Indonesi
an
Journa
l
of
El
ect
ri
cal Engineer
ing
an
d
Comp
ut
er
Scie
nce
Vo
l.
23
,
No.
1
,
Ju
ly
2021
, p
p.
471
~
47
8
IS
S
N: 25
02
-
4752, DO
I: 10
.11
591/ijeecs
.v
23
.i
1
.
pp
471
-
47
8
471
Journ
al h
om
e
page
:
http:
//
ij
eecs.i
aesc
or
e.c
om
Perform
ance of s
imi
larity
explicit
group it
eration f
or solvin
g 2D
un
stea
d
y conv
ection
-
diffu
s
ion equa
tion
Nu
r
Af
z
a M
at A
li
1
,
Ju
m
at S
ulaima
n
2
,
Az
ali Sa
udi
3
,
Nor
Syahid
a Moh
am
ad
4
1,2,4
Facul
t
y
of
Sc
ie
nc
e and
Natu
ra
l
Resourc
es,
Uni
ver
siti
Malay
si
a S
aba
h
(UM
S), M
al
a
y
s
ia
3
Facul
t
y
of
Com
puti
ng
and
Infor
m
at
ic
s,
Univ
ersi
ti
Ma
lay
sia
Saba
h
(UM
S),
Malays
ia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Ma
r
3
1,
2021
Re
vised
Jun
8
,
2021
Accepte
d
J
un
1
7
, 202
1
In
thi
s
pap
er,
a
s
imila
rity
finite
d
iffe
ren
ce
(SF
D)
soluti
on
is
addr
e
ss
ed
for
the
two
-
dimensional
(2D)
par
abolic
par
ti
a
l
diff
ere
nt
ia
l
equati
on
(PD
E),
spec
ifica
l
l
y
on
t
he
unstea
d
y
con
vec
t
ion
-
diffusio
n
proble
m
.
Stru
ct
uring
the
sim
il
ar
ity
tra
nsf
orm
at
ion
using
wave
var
i
ables,
we
red
uc
e
th
e
p
ara
bol
ic
PD
E
int
o
el
l
ipt
i
c
PD
E.
The
num
eri
c
al
solut
ion
of
t
he
cor
r
espondin
g
sim
il
ar
i
t
y
equa
t
ion
is
obta
i
ned
using
a
sec
o
nd
-
orde
r
ce
n
tra
l
SF
D
discre
ti
za
t
i
on
sche
m
e
to
get
the
sec
on
d
-
orde
r
SF
D
ap
proxima
ti
on
equ
at
ion
.
W
e
propose
a
four
-
point
sim
il
arit
y
expl
i
ci
t
group
(4
-
point
SEG)
it
er
at
iv
e
m
et
hod
as
a
num
eri
cal
soluti
on
of
th
e
la
rge
-
sca
l
e
an
d
sparse
li
n
ea
r
s
y
stems
der
iv
e
d
from
SF
D
discre
t
iz
a
ti
on
of
2D
unstea
d
y
con
vec
t
ion
-
diffusio
n
equa
t
ion
(CDE
).
To
show
the
4
-
point
SEG
it
erati
on
eff
icie
nc
y
,
two
itera
ti
v
e
m
et
hods,
such
as
Jac
obi
and
Gauss
-
Seide
l
(GS
)
it
er
ati
ons,
are
al
so
conside
red
.
Th
e
num
eri
ca
l
expe
riments
are
ca
rri
ed
out
usi
ng
thre
e
differe
nt
proble
m
s
to
i
ll
ustrate
our
proposed
it
erati
ve
m
et
hod'
s
pe
rfo
rm
anc
e.
Fina
lly
,
the
num
erica
l
result
s
show
ed
tha
t
our
proposed
i
te
r
at
i
ve
m
et
hod
is
m
ore
eff
icient
than
the
Jac
ob
i
and
GS
itera
ti
on
s in
t
erms
of
iter
at
ion
num
ber
an
d
execut
ion
t
ime.
Ke
yw
or
d
s
:
Convect
io
n
-
dif
fu
si
on equati
on
Partia
l dif
fer
e
nt
ia
l equ
at
io
n
Si
m
il
arity exp
l
ic
it
g
rou
p
Si
m
il
arity f
init
e d
if
fer
e
nce
Si
m
il
arity so
lu
ti
on
This
is an
open
acc
ess arti
cl
e
un
der
the
CC
B
Y
-
SA
l
ic
ense
.
Corres
pond
in
g
Aut
h
or
:
Ju
m
at
Su
la
i
m
a
n
Faculty
of
Scie
nce a
nd N
at
ur
a
l R
esources
Un
i
ver
sit
i M
al
ay
sia
Sab
a
h
Jal
an UMS,
88
400 K
ota K
i
na
balu, Sa
bah, M
al
ay
sia
Em
a
il
:
j
um
a
t@um
s.ed
u.
m
y
1.
INTROD
U
CTION
Convect
io
n
-
dif
fu
si
on
eq
uatio
n
(C
DE
)
is
one
of
the
m
os
t
chall
eng
i
ng
prob
le
m
s
and
f
re
qu
e
ntly
us
e
d
in
va
rio
us
bran
ches
of
en
gin
e
erin
g
a
nd
ap
plied
sci
e
nce,
e
spe
ci
al
ly
in
rad
ia
l
trans
port
in
a
por
ou
s
m
edium
[1
]
,
heat
trans
fer
i
n
a
nanof
l
uid
fill
ed
[
2],
heat
tr
a
ns
fe
r
in
a
dr
ai
ning
film
[3
]
,
and
water
tra
nsport
in
s
oil
[4
]
.
Also
,
the
ap
plica
ti
ons
of
CDE
can
be
fou
nd
in
[5
]
-
[
8].
D
ue
t
o
it
s
ap
plica
ti
on,
these
pr
oblem
s
hav
e
rec
ei
ved
extensi
ve
at
te
ntion,
a
nd
m
a
ny
rese
arc
her
s
at
tem
pted
to
so
lve
these
pr
ob
le
m
s
nu
m
erical
ly
to
achie
ve
t
he
lowest
com
pu
t
at
ion
al
com
plexity
and
highe
st
per
f
or
m
ance.
To
achie
ve
the
low
c
om
pu
ta
ti
on
al
com
pl
exity
,
there
a
re
m
any
stu
dies
on
sim
il
arit
y
so
luti
on
te
ch
niques
ha
ve
bee
n
e
xp
l
ored
by
m
any
res
earche
rs
a
nd
a
pp
li
e
d
in
pa
rtia
l
dif
fer
e
ntial
equa
ti
on
s
(PDEs)
.
For
in
sta
nce
,
A
fify
[
9]
pr
ese
nted
sim
il
arit
y
so
luti
on
s
in
m
agn
et
ohyd
rodynam
ic
,
wh
ic
h
is
ob
ta
in
ed
by
us
in
g
scal
in
g
trans
form
at
io
ns
the
n
so
l
ved
nu
m
erical
ly
by
us
in
g
the
sho
oting
t
echn
i
qu
e
with
four
t
h
-
orde
r
Runge
–
K
utta
integ
rati
on
sch
e
m
e.
M.
Sia
va
sh
i
et
al.
[10],
the
auth
or
s
obta
in
ed
a
sim
i
la
rity
so
luti
on
of
ai
r
and
na
noflui
d
i
m
pin
gem
ent
coo
li
ng
of
a
c
yl
ind
rical
por
ous
heat
sink
us
i
ng
si
m
il
arity
var
ia
bl
es
then
s
olv
e
d
nu
m
erical
ly
.
Usm
an
et
al.
[
11]
,
a
sim
i
la
rity
so
luti
on
of
the
water/m
agn
et
ite
na
noflui
d
m
od
el
le
d
P
DEs
su
bject
t
o
the
r
m
al
rad
ia
ti
on
and
L
or
e
ntz
f
or
ce
over
stret
chab
l
e
ro
ta
ti
ng
disks
is
ob
ta
ine
d
by
su
pport
in
g
preci
se
si
m
il
arity
transf
orm
at
i
on.
The
stu
di
es
on
the
sim
il
arit
y
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.
23
, N
o.
1
,
Ju
ly
2021
:
471
-
47
8
472
so
luti
on
can
al
so
be
f
ound
in
[12
]
-
[
15]
.
Ba
sic
al
ly
,
the
m
ai
n
idea
of
us
i
ng
a
si
m
i
la
rity
so
luti
on
is
to
re
duce
the
nu
m
ber
of
in
de
pende
nt
va
riables
of
a
syst
em
of
PDEs
at
le
ast
on
e
le
ss
than
that
of
th
e
or
igi
nal
eq
ua
ti
on.
Be
cause
of
th
e
low
com
pu
t
at
ion
al
c
om
pl
exity
and
high
c
om
pu
ta
ti
onal
perform
ance
of
this
m
eth
od,
it
s
i
m
ple
m
entat
io
n
is
of interest
.
Ma
ny
researc
he
rs
ha
ve
so
l
ve
d
the
CDE
usi
ng
the
finite
diff
e
ren
ce
m
et
ho
d
(FDM)
in
ad
diti
on
to
th
e
above
nu
m
erical
m
et
ho
ds
f
or
so
lvi
ng
t
he
pro
po
s
ed
C
DE
prob
le
m
as
no
te
d
in
the
fir
st
pa
r
agr
a
ph.
As
a
r
esult,
a
new
fi
nite
diff
e
re
nce
(FD)
disc
reti
zat
ion
sc
hem
e
has
been
propos
ed
by
com
bin
ing
se
ver
al
nu
m
erical
discreti
zat
ion
s
chem
es,
m
os
tly
fr
om
the
FD
schem
e
fa
m
ily.
For
instance
,
the
no
ns
ta
nd
ard
FD
[16
]
-
[
18]
,
th
e
expo
nen
ti
al
FD
[19
]
-
[
21
]
a
nd
pe
rturbati
onal
FD
[
22
]
are
im
ple
m
ented
f
or
so
l
ving
CDE
.
N
o
doubt
that
[16
]
-
[
22]
can
pr
ese
nt
ou
tst
a
nd
i
ng
pe
rfor
m
ance,
but
it
su
f
fer
s
fro
m
hig
h
com
pu
ta
ti
on
al
c
omplexit
y
because
the
obta
ined
a
ppr
ox
i
m
at
ion
eq
uatio
n
ge
ner
at
ed
a
seq
uen
ce
of
li
near
syst
em
s
fo
r
each
ti
m
e
lev
el
,
i
n
wh
ic
h
this
phen
om
eno
n
w
il
l
increase
c
om
pu
ta
ti
on
al
com
plexi
ty
.
T
he
com
pu
ta
ti
onal
com
plexity
of
al
gorithm
s
is
d
et
erm
ined
by
t
wo
norm
s:
the
nu
m
ber
of
nodes
that
ha
ve
t
o
be
e
xam
ined
and
t
he
opera
ti
on
al
cost
per
no
de.
The
m
or
e
li
ne
ar
syst
em
s,
the
m
or
e
node
s
w
il
l
be
use
d
that
will
be
af
fect
ed
by
c
om
pu
ta
ti
on
al
com
plexit
y.
In
conj
un
ct
i
on
w
it
h
new
FD
dis
creti
zat
ion
sc
hem
es
in
[16
]
-
[
22
]
,
we
ai
m
to
introd
uce
an
e
ntirel
y
new
FD
sc
he
m
e
based
on
com
bin
at
ion
of
the
sim
il
arity
and
FD
schem
es,
na
m
ed
as
si
m
i
la
rity
finite
diff
e
re
nce
(
SF
D)
disc
reti
zat
i
on sc
hem
es, to
inv
e
sti
gate it
s
f
easi
bili
ty
in
s
olv
in
g
C
DE.
Ba
sed
on
the
second
par
a
grap
h,
we
can
see
that
m
any
researc
her
s
ha
ve
pro
pose
d
a
new
F
D
discreti
zat
ion
schem
e
with
ge
tt
ing
hi
gh
c
om
pu
ta
ti
on
al
pe
rfor
m
ance
an
d
lo
wer
c
om
pu
ta
ti
on
al
c
omplexit
y.
Ap
a
rt
f
r
om
that,
we
at
tem
pt
to
e
xam
ine
t
he
feasibil
it
y
of
the
SF
D
di
screti
zat
ion
sc
hem
e
as
a
nu
m
erical
m
et
ho
d
to
s
ol
ve
2D
unste
a
dy
CDE.
Ba
se
d
on
the
pr
e
vi
ou
s
li
te
ratu
re
rev
ie
w
on
ap
plyi
ng
the
si
m
il
arity
so
luti
ons
i
n
th
e
first
pa
ra
gr
a
ph,
the
fin
ding
s
showe
d
t
hat
the
sim
il
arit
y
so
luti
on
te
ch
ni
qu
e
has
su
cces
sfu
ll
y
reduce
d
the
i
ndepe
ndent
va
riables
of
PD
E
s
an
d
the
n
tra
nsfo
rm
the
m
into
ordi
nar
y
di
ff
e
ren
ti
al
eq
ua
ti
on
(ODE).
It
m
e
ans
t
hat
the
c
om
pu
ta
ti
on
al
cost
of
the
sim
il
arity
te
chni
qu
e
will
dec
r
ease
an
d
m
ake
the
com
pu
ta
ti
on
al
com
plexity
of
the
sim
i
la
rity
a
ppr
ox
im
at
ion
equ
at
io
n
lo
w
c
om
par
ed
to
with
ou
t
us
i
ng
sim
i
l
arit
y
so
luti
ons.
Howev
e
r,
m
os
t
of
them
are
so
lvin
g
PD
E
s
us
i
ng
sim
il
ari
ty
s
olu
ti
ons
are
c
on
ce
r
ned
with
PD
E
pro
blem
s
being
re
duced
to
OD
E
pr
ob
le
m
s,
but
the
stu
di
es
con
ce
r
ning
reducin
g
para
bo
li
c
P
DEs
i
nt
o
the
corres
pondin
g
el
li
ptic
PD
Es
hav
e
not
been
fou
nd
ye
t.
T
he
m
a
in
idea
is
to
m
ai
ntain
the
dim
ension
of
t
he
m
od
el
so
that
the
cha
racteri
sti
c
of
the obj
ect
can
be
pr
e
ser
ve
d.
I
nspire
d
by
this
te
chn
i
qu
e'
s
low
c
om
pu
ta
ti
on
al
com
plexity
and
hi
gh
com
pu
t
at
ion
al
pe
rform
ance,
we
pro
po
s
e
ne
wly
SF
D
disc
reti
zat
ion
sc
hem
es.
W
i
th
that,
we
ap
plied
t
he
si
m
il
arity
so
luti
on
sp
eci
fical
ly
us
ing
wave
var
ia
bles
tra
nsfo
rm
at
ion
in
order
t
o
re
duc
e
2D
par
a
boli
c PD
E
s in deta
il
s 2D
un
ste
a
dy CD
E
pro
blem
into
2D ell
ipti
c PDE
s.
Th
e
2D
unste
a
dy
CDE
is
first
trans
form
ed
into
a
2D
el
li
ptic
PD
E
via
a
si
m
il
arity
so
luti
on
te
c
hniq
ue
to
so
lve
the
pro
po
se
d
pr
oblem
.
The
SFD
ap
pro
xim
a
tio
n
e
qu
at
i
on
i
s
then
f
or
m
ed
by
discreti
zi
ng
th
e
corres
pondin
g
el
li
ptic
PD
E
us
in
g
the
ne
w
ly
SFD
discre
ti
zat
ion
schem
es.
Since
t
he
SFD
a
ppr
oxim
at
ion
equ
at
io
n
produ
ces
a
la
rg
e
-
sca
le
and
s
par
se
li
near
syst
em
wi
th
it
s
m
at
rix
c
oeffici
ent,
an
e
ff
ect
ive
s
olv
e
r
m
us
t
so
lve
the
res
ulti
ng
la
rg
e
a
nd
s
par
se
li
near
sy
stem
.
Accor
ding
t
o
the
re
su
lt
s
in
[
23
]
-
[
25]
,
t
he
it
erati
ve
ap
proac
h
is
m
os
t
def
i
nitel
y
the
best
li
near
so
l
ver
f
or
la
rg
e
a
nd
s
pa
rse
li
nea
r
syst
e
m
s.
Seve
ral
it
erati
ve
m
et
ho
ds
f
or
so
lvi
ng
a
li
nea
r
syst
em
with
a
la
rg
e
-
scal
e
a
nd
sp
a
rse
c
oeffici
ent
m
at
rix
are
disc
us
se
d
i
n
the
li
te
ratu
re
.
The
i
m
ple
m
entat
io
ns
of
the
poi
nt
i
te
rati
on
f
a
m
ily,
su
c
h
a
s
su
cce
ssive
ov
e
r
-
r
el
axati
on
(SOR)
[
26
]
,
[
27
]
,
acce
le
rated
ov
er
-
r
el
axati
on
(
AO
R
)
[
28
]
,
[
29]
,
an
d
kaud
d
su
ccessi
ve
ove
r
-
relaxati
on
(
K
SO
R)
[
30
]
,
[
31]
,
can
be
us
e
d
to
s
ol
ve
this
li
near
m
et
ho
d.
E
va
ns
[
32
]
in
ven
t
ed
the
ex
plici
t
group
(E
G
)
it
erati
on
m
et
ho
d
f
or
ob
ta
ini
ng
an
a
ppr
ox
im
at
e
so
luti
on
by
gro
upin
g
the
li
near
syst
e
m
into
a
seq
uence
of
(
4x4)
li
near
sy
stems
base
d
on
t
he
c
oeffici
ent
m
at
r
ix'
s
char
act
eris
ti
cs.
Com
par
ed
to
the
G
S
it
er
at
i
on
,
t
his
blo
c
k
it
erati
ve
m
eth
od
i
s
faster.
Alth
ough
t
he
c
onve
r
ge
nce
rate
for
E
G
it
erati
on
has
bee
n
acce
le
rat
ed,
se
ve
ral
res
earche
rs
hav
e
create
d
new
ver
si
ons
of
the
E
G
it
erati
on
fam
ily,
i
nclu
ding
4E
G
SO
R
via
ni
ne
-
po
i
nt
Laplaci
an
(
4EGSOR
9L
)
[33],
4
po
i
nt
-
e
xp
li
ci
t
decou
pled
gro
up
(EDG
)
[34]
and
9
P
oi
nt
-
E
DGSOR
[
35]
in
w
hich
al
l
of
these
blo
c
k
it
erati
on
s
hav
e si
gn
i
fican
tl
y decreased
their co
nver
ge
nc
e rate. A
li
et
al.
[
36
]
and Be
e
et
a
l.
[37] u
s
ed
the EG
m
eth
od for
so
lvi
ng
the
2D
CDE
to
so
lve
the
li
near
syst
em
of
un
ste
a
dy
adv
ect
io
n
-
di
ffusio
n
pro
blem
it
erati
vely
,
an
d
their
resu
lt
s
ind
ic
at
ed
that
this
m
et
hod
has
a
fa
st
con
ve
r
gen
ce
du
e
to
the
la
r
ge
num
ber
of
po
i
nts
that
m
us
t
be
treat
ed
sim
ult
aneously
.
H
oweve
r,
withou
t
us
ing
sim
il
a
rity
so
luti
on
t
echn
i
qu
e
s,
the
hig
h
c
om
pu
ta
ti
on
al
com
plexity
will
occu
r
t
o
ge
t
the
appr
ox
i
m
at
e
so
luti
on
of
t
he
2D
unste
ady
CD
E.
To
ac
hieve
t
he
lo
w
com
pu
ta
ti
on
al
com
plexity
, w
e p
r
opose
a
ne
w varia
nt
of th
e EG
it
erati
on
fam
i
ly
to
so
lve
2D
un
ste
a
dy
CDE
by
so
lvi
ng the
ge
ne
rated linea
r
sy
stem
.
As
a
re
su
lt
,
t
he
rem
ai
nd
er
of
t
his
pa
per
fo
c
us
es
on
ev
al
uating
t
he
e
ff
ic
acy
of
the
four
-
po
i
nt
si
m
il
arity
exp
l
ic
it
gr
ou
p
(4
-
po
i
nt
SE
G)
it
erati
ve
m
et
ho
d
for
s
olv
i
ng
the
syst
em
of
SF
D
a
ppr
ox
i
m
at
ion
equ
at
io
ns,
w
hi
ch
wa
s
ins
pire
d
by
th
e
ne
wl
y
dev
el
ope
d
S
FD
disc
reti
zat
ion
schem
e.
T
he
com
bin
at
ion
of
t
he
SFD
sc
hem
e
a
nd
the
E
G
it
erati
ve
m
et
ho
d
can
be
us
e
d
to
form
ulate
the
4
point
-
S
EG
it
erati
ve
m
e
tho
d.
To
il
lustrate
the
f
easi
bili
ty
of
4
-
po
i
nt
SE
G
it
er
at
ion
,
t
he
f
ollow
i
ng
ge
ner
al
eq
uation
f
or
2D
unste
ady
CDE
i
s
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
Perf
orma
nce
of
simil
ar
it
y exp
li
ci
t gro
up it
er
ation f
or
so
lv
in
g 2D
…
(
Nu
r Af
za
M
at Ali
)
473
consi
d
ere
d
i
n
t
his
pa
pe
r
to
i
nvest
igate
the
f
easi
bili
ty
of
th
e
SF
D
disc
reti
zat
ion
sc
hem
e
s
an
d
perform
ance
of
the
4 p
oin
t
-
S
E
G
it
erati
ve
m
eth
od
:
+
1
+
2
=
1
2
2
+
2
2
2
+
(
,
,
)
,
Ω
×
(
0
,
]
,
(1)
su
bject
t
o
the
foll
ow
i
ng con
diti
on
s
(
,
,
)
=
(
,
,
)
,
(
,
)
∈
Ω
,
∈
(
0
,
]
,
(
,
,
0
)
=
ℎ
(
,
)
,
(
,
)
∈
Ω
,
(2)
wh
e
re
1
and
2
are
co
ns
ta
nt
s
pe
eds
of
co
nve
ct
ion
in
the
directi
on
of
an
d
resp
ect
ively
;
1
>
0
and
2
>
0
are
the
c
oeffici
ents
of
di
ffu
sivit
y
in
t
he
−
a
nd
−
di
recti
ons,
res
pecti
vely
;
Ω
is
a
s
ubset
of
ℝ
2
;
(
0
,
]
is t
he
ti
m
e int
erv
al
;
(
,
,
)
an
d
ℎ
(
,
)
de
no
te
t
he
sm
oo
t
h functi
ons.
2.
SFD
AP
PR
O
X
I
M
ATIO
N E
QUA
TI
ON
Be
fore
we
sta
r
t
the
discr
et
iz
at
ion
process
of
pro
blem
(1
),
f
irstl
y,
we
tra
nsfo
rm
pr
oble
m
(1)
int
o
the
2D
el
li
ptic
PDE
us
i
ng
sim
i
lar
it
y
transfo
rm
a
ti
on
s
pecifica
ll
y
on
wa
ve
va
r
ia
bles
as
e
xpla
ined
i
n
t
he
pr
e
viou
s
sect
ion
.
To sta
r
t t
he
tra
ns
f
or
m
at
ion
process
, l
et
u
s c
onside
r
t
he
wav
e
v
a
riab
le
s as foll
ows
[
27
]
,
[
38
]
,
[
39
]
;
=
−
,
=
−
,
(3)
and w
e
use t
he
tran
s
f
or
m
at
ion
(
,
,
)
=
(
,
)
. Usin
g
(3),
in
(1) red
uce
s
t
o an ell
ipti
c P
DE
as
;
(
+
)
+
(
2
2
+
2
2
)
=
−
(
,
)
,
Ω
∈
[
,
]
×
[
,
]
,
(4)
wh
e
re
=
2
−
,
=
1
=
2
,
=
1
=
2
.
Let
us
buil
d
the
di
stribu
ti
on
of
un
i
form
l
y
node
points
as
sh
ow
n
in
Fi
gur
e 1
t
o help
us e
xp
l
or
e
the
der
i
vation o
f
t
he
si
m
il
arity approxim
a
ti
on
e
qu
at
ion
.
Fr
om
the
Fig
ure
1,
we
m
us
t
di
screti
ze
the
sol
ution
do
m
ai
n,
(
Ω
)
un
if
orm
l
y
in
bo
th
and
directi
ons
with
a
m
esh
siz
e,
ℎ
w
hich
is
def
i
ned
as
Δ
=
−
,
Δ
=
−
,
ℎ
=
Δ
=
Δ
an
d
=
+
1
.
Using
the
SFD schem
e and the
finite
gri
d
net
wor
k
in
F
ig
ure
1,
we dis
creti
ze the
2D
el
li
ptic PDEs (
4) as
fo
ll
ows:
|
=
+
1
,
−
−
1
,
2ℎ
2
2
|
=
+
1
,
−
2
,
+
−
1
,
ℎ
2
|
=
,
+
1
−
,
−
1
2ℎ
2
2
|
=
,
+
1
−
2
,
+
,
−
1
ℎ
2
(5)
The
n
s
ubsti
tute (
5)
i
nto
(
4)
,
we ha
ve
the
foll
ow
i
ng app
roxi
m
at
ion
equati
on
;
(
+
1
,
−
−
1
,
2ℎ
+
,
+
1
−
,
−
1
2ℎ
)
+
(
+
1
,
−
2
,
+
−
1
,
ℎ
2
+
,
+
1
−
2
,
+
,
−
1
ℎ
2
)
=
−
,
,
(6)
By
si
m
plifyi
ng
(6
),
we
get
;
−
1
,
+
+
1
,
+
,
−
1
+
,
+
1
−
4
,
=
,
(7)
wh
e
re
=
1
−
2
1
,
=
1
+
2
1
,
1
=
ℎ
2
,
2
=
2ℎ
and
,
=
−
,
1
.
Ba
sed
on
the
SFD
ap
pro
xi
m
at
ion
(
7),
a
li
near
syst
em
gen
e
rates
with
the
c
oe
ff
ic
ie
nt
m
at
rix
is
la
rg
e
-
scal
e an
d spa
rse
i
n
m
at
ri
x form
as (
8)
by
taking
=
1
,
2
,
3
,
…
,
=
(8)
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.
23
, N
o.
1
,
Ju
ly
2021
:
471
-
47
8
474
wh
e
re;
=
[
2
3
0
0
⋯
0
1
2
3
0
⋯
0
0
1
2
3
⋯
0
⋮
⋮
⋮
⋮
⋱
⋮
0
0
0
1
2
3
0
0
0
0
1
2
]
,
=
[
1
2
3
4
⋮
]
,
=
[
1
2
3
4
⋮
]
and
1
=
[
0
0
⋯
0
0
0
⋯
0
0
0
⋯
0
⋮
⋮
⋮
⋱
⋮
0
0
0
0
]
,
2
=
[
−
4
0
⋯
0
−
4
⋯
0
0
−
4
⋯
0
⋮
⋮
⋮
⋱
⋮
0
0
0
−
4
]
,
3
=
[
0
0
⋯
0
0
0
⋯
0
0
0
⋯
0
⋮
⋮
⋮
⋱
⋮
0
0
0
0
]
,
=
[
1
,
2
,
3
,
4
,
⋯
,
]
,
for
=
1
,
2
,
3
,
…
,
.
3.
DERIV
ATIO
N
O
F
4
-
P
OINT
SEG ITER
ATIO
N
Since
the
coe
f
fici
ent
m
at
rix
for
the
li
ne
ar
s
yst
e
m
(8
)
has
la
rg
e
-
scal
e
an
d
sp
a
re
cha
racte
risti
cs
,
this
stud
y
pro
pose
d
a
faster
nu
m
erical
so
l
ver
by
em
plo
yi
ng
a
4
-
point
SE
G
it
erati
on
.
N
ow,
in
this
sect
io
n,
t
he
form
ulati
on
of
4
po
i
nt
-
S
EG
at
tem
pts
to
be
est
ablished
.
To
obta
in
the
fo
rm
ulati
on
of
4
-
po
i
nt
SE
G,
we
consi
der
t
he
gr
id
netw
ork
in
Fig
ure
1
a
nd
a
group
of
blo
c
k
no
de
points
con
ce
pt
in
Fi
g
ure
2
The
finit
e
gr
id
netw
ork
of
the
SFD
a
ppr
ox
i
m
at
ion
equat
io
n
is
de
picte
d
i
n
Fig
ure
2
,
w
he
re
the
blo
c
k
it
erati
on
a
ppr
oa
ch
ha
s
been m
ade u
nti
l i
te
rati
on
c
onve
rg
e
nce is a
chi
eved
.
Figure
1. Finit
e grid
netw
ork
s
at
m
=8
Figure
2.
Im
ple
m
entat
ion
of
t
he
4
-
point S
EG
it
erati
on
at s
ol
ution d
om
ai
n
,
Ω
Appl
y
(7)
on
any
gro
up
of
four
node
points
in
the
s
ol
ution
dom
ai
n,
the
4
-
point
S
EG
it
erati
on
s
c
hem
e can
be
form
ulate
d
as
;
[
4
−
0
−
−
4
−
0
0
−
4
−
−
0
−
4
]
[
,
+
1
,
+
1
,
+
1
,
+
1
]
=
[
1
2
3
4
]
(9)
w
he
re
,
1
=
−
1
,
+
,
−
1
−
,
,
2
=
+
2
,
+
+
1
,
−
1
−
+
1
,
,
3
=
+
2
,
+
1
+
+
1
,
+
2
−
+
1
,
+
1
,
4
=
−
1
,
+
1
+
,
+
2
−
,
+
1
.
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
Perf
orma
nce
of
simil
ar
it
y exp
li
ci
t gro
up it
er
ation f
or
so
lv
in
g 2D
…
(
Nu
r Af
za
M
at Ali
)
475
By
est
ablishin
g
the
in
ve
rse
m
at
rix
of
th
e
coeffic
ie
nt
sys
tem
(9
)
an
d
m
anip
ulati
ng
it
,
the
ge
ner
al
schem
e o
f
the
4
-
point S
E
G
it
erati
on can
b
e
wr
it
te
n
as
;
[
,
+
1
,
+
1
,
+
1
,
+
1
]
(
+
1
)
=
1
4
[
1
1
+
3
2
+
6
3
+
3
4
2
1
+
1
2
+
3
3
+
4
4
5
1
+
2
2
+
1
3
+
2
4
2
1
+
4
2
+
3
3
+
1
4
]
(10)
wh
e
re
4
=
8
(
4
−
)
,
1
=
8
−
,
2
=
2
,
3
=
2
,
4
=
,
5
=
2
and
6
=
2
.
T
he
4
-
po
int
SEG
it
erati
on
wh
ic
h has
bee
n use
d
to
so
l
ve
the
pro
po
se
d p
roblem
(1)
,
is
s
umm
arised in
Algorithm
1
.
Algorithm
1
:
4
-
point
SEG
it
er
at
ion
i.
In
it
ia
li
ze
←
0
an
d
←
10
−
10
ii.
Ca
lc
ulate
an
d
iii.
iv.
Fo
r
=
1
,
2
,
3
,
…
,
,
cal
culat
e
the e
qu
at
io
n (
10)
.
Perfo
rm
the
conve
rg
e
nce
te
s
t,
|
(
+
1
)
−
(
)
|
<
=
10
−
10
.
If
ye
s,
go
to
ste
p
(
v).
Othe
rw
ise
,
r
e
pe
at
step (
ii
i).
v.
Disp
la
y t
he
nu
m
erical
o
utput
s.
4.
NUMER
IC
A
L E
X
PERI
M
ENT
T
hr
ee
sel
ect
ed
num
erical
experim
ents
wer
e
perform
ed
in
t
his
sect
io
n
t
o
i
ll
us
trat
e
the
fe
asi
bili
ty
of
4
-
point
SE
G
it
erati
on
i
n
s
olv
i
ng
t
he
2D
unste
ady CDE
(1) a
s co
m
par
ed w
it
h
the Jaco
bi a
nd
GS
it
erati
on
s.
F
or
the
sake
of
c
om
par
ison,
we
con
si
der
e
d
th
ree
crit
eria
in
cl
ud
e
it
erati
on
nu
m
ber
(
Iter.
),
exec
utio
n
ti
m
e
in
seco
nd
(
Time
)
and
m
axim
u
m
abso
lute
e
rror
(
Err
.
).
All
the
num
erical
exp
e
rim
ents
wer
e
r
un
with
dif
fer
e
nt
m
esh
siz
es (
m
)
su
c
h
as
64, 1
28, 2
56, 512,
and
1024.
Pr
oble
m
1
[40
]
Con
si
der
(
1)
in
the
unit
sq
ua
re
do
m
ai
n
[
0
,
1
]
×
[
0
,
1
]
wi
th
dif
fus
ion
c
oeffici
en
ts
=
0
.
05
,
convecti
on c
oe
ff
ic
ie
nts
=
0
.
8
and
an
a
naly
ti
c so
luti
on is
;
(
,
,
)
=
1
1
+
4
(
−
(
−
1
−
0
.
5
)
2
1
(
1
+
4
)
−
(
−
2
−
0
.
5
)
2
2
(
1
+
4
)
)
.
(11)
Pr
oble
m
2
[41
]
To
dem
on
strat
e
the
be
ne
fits
of
t
he
4
-
po
i
nt
SEG
it
erati
on
,
we
c
reated
a
new
te
st
pr
oblem
with
a
n
analy
ti
c so
luti
on as
foll
ow
;
(
,
,
)
=
(
−
2
)
cos
(
)
sin
(
2
)
.
(12)
and
the
source
te
rm
is
;
(
,
,
)
=
(
−
2
)
[
cos
(
+
2
)
+
cos
(
)
(
1
2
sin
(
2
)
+
cos
(
2
)
)
]
.
(13)
Con
si
der (
1) in
[
0
,
1
]
×
[
0
,
1
]
with
d
i
ffusio
n coe
ff
ic
ie
nts
=
0
.
5
, a
nd con
vecti
on
coeffic
ie
nts
=
1
.
0
.
Pr
oble
m
3
[42
]
In
t
his
pro
ble
m
,
we
c
on
sid
e
r
(
1)
i
n
the
unit
sq
ua
re
do
m
ain
[
0
,
1
]
×
[
0
,
1
]
wi
th
di
ffusi
on
c
oe
ff
ic
ie
nts
=
=
=
1
.
0
, con
vecti
on c
oe
ff
ic
ie
nts
=
64
.
0
an
d
the
analy
ti
c so
l
ution i
s
(
,
,
)
=
−
(
3
+
3
)
sin
(
+
)
(14)
and the
source
te
rm
is
;
(
,
,
)
=
(
+
)
−
(
3
+
3
)
cos
(
+
)
(15)
All
the
nu
m
erical
resu
lt
s
fo
r
4
-
po
i
nt
SEG
it
erat
ion
tog
et
her
with
Jaco
bi
and
GS
it
er
at
ion
s
in
so
l
vi
ng
the
above t
hr
ee
pr
ob
le
m
s w
ere c
ollec
te
d
a
nd tabu
la
te
d i
n T
abl
es 1,
2
a
nd 3, re
sp
ect
ively
.
Ba
sed
on
Tabl
es
1
t
o
3,
obvi
ou
sly
,
it
s
hows
that
our
pro
po
sed
it
erati
ve
m
et
hod,
nam
el
y
4
-
point
SE
G
it
erati
on
, gives
trem
end
ously
i
m
pr
ove in te
r
m
o
f
it
erati
on
nu
m
ber
a
nd exe
cution t
i
m
e, w
hic
h
sig
nifica
ntly
h
a
s
appr
ox
im
at
ely
red
uce
d
it
era
ti
on
num
ber
and
e
xecu
ti
on
tim
e
by
58
.
32
-
68.34%
an
d
50.
86
-
66.
89
%
fo
r
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.
23
, N
o.
1
,
Ju
ly
2021
:
471
-
47
8
476
Pr
oble
m
1,
69
.
83
-
72.17%
and
64.55
-
71.
89%
for
Probl
e
m
2
and
74.43
-
79.
33
%
a
nd
69.44
-
73.
53%
f
or
Pr
oble
m
3
re
sp
ect
ively
.
Al
so
,
for
the
m
axim
u
m
abso
lute
er
ror,
al
l
it
erati
ve
m
e
tho
ds
sho
w
e
xc
el
le
nt
agr
eem
ent and
cl
os
e
to
the e
xa
ct
so
luti
on
.
Table
1.
C
om
par
iso
n of
Iter.
,
Time
, a
nd
Er
r.
for
Jac
obi,
GS
and 4
-
po
i
nt SE
G
it
erati
ons
of
Pr
oble
m
1
Metho
d
Mesh
Size
6
4
x
6
4
1
2
8
x
1
2
8
2
5
6
x
2
5
6
5
1
2
x
5
1
2
1
0
2
4
x
1
0
2
4
Iter.
Jaco
b
i
2236
6732
1
9
0
1
7
4
9
8
6
8
1
1
9
7
2
9
GS
1268
3919
1
1
4
3
4
3
1
1
4
6
7
8
2
8
7
4
-
p
o
in
t SE
G
708
2242
6744
1
9
0
3
8
4
9
9
0
8
Time
Jaco
b
i
1
.48
4
.79
3
8
.20
3
9
1
.40
3
7
8
5
.6
4
GS
0
.82
2
.77
2
5
.30
2
6
1
.91
2
6
2
8
.6
7
4
-
p
o
in
t SE
G
0
.49
1
.76
1
6
.91
1
7
9
.43
1
8
6
0
.2
7
Err.
Jaco
b
i
1
.50
5
8
1
7
E
-
05
1
.51
5
9
1
9
E
-
05
1
.53
7
6
1
4
E
-
05
1
.62
0
9
3
4
E
-
05
1
.92
7
4
4
5
E
-
05
GS
1
.50
4
6
5
1
E
-
05
1
.51
1
8
9
7
E
-
05
1
.52
2
8
4
4
E
-
05
1
.56
5
9
5
8
E
-
05
1
.72
5
0
0
4
E
-
05
4
-
p
o
in
t SE
G
1
.50
4
1
1
2
E
-
05
1
.50
9
9
3
7
E
-
05
1
.51
5
3
3
0
E
-
05
1
.53
7
6
9
9
E
-
05
1
.62
0
5
4
3
E
-
05
Table
2
.
C
om
par
iso
n of
Iter.
,
Time
, a
nd
Er
r.
for
Jac
obi,
GS
and 4
-
po
i
nt SE
G
it
erati
ons
of
Pr
oble
m
2
Metho
d
Mesh
Size
6
4
x
6
4
1
2
8
x
1
2
8
2
5
6
x
2
5
6
5
1
2
x
5
1
2
1
0
2
4
x
1
0
2
4
Iter.
Jaco
b
i
9906
3
5
1
4
7
1
2
2
6
4
0
4
1
8
7
4
6
1
3
8
7
7
0
0
GS
5231
1
8
6
8
8
6
5
7
9
3
2
2
7
2
9
5
7
6
5
6
0
4
4
-
p
o
in
t SE
G
2757
9907
3
5
1
4
3
1
2
2
6
2
7
4
1
8
7
1
4
Time
Jaco
b
i
5
.94
2
2
.20
2
4
7
.69
3
5
2
2
.9
5
4
4
3
0
1
.54
GS
3
.14
1
2
.52
1
4
5
.05
1
9
2
3
.9
3
2
6
9
9
1
.47
4
-
p
o
in
t SE
G
1
.67
7
.45
8
6
.06
1
1
6
7
.6
3
1
5
7
0
3
.74
Err.
Jaco
b
i
3
.26
8
6
6
8
E
-
03
3
.26
9
3
2
6
E
-
03
3
.27
0
3
2
6
E
-
03
3
.27
3
9
1
9
E
-
03
3
.28
8
1
9
0
E
-
03
GS
3
.26
8
6
3
1
E
-
03
3
.26
9
1
7
6
E
-
03
3
.26
9
7
3
0
E
-
03
3
.27
1
5
4
9
E
-
03
3
.27
8
6
9
6
E
-
03
4
-
p
o
in
t SE
G
3
.26
8
6
1
2
E
-
03
3
.26
9
1
0
2
E
-
03
3
.26
9
4
3
4
E
-
03
3
.27
0
3
6
5
E
-
03
3
.27
3
9
5
3
E
-
03
Table
3
.
C
om
par
iso
n of
Iter.
,
Time
, a
nd
Er
r.
for
Jac
obi,
GS
and 4
-
po
i
nt SE
G
it
erati
ons
of
Pr
oble
m
3
Metho
d
Mesh
Size
6
4
x
6
4
1
2
8
x
1
2
8
2
5
6
x
2
5
6
5
1
2
x
5
1
2
1
0
2
4
x
1
0
2
4
Iter.
Jaco
b
i
421
1697
6611
2
5
5
1
9
9
8
0
7
3
GS
158
750
3138
1
2
5
5
2
4
9
1
5
2
4
-
p
o
in
t SE
G
87
387
1604
6403
2
5
0
8
2
Time
Jaco
b
i
0
.34
1
.23
1
4
.90
2
0
0
.16
3
0
8
2
.4
6
GS
0
.13
0
.62
7
.18
1
0
7
.98
1
6
5
7
.8
1
4
-
p
o
in
t SE
G
0
.09
0
.33
4
.01
6
1
.16
9
3
2
.81
Err.
Jaco
b
i
1
.14
7
1
3
2
E
-
03
1
.13
8
3
1
4
E
-
03
1
.13
6
0
6
7
E
-
03
1
.13
5
5
3
7
E
-
03
1
.13
5
5
5
3
E
-
03
GS
1
.14
7
1
3
2
E
-
03
1
.13
8
3
1
3
E
-
03
1
.13
6
0
6
1
E
-
03
1
.13
5
5
1
3
E
-
03
1
.13
5
4
4
9
E
-
03
4
-
p
o
in
t SE
G
1
.14
7
1
3
2
E
-
03
1
.13
8
3
1
2
E
-
03
1
.13
6
0
5
8
E
-
03
1
.13
5
5
0
1
E
-
03
1
.13
5
3
9
7
E
-
03
5.
CONCL
US
I
O
N
In
this
pa
pe
r,
we
hav
e
be
en
su
cces
sf
ully
red
uce
d
2D
parab
olic
PD
Es
,
par
ti
c
ul
arly
on
the
convecti
on
-
dif
fu
si
on
pro
blem
,
into
2D
el
li
ptic
PD
Es
usi
ng
the
sim
ilarity
so
luti
on
s
te
chn
iq
ue
via
wav
e
var
ia
bles
in
w
hich
we
m
anage
to
get
a
low
com
pu
ta
ti
on
al
com
plexity
as
desire
d
in
this
stud
y.
T
he
sim
il
arity
appr
ox
im
at
ion
eq
uation
ha
s
bee
n
discreti
zi
ng
b
y
us
in
g
the
S
FD
dis
creti
zat
ion
sc
hem
e
to
get
th
e
SF
D
appr
ox
im
at
ion
equ
at
io
n.
T
his
approxim
ation
eq
uation
ge
ne
rated
a
la
rg
e
-
scal
e
and
spa
r
se
li
near
syst
em
then
so
lve
d
usi
ng
4
-
point
SE
G,
G
S
an
d
Jaco
bi
it
erati
on
s.
T
he
4
-
point
SE
G
it
erati
on
ha
s
ac
hieve
d
the
hi
ghest
perform
ance
ba
sed
on
the
im
plem
entat
ion
of
these
th
ree
it
erati
on
s
since
the
it
erati
on
num
ber
an
d
e
xe
cution
tim
e
was
s
m
aller
than
Jaco
bi
and
GS
it
erati
on
s
.
Th
us
,
we
can
co
nclu
de
that
our
pro
pos
ed
m
e
tho
d
is
m
or
e
eff
ic
ie
nt
t
han
GS
a
nd
Jac
ob
i
it
erati
on
s.
T
his
re
searc
h
will
be
furthe
r
e
xp
a
nded
in
to
the
us
e
of
SF
D
discreti
zat
ion s
chem
e v
ia
h
al
f
-
swee
p
an
d quarte
r
-
swee
p
it
e
rati
on f
am
il
ie
s
for
s
olv
i
n
g t
he
2D
unste
a
dy CDE.
ACKN
OWLE
DGE
MENTS
Fo
r
the
c
om
pleti
on
of
t
his
pa
per,
the
a
utho
r
s
ap
pr
eci
at
e
th
e
fun
d
ob
ta
ine
d
f
r
om
Un
iver
sit
i
Ma
la
ys
ia
Saba
h
,
Ma
la
ys
ia
u
nde
r
the
res
earch
grant
sch
e
m
e (GUG0
491
-
1/20
20).
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
Perf
orma
nce
of
simil
ar
it
y exp
li
ci
t gro
up it
er
ation f
or
so
lv
in
g 2D
…
(
Nu
r Af
za
M
at Ali
)
477
REFERE
NCE
S
[1]
E.
J.
Veli
ng
,
“
Radi
a
l
Tra
nsport
in
a
Porous
Medium
with
Dirichle
t
,
Neum
ann
a
nd
Robin
-
T
y
pe
Inhom
ogene
ous
Boundar
y
Valu
es
a
nd
Gene
ral
Init
ia
l
Dat
a:
Anal
y
tical
Solu
ti
on
a
nd
Eva
lu
at
ion
,
”
Journal
of
Engi
nee
rin
g
Mathe
mati
cs
,
vo
l.
75
,
pp
.
173
–
18
9,
2012
,
doi
:
10
.
1007/s10665
-
011
-
9509
-
x
.
[2]
M.
Sheikhol
esl
a
m
i,
R.
E
ll
ah
i,
an
d
M.
Hass
an,
“
A
Stud
y
o
f
Nat
ura
l
Convection
He
at
Tr
ansfe
rs
in
a
Na
nofuid
Fll
ed
Enc
losure
w
it
h
El
li
p
ti
c
Inn
er
C
y
l
inde
r
,
”
Int
ernati
onal
Journal
of
Numerical
M
et
hods
for
Heat
and
Fl
uid
Fl
o
w
,
vol.
24
,
no
.
8
,
pp
.
1906
–
1927
,
20
14
,
doi
:
10
.
1108
/HFF
-
07
-
2013
-
0
225
.
[3]
J.
Isenbe
rg
and
C.
Gutfnge
r,
“
Hea
t
T
ran
sfer
to
a
Drain
ing
Film
,
”
Inte
rnational
Jo
urnal
of
Heat
and
Mass
T
rans
fe
r
,
vol.
16
,
no
.
2
,
pp
.
505
–
512
,
1973
,
doi:
10.
1016
/00
17
-
9310(73)90075
-
6
.
[4]
J.
Y.
P
arl
ange,
“
W
at
er
Tra
nsport
in
S
oil
s,”
A
nnual
Re
v
ie
w
o
f
Fl
uid
M
e
chan
ic
s
,
vol
.
12,
pp
.
77
–
102,
1980
,
doi:
10
.
1146/a
nn
ure
v.
fl
.
12.
0101
8
0.
000453
.
[5]
P.
A.
Garc
í
a
-
Sal
abe
rri
,
“
Modeli
n
g
Diffusion
a
nd
Convec
ti
on
i
n
T
hin
Porous
Tra
n
sport
Lay
ers
Us
i
ng
a
Com
posite
Conti
nuum
-
Network
Model:
Applicati
on
to
Gas
Diffusion
Lay
er
s
i
n
Pol
y
m
er
E
lectr
ol
y
t
e
Fuel
Cells,”
Int
ernati
on
a
l
Journal
of
Heat
and
Mass
Tr
ansf
er
,
vol
.
167
,
pp
.
120824
,
2021
, d
oi:
10
.
1016/j.ij
h
ea
tmass
tra
nsfer
.
2020.
120824
.
[6]
C.
Japhe
t
,
F.
Nata
f,
and
F.
Rogle
r
,
“
Th
e
O
pti
m
iz
ed
Ord
er
2
Method:
Ap
pli
c
at
ion
t
o
Co
nvec
t
ion
–
Diffusi
o
n
Problems
,
”
Futur
e
Gene
ration
Computer
Sys
te
ms
,
v
ol
.
18
,
no.
1,
pp.
17
-
30
,
2001
,
doi:
0.
1016/S016
7
-
739X(00)00072
-
8
.
[7]
A.
Baz
ile,
E
.
Hac
hem,
J.
C.
L
.
Huget
,
and
Y.
Mesri
,
“
Vari
at
i
onal
Multi
sca
le
Err
or
Esti
m
at
o
r
f
or
Anisotropic
Adapti
ve
Fluid
Mec
hanic
Sim
ula
ti
ons:
Appli
ca
t
ion
t
o
Convec
t
i
on
–
Diffusion
Problems
,
”
Computer
Me
thods
in
Appl
ie
d
Me
chan
ic
s and Engin
ee
r
ing
, v
ol
.
331,
pp.
94
-
115,
2018
,
d
oi:
10
.
1016/j.c
m
a.
2017.
11
.
019
.
[8]
D.
Shongs
heng
and
P.
Ji
ani
ng
,
“
Applic
at
ion
of
Convec
ti
on
-
Diff
usion
Equation
t
o
the
Anal
y
ses
o
f
Conta
m
inatio
n
Bet
wee
n
B
atche
s
i
n
Mu
lti
-
Prod
uct
s
Pipeline
Tr
ansport,
”
Applie
d
Mathe
mati
cs
and
Me
chanics
,
vol.
19
,
no.
8,
pp.
757
-
764
,
19
98.
[9]
A.
A.
Afif
y
,
“
Si
m
il
ari
t
y
So
lut
ion
i
n
MH
D
:
E
ffe
c
t
s
of
Therm
al
Dif
fusion
and
Dif
fu
sion
The
rm
o
on
Free
Conve
ct
iv
e
Hea
t
and
Mass
Tra
nsfer
ov
er
a
Stret
chi
ng
Surfa
ce
C
onsid
eri
ng
S
uct
ion
or
Inject
ion,
”
Comm
unications
in
Non
li
n
ear
Sci
en
ce and
Nu
meric
al
S
imulat
i
on
,
vol
.
14
,
no
.
5,
pp
.
2202
-
2214
,
2009
,
doi:
10.
10
16/j
.
cnsns.2008.07.001
.
[10]
M.
Siava
shi
,
H.
Rasam,
and
A
.
Iz
adi
,
“
Sim
il
a
rity
Solut
ion
of
Air
a
nd
Nanof
lui
d
Im
pinge
m
e
nt
Cooli
ng
o
f
a
C
y
l
indri
c
al Porous
Hea
t
Sink
,
”
J
ournal
of
Therm
al
Ana
ly
sis and
Calorimet
ry
,
vol
.
135
,
no
.
2
,
pp
.
1399
-
1415,
201
9.
[11]
Us
m
an,
P.
Li
n,
A.
Ghaffa
ri,
and
I.
Mus
ta
fa
,
“
Si
m
il
ari
t
y
Soluti
o
n
of
the
Parti
al
Diffe
ren
t
ia
l
Equ
at
ions
tha
t
Mod
e
l
W
at
er/
Magne
ti
t
e
Nanofl
uid
Flow
and
Hea
t
Tra
nsfer
on
a
Str
et
chable
Rota
t
ing
Disk
Subjec
t
to
Therm
al
Radi
at
io
n
a
nd
Lore
nt
z
Force
,”
Numerical
Me
thods
for
Parti
al
Diff
ere
nt
ial
Equat
ions
,
2
020
,
doi:
10
.
1002/nu
m
.
22677
.
[12]
O.
K.
Koriko
and
I.
L.
Anim
asa
un,
“
New
Sim
il
ari
t
y
Solu
ti
on
o
f
Micr
o
pola
r
Flui
d
Flow
Problem
Over
a
n
Uhs
pr
i
n
The
Presenc
e
of
Quart
ic
Kind
o
f
Autoca
t
aly
tic
C
hemica
l
R
ea
c
ti
o
n,
”
Frontiers
in
Heat
and
Mass
Tr
ansfer
,
vol.
8,
no.
26
,
pp
.
1
-
12
,
2017
,
doi:
10.
50
98/hmt.8.
26
.
[13]
M.
Hasanuz
za
m
an,
B.
Manda
l
,
a
nd
M.
M.
T.
Hossai
n
,
“
A
Study
o
f
Sim
il
ari
t
y
Solu
ti
on
of
Uns
te
ad
y
Com
bine
d
Free
a
nd
Force
Convec
t
ive
La
m
ina
r
Boundar
y
L
a
y
er
Flow
About
a
V
ert
i
ca
l
Porous
S
urfa
ce
with
Suct
ion
and
Blowing,
”
Annal
s of
Pure
a
nd
Applied Math
emati
cs
,
vol
.
6
,
no.
1
,
pp
.
85
-
97
,
2014.
[14]
Y.
M.
Y
atim
,
B.
R.
Du
ff
y
,
and
S.
K.
W
il
son
,
“
Tra
velling
-
W
ave
S
imila
rity
Solu
ti
o
ns
for
a
n
Uns
te
ad
y
Shear
-
Str
ess
-
Dri
ven
Dr
y
Pa
tch i
n
A
Flowing
Film
,
”
Sains
Ma
lay
sina
,
vol
.
42
,
no.
7
,
pp
.
975
-
9
80,
2013
.
[15]
S.
S.
Abas
and
Y.
M.
Yat
im,
“
Sim
il
ari
t
y
So
lut
io
n
for
an
Uns
te
ad
y
Dr
y
Pa
tc
h
i
n
A
Shear
-
Stress
-
Drive
n
Thi
n
Film
o
f
Non
-
Newtonia
n
Pow
er
-
La
w
Fluid,
”
In
AI
P
Confe
ren
ce
Proceedi
ngs
,
v
ol.
1605,
2014
,
pp.
361
-
366
,
doi:
10
.
1063/1
.
4
887616
[16]
M.
Ehrha
rd
t
and
R.
E
.
Mi
cke
ns,
“
A
Nons
ta
ndar
d
Finit
e
Diffe
r
e
nce
Sch
eme
f
or
Convec
ti
on
–
Di
ff
usion
Equa
t
io
ns
Having
Constan
t
Coeff
i
cients,
”
Appl
ie
d
Math
emati
cs
and
Comp
utat
ion
,
vol
.
21
9
,
no.
12,
pp
.
6
591
-
6604,
2013
,
doi:
10
.
1016/j.a
m
c.
2012.
12.
068
.
[17]
P.
M.
Jordan,
“
A
Nons
ta
ndar
d
Finit
e
Diff
ere
n
c
e
Scheme
f
or
Nonlinear
Hea
t
Tr
ansfe
r
in
A
Thi
n
Finit
e
Rod,
”
The
Journal
of
Diff
ere
n
ce
Eq
uati
ons
and
Appl
i
cat
ions
,
vol.
9
,
n
o.
11,
pp
.
1015
-
1021,
2003
,
doi:
10
.
1080/10
23619031000146922
.
[18]
L.
W
ang
and
L.
I
.
Roeg
e
r,
“
Nons
ta
ndar
d
Finit
e
Diff
e
ren
ce
Sch
emes
for
A
Cla
ss
o
f
Gene
ra
li
z
ed
Convec
ti
on
–
Di
ff
usion
–
Rea
c
ti
on
Equa
ti
ons
,”
Nu
meric
al
M
et
hods
for
Parti
a
l
Dif
f
ere
nti
al
Equa
ti
o
ns
,
vol.
31
,
no
.
4,
pp.
1288
-
1309
,
2015.
[19]
L.
Vulkov
and
A.
I.
Za
dorin
,
“
Two‐gri
d
Algorit
hm
s
for
The
Soluti
on
of
2D
Sem
il
ine
ar
Singula
rl
y
‐p
ert
urbe
d
Convec
ti
on
‐diff
usion
Equa
ti
ons
Us
ing
an
Expon
ent
i
al
Finit
e
Diffe
ren
c
e
Scheme,”
In
AIP
Confe
re
nce
Proceedi
ngs
,
vol.
1186
,
2009
,
pp.
371
-
379
,
doi
:
10.
1063
/1.
3265
351
.
[20]
A.
C.
Radh
akr
is
hna
Pill
ai
,
“
Fourth‐Order
Expon
e
nti
al
Finit
e
Diffe
re
nce
Methods
f
or
Boundar
y
Val
ue
Proble
m
s
of
Convec
ti
v
e
Diff
usion
T
y
p
e,”
Int
ernati
onal
Jour
nal
f
or
Numeric
al
Me
thods
i
n
F
lui
ds
,
vol
.
37
,
n
o.
1,
pp.
87
-
106
,
2001
,
doi
:
10
.
10
02/fl
d.
167
.
[21]
C.
W
ang,
“
A
New
W
ay
to
Gene
ra
te
a
n
Exp
one
ntial
Finit
e
Diffe
ren
c
e
Sche
m
e
f
or
2D
Con
vec
t
ion
-
Diffusio
n
Equa
ti
ons
,”
Jour
nal
o
f
Applied
M
athe
mati
cs
,
vol
.
2014,
no
.
3
,
pp
.
1
-
14
,
2014
,
doi
:
10.
1155/2014/
45
7938
.
[22]
G.
B.
L
i
,
L
.
M.
Jung
and
G.
Z
hi
,
“
The
Perturb
at
ion
al
Finit
e
D
iffe
ren
ce
(PF
D
)
Scheme
of
Var
ia
ti
on
al
Step
o
f
Convec
ti
on
-
Diff
usion E
qua
ti
on,
”
Journal
o
f
Hydr
odynamic
s
,
vol.
20
,
no
.
3
,
pp
.
29
3
-
299
,
2005
.
[23]
D.
M.
Young
,
“
Ite
ra
ti
ve
Soluti
on
of
L
arg
e
L
ine
ar
S
y
stems
,
”
Lond
on:
Ac
ad
emic P
r
ess
,
1971.
[24]
W
.
Hac
kbusch
,
“
Ite
rat
iv
e
Solu
tion of
L
a
rge
Spar
se
S
y
st
ems
of
E
quat
ions,
”
Sprin
ger
-
Ve
rlag
,
199
5.
[25]
Y.
Saad
,
“
Ite
r
at
i
ve
Methods
for S
par
se
Li
n
ea
r
Sy
stems
,
”
Int
ernati
onal
T
homas
P
ubli
shing
,
1996.
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.
23
, N
o.
1
,
Ju
ly
2021
:
471
-
47
8
478
[26]
R.
Rahman,
N.
A.
M.
Ali
,
J.
Su
la
iman,
and
F.
A.
Muhiddin
,
“
Caput
o’s
Finit
e
Diffe
ren
c
e
Solut
ion
o
f
Frac
ti
ona
l
Two
-
Point
Boundar
y
V
al
ue
Pr
oble
m
s
Us
ing
S
OR
Ite
ration,
”
I
n
AIP
Confe
ren
ce
Proc
ee
dings
,
vol.
2013
,
201
8
,
pp.
020034
,
doi
:
10.
1063/1.
50542
33
.
[27]
N.
A.
M.
Ali,
R.
Rahman,
and
J.
Sulai
m
an,
“
SO
R
Ite
rative
Metho
d
w
it
h
W
ave
Va
ria
bl
e
Tra
nsform
at
ion
f
or
Solving
Advec
ti
on
-
Diffu
sion E
qua
ti
ons,”
In
AI
P
Con
fe
re
nce
Proceedi
ngs
,
vol
.
2013
,
no
.
1
,
2018
,
doi
:
10
.
1
063/1.
5054235
.
[28]
A.
Sunarto,
J.
S
ula
iman,
and
A.
Saudi
,
“
Im
p
li
cit
Finit
e
Diff
e
ren
c
e
Soluti
on
f
or
Ti
m
e
-
Frac
ti
onal
Diffusion
Equa
ti
ons
Us
in
g
AO
R
Meth
od,
”
In
Journ
al
of
Phy
si
cs:
Confe
renc
e
Serie
s
,
vol
.
495
,
no.
1,
2014
,
doi:
10
.
1088/17
42
-
6596/495/
1/0
12032
[29]
A.
Sunarto,
P.
Agarwal
,
J.
Sulai
m
an
,
J.
V.
L.
Chew,
and
E.
Aruchuna
n
,
“
Ite
rat
ive
Me
t
hod
fo
r
Solving
One
-
Dim
ensiona
l
Frac
ti
ona
l
Ma
the
m
at
i
ca
l
Ph
y
si
cs
Model
Via
Q
uar
te
r
-
Sw
ee
p
a
n
d
PA
OR,”
Adv
a
nce
s
in
Dif
fe
ren
ce
Equat
ions
,
vol
.
2021,
pp
.
1
-
12
,
2021.
[30]
I.
K.
Yous
sef
and
A.
A.
Ta
ha
,
“
On
Modifie
d
S
uc
ce
ss
ive
Ove
rre
la
x
at
ion
Me
t
hod,
”
App
li
ed
Mathe
mati
cs
and
Computati
on
,
vo
l.
219
,
no
.
9
,
pp
.
4601
-
4613,
201
3
,
doi
:
10
.
1016/j.a
m
c.
2012
.
10.
07
1
.
[31]
F.
A.
Muhiddin,
J.
Sulai
m
an,
an
d
A.
Sunarto,
“
Num
eri
ca
l
Ev
aluati
on
of
Quart
e
r
-
Sw
ee
p
KS
OR
Method
to
Solve
Ti
m
e
-
Frac
t
io
nal
Para
bolic
Equ
at
i
ons,”
SSRG
In
ter
nati
onal
Journ
al
of
Engi
ne
erin
g
Tr
ends
and
Technol
ogy
,
vol
.
1
,
pp.
63
-
69
,
2020
,
doi:
10.
14445
/2
2315381/CATI2P210
.
[32]
D.
J.
Eva
ns,
“
Group
Expl
ic
i
t
Ite
ra
ti
ve
Me
tho
ds
for
Solving
La
rge
L
inear
Sy
stems
,
”
Int
ernati
onal
Journal
of
Computer
Mathem
ati
cs
,
vo
l. 17
,
pp.
81
-
108
,
198
5
,
doi
:
10
.
1080/
00207168508803452
.
[33]
A.
Saudi
and
J.
Sulai
m
an,
“
Path
Planni
ng
f
or
Mobile
Ro
bot
Us
ing
4EGSO
R
Via
Nine
-
Point
La
pl
ac
ian
(4EGSO
R9L)
Ite
rative
Method
,
”
Inte
rnational
Jo
urnal
o
f
Computer
Appl
ic
a
ti
on
s,
vol.
53
,
no
.
16
,
pp.
38
-
42
,
2012
,
doi:
10
.
5120/85
09
-
2568
.
[34]
M.
K.
M.
Akhir,
M.
Othm
an,
J.
Sulai
m
an,
an
d
Z.
A.
Maj
id
,
“
Four
Point
Expl
ic
i
t
Dec
oup
led
Group
Ite
rati
ve
Method
Applied
t
o
Two
-
Dim
e
nsional
Helmho
lt
z
Equa
t
ion
,”
I
nte
rnational
Jo
urnal
of
Math
e
matic
al
Anal
ysi
s
,
vol.
6
,
no
.
20
,
pp
.
963
-
974
,
2012
.
[35]
J.
Sulai
m
an,
M.
Othm
an,
and
M.
K.
Hasan,
“
Nine
Point
-
EDGS
OR
Ite
ra
ti
ve
Me
th
od
for
the
Fini
te
El
ement
Solut
io
n
of
2D
Pois
son
Equa
ti
ons,
”
In
Co
nfe
renc
e:
Proc
e
edi
ngs
of
th
e
In
t
ernati
onal
Con
f
ere
nce
on
Computati
ona
l
Scienc
e
and
Its
App
li
ca
tions
:
Part
I
,
200
9
,
doi
:
10
.
1007/
978
-
3
-
642
-
02454
-
2_59
[36]
N.
A.
M.
Ali,
R
.
Rahman,
J.
Sulai
m
an,
and
K.
Ghaz
ali
,
“
Num
eri
cal
Soluti
ons
of
Uns
te
ad
y
Adv
ec
t
ion
-
Diffusion
Equa
ti
ons
b
y
Us
ing
EG
Ite
r
ation
with
W
ave
Var
ia
bl
e
Tra
nsform
at
ion
,
”
In
Journal
of
Ph
ysic
s:
C
onfe
renc
e
Seri
es
,
vol.
1358
,
2019
,
doi:
10
.
1088/17
42
-
6596/1358/
1/
012049
.
[37]
T.
K.
Be
e,
N.
M
.
Ali,
and
L.
C
.
Hong
,
“
Expl
ic
i
t
Group
Methods
in
The
Solut
ion
o
f
the
2
-
D
Convec
t
ion
-
Diffusion
Equa
ti
ons
,
”
In
P
roce
edi
ngs o
f th
e
World
Congre
ss
on
Engi
ne
erin
g
, v
ol
.
3
,
2010
.
[38]
M.
Alqura
n,
“
S
oli
tons
and
Peri
odic
Soluti
ons
t
o
Nonline
ar
Parti
al
Diff
ere
nt
ia
l
Equa
ti
ons
b
y
T
he
Sine
-
Cosin
e
M
et
hod
,
”
App
li
e
d
Mathe
mat
ic
s a
nd
Information
S
ci
en
ce
s
,
vol. 6
,
n
o.
1
,
pp
.
85
-
88
,
2012.
[39]
A.
M.
W
az
waz
,
“
Tra
vel
l
ing
W
av
e
Soluti
ons
for
Com
bine
d
a
nd
Double
Com
bine
d
Sine
–
Cosine
-
Gordon
Equa
ti
o
ns
b
y
T
h
e
Vari
able
Separ
ated
OD
E
Method,
”
Appl
i
ed
Mathe
mat
ic
s
and
Computati
o
n
,
vol.
177
,
no.
2,
pp.
755
-
760
,
2006
,
doi
:
10
.
10
16/j
.
amc.
2005.
0
9.
104
.
[40]
R.
C.
Mitt
al
and
A.
Tri
pat
h
i,
“
Num
eri
ca
l
Soluti
on
s
o
f
Two
-
Dim
ensiona
l
Uns
te
ad
y
Convec
ti
on
–
Di
ff
usion
Problem
s
Us
ing
Modifie
d
Bi
-
Cubic
B
-
Spline
Fin
it
e
Elem
ent
s,”
Inte
rnat
io
nal
Journal
o
f
Computer
Math
emati
cs
,
vol.
94
,
no.
1
,
pp
.
1
-
21
,
2017
,
doi
:
10
.
10
80/00207160.
20
15.
1085976
.
[41]
R.
C.
Mit
ta
l
,
S.
Kum
ar,
and
R.
Jiww
ari
,
“
A
Cubic
B
-
S
pl
ine
Quasi
-
In
t
erp
olation
Met
hod
f
or
Solving
Two
-
Dim
ensiona
l
Uns
te
ad
y
Adv
ec
t
ion
Diffusion Equat
ions
,
”
In
ter
nati
onal
Journa
l
of
Numeri
cal
Me
thods f
or Hea
t
&
Fl
uid
Fl
ow
,
v
ol.
30
,
no
.
9
,
pp
.
4281
-
4306,
202
0
,
doi
:
10
.
1108/
HF
F
-
07
-
2019
-
0
597
.
[42]
H.
W
.
Sun
and
L.
Z
.
L
i,
“
A
CC
D
–
AD
I
Method
for
Uns
te
ad
y
Co
nvec
t
ion
–
Diffusi
on
Equa
t
ions,”
Computer
Phy
si
cs
Comm
unic
ati
ons
,
vol
.
185
,
no
.
3
,
pp.
790
-
797
,
20
14
,
doi
:
10
.
1016
/j
.
cpc.2013.
11
.
0
09
.
Evaluation Warning : The document was created with Spire.PDF for Python.