
e
le
ct
ric
fiel
d
on
the
surf
ace
of
the
ar
rester
i
nsula
ti
on
housi
ng
can
be
ac
hi
eved,
wh
ic
h
in
tu
r
n
m
ay
increase
the
ser
vice
li
fe
sp
a
n
of
the
s
urge
ar
rester.
C
om
par
ison
of
th
e
pro
po
se
d
m
e
thod
sh
ows
t
hat
the
op
ti
m
ise
d
su
r
ge
ar
rester
des
ign
m
od
el
co
m
pl
ie
s
with
th
e
m
anu
fact
ur
e
r
’s
te
st
data
an
d
yi
el
ds
lowe
r
m
ini
m
um
el
ect
ric
fiel
d
m
agn
it
ud
e
at
the
ene
rg
iz
e
d
e
nd
of
th
e
surge
arr
est
er
.
It
wa
s
al
so
fou
nd
t
ha
t
the
m
ini
m
u
m
elec
t
ric fiel
d
is
lo
w
er
us
in
g ICA t
han GSA
and
PSO.
ACKN
OWLE
DGE
MENTS
The
a
uthor
s
would
li
ke
to
ack
nowled
ge
Faculty
of
En
gin
eeri
ng
a
nd
Buil
t
En
vi
ronm
ent
an
d
Un
i
ver
sit
i
Kebangsaa
n
Ma
la
ysi
a
for
t
he
facil
it
ie
s
and
fina
nc
ia
l
su
pp
or
ts
t
hr
ough
resea
rc
h
gr
a
nt
c
ode
num
ber
GGPM
-
2018
-
056.
REFERE
NCE
S
[1]
Le
e
SB
,
Lee
SJ
,
Le
e
BH.
Anal
ysis
of
The
rm
al
and
El
ec
t
rical
Properties
of
ZnO
Arreste
r
Bloc
k.
Curr
ent
Appl
ied
Phy
sics
.
2010;
1
0(1):
176
-
180
.
[2]
Hos
seini
SA
,
M
irz
a
ie
M,
Bar
for
oshi
T.
Reliab
il
i
t
y
Ana
l
y
s
is
of
Surge
Arreste
r
Loc
a
ti
on
Eff
ect
in
High
Volta
g
e
Subs
ta
ti
ons.
Ind
onesian
Journal
of
E
le
c
tric
al
En
gine
ering
and
C
omputer
Scienc
e
.
2014;
12(8):
58
14
-
5826.
[3]
He
JL,
Ze
ng
R,
Chen
SM
,
Gu
an
ZC.
Potentia
l
Distribut
ion
A
naly
s
is
of
Sus
p
ende
d
-
T
y
p
e
Me
ta
l
-
Oxid
e
Surge
Arreste
r
'
s.
I
EE
E
Tr
ansacti
ons on Power
Del
iv
er
y
.
2003;
18(4):
121
4
-
1220.
[4]
Han
SJ
,
Zou
J,
Gu
SQ
,
He
JL,
Yuan
JS
.
Cal
culati
on
of
The
Pot
ent
i
al
Distributi
on
of
High
Volta
ge
Metal
Oxide
Arreste
r
b
y
Us
i
ng
an
Im
prove
d
Sem
i
-
Anal
y
tic
Finit
e
Elem
ent
Method.
I
EEE
Tr
ansacti
ons
on
Magne
t
ic
s
.
200
5;
41(5):
1392
-
139
5.
[5]
Hadda
d
A,
Na
y
lor
P.
Finit
e
-
E
le
m
ent
Co
m
put
at
ion
of
C
apaci
ta
nc
e
Networks
in
Multi
p
le
-
E
l
ec
trod
e
S
y
st
ems
:
Applic
a
ti
on
to
ZnO
Surge
Arre
sters.
IE
EE
Proce
ed
ings
-
Sci
en
c
e
Me
asur
eme
nt
and
Technol
ogy
.
1998;
145:
129
-
135.
[6]
Ekonomou
L,
C
hristodoul
ou
C
A,
Mlade
nov
V.
Esti
m
at
ion
of
t
he
Elec
tri
c
Field
Across
Mediu
m
Volta
ge
Surg
e
Arreste
rs Us
ing A
rti
ficial Neural
Networks.
Enginee
ring
Appl
i
cations of
N
eural
N
et
works
.
2014
;
4
59:
227
-
236.
[7]
Kum
ar
U,
Mogave
er
a
V.
Vo
lt
ag
e
Distributi
on
Studie
s
on
ZnO
Arreste
r
s.
IEEE
Proc
e
edi
ngs
-
Gene
rati
on
Tr
ansm
i
ss
ion
and
Distributi
on
.
2
002;
149(4):
457
-
462.
[8]
Rashedi
E
,
Ne
z
amaba
di
PH
,
Sa
r
y
a
zdi
S.
GS
A:
A
Gravi
tational
Sear
ch
Algor
it
h
m
.
Information
Sci
en
ce
s
.
2009
;
179(13):
2232
-
2
248.
[9]
Za
m
ani
MK
M,
Mus
iri
n
I,
Om
ar
MS
,
Sulim
an
S
I,
Ghani
NA
,
K
amari
NA
.
Grav
it
ational
sea
rch
al
gorit
hm
base
d
te
chn
ique
for
vo
lt
ag
e
stabilit
y
improvem
ent
.
Indo
nesian
Journal
o
f
El
e
ct
ri
cal
Eng
i
nee
ring
and
Co
mputer
Sci
en
ce
.
2018;
9(1):
123
-
130.
[10]
Atashpa
z
-
Garg
a
ri
E
,
Luc
as
C.
Im
per
ia
li
st
Com
pet
itive
Algorithm
:
An
Algorithm
for
Optimiz
at
ion
Inspire
d
b
y
Im
per
ia
li
st
ic
Co
m
pe
ti
ti
on
.
I
EE
E
Congress
on
Evoluti
onary
Computati
on
.
2007;
4
661
-
4667.
[11]
Xing
B,
Gao
W
J.
Innova
ti
v
e
Com
puta
ti
onal
In
telli
gen
ce:
A
Rough
Guide
to
134
Cle
ver
Algori
th
m
s.
Sw
it
ze
rla
nd:
Springer;
2014
.
[12]
W
aha
b
MN
A,
Neft
i
-
Mez
ia
ni
S,
At
y
ab
i
A.
A
Com
pre
hensive
Revi
ew
of
Sw
ar
m
Optimiza
ti
on
Algorit
hm
s.
Plos
One
.
2015;
10(5
).
[13]
Ill
ia
s
HA
,
Chai
XR,
Baka
r
AH
A,
Mokhlis
H.
Tra
nsform
er
Inc
ipi
ent
Fau
lt
Pre
dic
ti
on
Us
ing
C
om
bine
d
Artifi
cial
Neura
l
N
et
work
and
Vari
ous
Part
ic
l
e
Sw
arm Optim
isat
ion
T
ec
hni
ques
.
P
los One
.
2015;
10(6).
[14]
Li
u
C
,
Du WB,
W
ang
W
X.
Particle Sw
arm Optim
iz
at
ion
with
Sc
al
e
-
Fre
e
In
te
ra
ctions.
P
los One
.
2014;
9(5).
[15]
Ali
AM
,
Ebra
h
i
m
MA
,
Hass
an
MM
.
Autom
at
ic
volt
ag
e
gen
era
t
ion
cont
ro
l
for
t
wo
are
a
power
s
y
stem
base
d
on
par
ticle
sw
arm
o
pti
m
iz
at
ion
.
Ind
onesian
Journal
of
El
e
ct
ri
cal
En
gine
ering
and
C
omputer
Sci
en
ce
.
2016;
2(1):
132
-
144.
Evaluation Warning : The document was created with Spire.PDF for Python.