Indonesian J
ournal of Ele
c
trical Engin
eering and
Computer Sci
e
nce
Vol. 1, No. 3,
March 20
16, pp. 431 ~ 4
4
5
DOI: 10.115
9
1
/ijeecs.v1.i3.pp43
1-4
4
5
431
Re
cei
v
ed
No
vem
ber 2
3
, 2015; Re
vi
sed
Jan
uar
y 24, 2
016; Accepte
d
February 8,
2016
Optimal Power Flow using the Moth Flam Optimizer: A
Case Study of the Algerian Power System
Bachir
Ben
t
ouati*, Lakh
dar Ch
aib, Saliha Che
ttih
LACoSE
RE La
borator
y, Electr
ical En
gin
eeri
n
g Dep
a
rt
ment, Amar T
e
lidji U
n
iversit
y
of La
g
hou
at, Algeri
a
email : b.be
nto
uati@l
a
g
h
-un
i
v
.
dz
A
b
st
r
a
ct
In
this pa
per,
a n
e
w
tech
niq
ue
of o
p
ti
mi
z
a
t
i
on
know
n
as
Moth-F
la
m Opt
i
mi
z
e
r (MF
O)
has
be
e
n
proposed to s
o
lve the
prob
lem
of the Optim
a
l Power Flow (OP
F
) in the interconnected power syst
em
,
taking i
n
to ac
count the set of equa
lity an
d ine
qua
lit
y constrai
nts. T
he prop
osed a
l
gorith
m
h
a
s b
een
presented to t
he A
l
ger
ian power system
network for a
variety
of objectives. T
he obtained
res
u
lts ar
e
compar
ed w
i
th
rece
ntly p
u
b
lis
hed
al
gor
ith
m
s
such
as;
as
th
e Artifici
al B
e
e
Colo
ny (AB
C
),
and
oth
e
r
meta
-
heur
istics. Si
mulati
on r
e
sults
clearly
reve
al
the
effe
ctiven
es
s an
d the
rob
u
s
tness of th
e p
r
opos
ed
alg
o
rit
h
m
for solvin
g the OPF
proble
m
.
Ke
y
w
ords
: Moth flam
optim
i
z
e
r, Optim
a
l
power
flow, Power system
optim
iz
ation, Voltage
profile
1. Introduc
tion
Optimal
power flow
(OPF
) is a
well
studied
optimiza
t
ion p
r
oble
m
in po
we
r
syst
ems. In
1962, such an issue wa
s
first introd
uced by Carpe
n
tier [2]. The proble
m
of the OPF ca
n be
defined a
s
a
nonlin
ear p
r
o
g
rammi
ng p
r
oblem [1]. The main obje
c
t
i
ve of the OPF probl
em is
to
optimize ch
o
s
en obje
c
tive
function
s su
ch
a
s
pie
c
e
w
ise q
uad
ratic co
st functio
n
, fuel co
st
with
valve point e
ffects an
d voltage profile improv
em
ent, by optimal adju
s
ting the
powe
r
syste
m
control va
ria
b
les an
d
sati
sfying vari
ou
s
system
op
erating
such
as
po
wer flo
w
e
quatio
ns
and
inequ
ality con
s
traint
s, simul
t
aneou
sly [3–
6
].
To solve thi
s
pro
b
lem, the
re
sea
r
che
r
s
prop
osed
a n
u
mbe
r
of o
p
timization
alg
o
r
ithms
over the yea
r
s. The
r
efo
r
e
optimization
is kn
o
w
n a
s
one of the most cu
rrent
problem fa
ci
n
g
resea
r
ch, a good o
p
timization lead
s t
o
an optim
al solution fo
r
an efficient
system. The fi
rst
solutio
n
meth
od for the O
P
F proble
m
wa
s pro
p
o
s
e
d
by Domme
l and Tinney
[7] in 1968, and
sin
c
e the
n
n
u
merou
s
oth
e
r m
e
thod
s
have be
en
p
r
opo
se
d, so
me of them
are: Ant
Col
ony
Optimizatio
n
(ACO
) [8], G
enetic Alg
o
rit
h
m (GA
)
[9-1
0], enhan
ce
d
geneti
c
algo
rithm (EGA
) [11-
12], Hybrid
Geneti
c
Algo
rithm (HGA) [
13], artifi
cial neural netwo
rk (A
NN
) [14]
, Particle swarm
optimizatio
n
(PSO) [1
5], fuzzy ba
sed
hybrid
par
tic
l
e swarm optimiz
a
tion (fuzz
y
HPSO) [16]
,
Tabu Sea
r
ch
(TS) [17], Gravitational
Sear
ch Algorithm (GSA
) [18]. Biogeography ba
sed
optimizatio
n
algorith
m
(B
BO) [19], ha
rmony
sea
r
ch algo
rithm
(HS) [20],
kril
l herd alg
o
rit
h
m
(KHA) [21], Cuckoo Sea
r
ch (CS) [2
2], adaptive gr
o
up se
arch op
timization (A
GSO) [23], Black-
Hole
-Ba
s
ed
Optimizatio
n
(BHBO
)
[24]. The repo
rte
d
results we
re promi
s
ing
and en
cou
r
a
g
ing
new resea
r
ch
in this dire
ction.
A new
metho
d
kn
own a
s
t
he Moth
-Fla
me Optimi
zat
i
on MFO
alg
o
r
ithm this met
hod h
a
s
been
propo
sed by Seye
d
a
li Mirj
alili [2
5] in 2
015
is a n
a
ture
-in
s
pired
meth
od
from
navigat
ing
mech
ani
sm o
f
moths i
n
n
a
t
ure
calle
d transve
rse
o
r
ie
ntation, which ha
s n
o
t received yet mu
ch
attention in the power sy
stem.
He
nce, the first obje
c
t
i
ve of this paper is to ap
pl
y a new meth
od
that is the M
F
O in
order to solve the
OPF probl
em
. In what foll
ows, we
will
briefly give t
h
e
mathemati
c
al
mod
e
l o
n
th
e p
r
opo
se
d
a
l
gorithm
of
spiral
flying p
a
th of m
o
ths around
a
r
tificial
lights (flam
e
s) [25].
In this pape
r, an approa
ch
base
d
on M
F
O is propo
sed to solve th
e OPF probl
e
m
. This
probl
em ha
s
been formula
t
ed as a no
nl
inear o
p
timization pro
b
le
m with equali
t
y and inequ
ality
con
s
trai
nts. I
ndee
d, different obj
ective
s a
r
e
consi
d
e
r
ed i
n
thi
s
wo
rk to mini
mize the
co
st of
fuel,
emission,
an
d imp
r
ove th
e
voltage
profil
e. Moreover,
this met
hod i
s
simulate
d a
nd te
sted
on t
he
Algeria
n p
o
wer
system
n
e
twork. In
a
ddition,
the
result
s a
r
e
compa
r
ed
wit
h
othe
r m
e
th
ods
repo
rted in a
n
y relevant literatu
r
e de
alin
g with the su
bject.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
1, No. 3, March 20
16 : 431 – 445
432
The
org
ani
zation of
this pap
er is a
s
follo
ws: S
e
ction
2
discu
s
ses the
probl
em
formulatio
n of OPF while brief de
script
ion of
MFO, It is followed
by OPF implementatio
n
in
solving
OPF
probl
em in S
e
ction
3. Section 4 pr
esen
ts the sim
u
la
tion re
sults
a
nd di
scussio
n
.
Finally, Section 5 state
s
the con
c
lu
sio
n
of this pape
r.
2. The Formulation of O
P
F Problem
In general, the mathemati
c
al formulatio
n
of OPF probl
em can b
e
formulated a
s
a
n
optimizatio
n probl
em subj
ect to nonlin
e
a
r co
nst
r
aint
s:
,
(1)
,
0
(2)
,
0
(3)
Equation
s
(4
) and (5
) give respe
c
tively the vectors of
control varia
b
l
es
'u'
a
nd sta
t
e variable
s
'x'
of the proble
m
of OPF:
,
,
,
(4)
whe
r
e
:
active powe
r
Gen
e
rato
r output
at PV
buses except at the sla
ck bu
s.
:
voltages
Gene
ration b
u
s
:
T
r
a
nsf
o
rme
r
t
a
p
s
se
tt
i
n
gs
.
:
S
huntV
ARcompensation.
,
,
,
(5)
whe
r
e
:
voltage profile to lo
ad bu
se
s
:
Argument volta
ges of all the
buses, excep
t
the beam
node (sla
ck b
u
s)
:
Active power g
ene
rate
d to the balan
ce bu
s (sla
ck
bus).
:
rea
c
tive powers
gene
rated of
gene
rato
rs b
u
se
s.
2.1. Equalit
y
Cons
train
t
s
Ties
co
nst
r
ai
nts of th
e O
P
F refle
c
t th
e phy
sical
system of
ele
c
tri
c
al e
n
e
r
g
y
. They
rep
r
e
s
ent the
flow e
quatio
ns
of active
and
rea
c
tive power
i
n
an
electri
c
network, whi
c
h
are
rep
r
e
s
ente
d
resp
ectively b
y
equation
s
(6) and
(7):
0
(6)
0
(7)
,
Elements of the admittan
c
e matrix (con
duc
ta
nce and
susce
p
tan
c
e
resp
ectively).
2.2. Inequality
Constraints
g
i
g
i
g
i
P
P
P
max
,
min
,
i=
1…..
n
g
(8)
g
i
g
i
g
i
Q
Q
Q
max
,
min
,
i=
1…..
n
g
(9)
g
i
g
i
g
i
V
V
V
max
,
min
,
i=
1…..
n
g
(10
)
sh
i
sh
i
sh
i
Q
Q
Q
max
,
min
,
i=
1…..
n
sh
(11
)
g
i
g
i
g
i
T
T
T
max
,
min
,
i=
1…..
n
T
(12
)
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
Optimal Power Flow us
ing the Moth Flam Opti
miz
e
r: A c
a
s
e
s
t
udy
of the Algerian ... (Bac
hir B.)
433
whe
r
e
g
i
P
min
,
,
g
i
P
max
,
,
g
i
Q
min
,
, and
g
i
Q
max
,
are the ma
ximum active
powe
r
, mini
mum active p
o
we
r,
maximum rea
c
tive po
wer,
and mini
mum
rea
c
tive po
wer of the
ith
g
eneration u
n
it, respectively
.
In addition,
g
i
V
min
,
,
g
i
V
max
,
are the
maximum an
d minimum
limits of voltage am
plitud
e,
r
e
spec
tively.
sh
i
Q
min
,
stands fo
r lower
and
sh
i
Q
max
,
stand
s for u
pper limit
s of comp
ensa
t
or
cap
a
cit
o
r.
Fi
nally
,
g
i
T
min
,
and
g
i
T
max
,
prese
n
ts l
o
wer and
up
pe
r
boun
ds of ta
p chan
ger in
ith
trans
former.
Secu
rity con
s
traints: involve the con
s
trai
nt
s of voltage
s at load bu
ses an
d tran
smissi
on
line loadin
g
a
s
:
L
i
L
i
L
i
V
V
V
max
,
min
,
i=
1…..
n
L
(13
)
L
i
L
i
S
S
max
,
i=
1…..
nl
(14
)
whe
r
e
L
i
V
max
,
and
L
i
V
min
,
are the mini
mum and ma
ximum load voltage of
ith
unit,
L
i
S
define
s
appa
rent po
wer flo
w
of
ith
bran
ch.
L
i
S
max
,
defines maxim
u
m appa
rent
powe
r
flow limit of
ith
bran
ch.
A pen
alty function
[26] i
s
add
ed
to
the
obj
ective functio
n
, if th
e functional
operating
con
s
trai
nts vi
olate any of t
he limits. Th
e
initial va
lue
s
of the pen
alty weig
hts a
r
e
con
s
id
ere
d
a
s
in
[27].
3. Moth Flam Optimizer [1
9]
Moth Flam Optimize
r (M
FO) was first in
troduced
in [25]. The MFO ha
s proved its
comp
etitiveness with ma
ny other
opt
imization
alg
o
rithm, whi
c
h is in
spired
from physi
cal
phen
omen
a in nature. T
he main in
spiration of
t
he pro
p
o
s
ed
algorithm i
s
the navigating
mech
ani
sm
o
f
moths in n
a
ture th
at is cal
l
ed tr
a
n
sve
r
se ori
entation.
Figure 1
sho
w
s a
co
ncept
ual
model of tran
sverse o
r
ient
ation. Moths
are fan
c
y
insects that fligh
t
in the night usin
g moonli
ght,
that have a speci
a
l naviga
t
ion method in the ni
ght. Their movem
ent is done b
y
maintaining
a
fixed angle
with respe
c
t to the moon,
which
allows th
em to fly in a straig
ht light. This m
e
thod
is
calle
d tra
n
sv
erse o
r
ie
ntation. However,
due
to a
r
tifici
al light, the
m
o
ths
do
not
move
straig
ht but
spiral. For th
at, this light i
s
con
s
ide
r
ed
as a
ne
w goa
l for moth
s th
at have to be
conve
r
ge
d. The
optimizatio
n algorith
m
of this move is called Mo
th-Flame Optimization (MFO). In what follows
,
we will present the mathematical model of the pr
oposed al
gorithm
of spiral flying path of moths
arou
nd artificial lights (fla
mes). The a
pplicati
on include
s the finding of the optimal values of
control varia
b
l
es to minimi
ze the obje
c
tive function.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
1, No. 3, March 20
16 : 431 – 445
434
Figure 1. Tra
n
sverse o
r
ie
n
t
ation [25]
3.1. Mathem
atical Modell
ing
It is assum
e
d that the
candid
a
te sol
u
tions
are th
e moth
s, an
d the vari
abl
es of th
e
probl
em a
r
e
the p
o
sition
o
f
moths i
n
sp
ace. T
he
mot
h
s
ca
n fly hyper
dimen
s
io
nal
spa
c
e
wit
h
cha
ngin
g
thei
r po
sition ve
ctors. MFO a
l
gorithm
i
s
b
a
se
d on the
popul
ation. All the moths
are
repres
ented in a matrix as
follows
:
M
,
⋯
,
⋮⋱
⋮
,
⋯
,
(15
)
Whe
r
e
n
i
s
the num
ber
of moths an
d d
is the num
b
e
r of
v
a
ria
b
le
s (dim
en
sion
)
.
We as
su
me
a
table to store
the values of
all moths the
formatting a
s
follows:
OM
⋮
(16
)
Whe
r
e
n
i
s
t
he n
u
mbe
r
o
f
moths. An
o
t
her
key
co
mpone
nts i
n
the p
r
op
ose
d
alg
o
rithm
are
flames
. A matrix s
i
milar to t
he moth
matrix is
c
o
ns
idered as
follows
:
F
,
⋯
,
⋮⋱
⋮
,
⋯
,
(17
)
whe
r
e
n
is th
e numbe
r of moths, and d
is the numb
e
r of variables
(dimen
sion
).
It is also sup
posed that th
ere i
s
an a
r
ray for stori
n
g
the corre
s
po
nding
sha
p
in
g of values a
s
follows
:
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
Optimal Power Flow us
ing the Moth Flam Opti
miz
e
r: A c
a
s
e
s
t
udy
of the Algerian ... (Bac
hir B.)
435
OF
⋮
(18
)
whe
r
e
n
i
s
th
e numb
e
r of
moths. Th
e moths a
r
e a
c
tual
sea
r
ch a
g
ents that m
o
ve on the
search
spa
c
e,
while
the flames
are the be
st po
sition moth
s.
The po
sition
of each moth
is up
dated
with
respe
c
t to a flame usi
ng th
e followin
g
eq
uation:
)
,
(
j
i
i
F
M
S
M
(19
)
whe
r
e
M
i
indi
cate
s the
i
-th moth,
F
j
indicates the
j
-th fl
ame, and
S
is the s
p
iral func
tion.
Con
s
id
erin
g these point
s, a logarith
m
ic
spiral
is defin
ed for the MF
O algorith
m
a
s
follows:
,
.
.
cos
2
(20
)
whe
r
e
D
i
i
ndi
cate
s the
dist
ance of the
i
-t
h moth for the
j
-th flam
e,
b
is
a co
ns
ta
n
t
fo
r
de
fin
i
n
g
th
e
sha
pe of
the
loga
rithmi
c spiral,
an
d
t are a ran
d
o
m
num
b
er in
[-1, 1]. Th
e
logarith
m
ic spiral,
spa
c
e
aro
u
n
d
the flame,
and the
po
sition con
s
ide
r
i
n
g differe
nt
t
on the
curve
are illustrated in
Figure 2.
D
is calculated as follows:
|
(21
)
whe
r
e Di in
di
cate
s the dist
ance of the
i-th
moth for the
j-th
flame.
Numb
er
of flames is
red
u
c
ed
by incre
a
sin
g
the
n
u
m
ber
of iterat
ions. T
he foll
owin
g form
ul
a is
utilized in thi
s
subject:
∗
1
(22
)
whe
r
e
is the curre
n
t numb
e
r of iteration,
is the maximum numb
e
r of flames, an
d
denote
the maximum
numbe
r of iteration
s
.
Figure 2. Log
arithmi
c
spi
r
a
l
, space aro
u
nd a flam
e, and the po
sitio
n
with re
spe
c
t to t [25]
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
1, No. 3, March 20
16 : 431 – 445
436
3.2. Implementa
tion of th
e Proposed
MFO Algori
t
hm to the O
P
F Problem
The su
mma
ri
ze flowch
art of the propo
sed moth
-flam
e
algorith
m
is given in Figu
re 3.
Figure 3. Flow of prop
ose
d
MFO for sol
v
ing OPF
whe
r
e, the bo
unda
ry limits and vari
a
b
le
of control are
given by;
lb=[lb
1
,lb
2
,...,l
b
n
] is the lower bou
nd of variabl
e n;
ub=[u
b
1
,ub
2
,...,ub
n
] the upper bou
nd of variabl
e n.
Our o
b
je
ctive is to solve th
e OPF proble
m
. Hen
c
e, we will apply th
e moth-flam
e
method for t
h
is
purpose as
follows
;
The co
ntrol v
a
riabl
es a
r
e:
P, V, T and Q
c
W
h
er
e
Lb
=[
P
mi
n
,
V
min
,T
mi
n
,Q
c m
i
n
] (23
)
Ub
=[
P
ma
x
,
V
ma
x
,T
ma
x
,Q
c ma
x
]
(24
)
4. Applicatio
n and Re
sults
In orde
r to show the rob
u
stne
ss and
e
ffectiveness of prop
os
e
d
MFO app
roach for
solving
OPF
probl
em in
large
r
po
we
r system
s,
it has
been te
sted o
n
Alge
rian 5
9
-b
us t
e
st
system
sho
w
n in Figu
re 4
.
Which ha
s
a 20
control
variable
s
. Thi
s
net
work i
s
comp
osed 1
0
gene
rato
r, 36
load
s of 684
.10MW a
nd 8
3
bra
n
che
s
, kno
w
in
g that
the gen
erato
r
of the bu
s No.
13 is not in service. Th
e value
s
of coefficient
s fuel costs a
nd emi
ssi
on
s of the ten generato
r
s,
the vario
u
s n
e
twork
co
ntrol varia
b
le
s
and th
ei
r
ra
n
ges con
s
ide
r
ed throug
hou
t this
study
and
other
param
eters a
r
e giv
en in [2
8]. T
he pa
ram
e
te
r settings to
execute
MF
O is:
Num
b
e
r
of
popul
ation =
40, maximum
of iteration =
150.
St
ar
t
Get
the
f
u
n
c
ti
on
de
ta
i
l
s
(
low
e
r
bo
und
,
up
per
bo
und,
v
a
r
i
able
s
dim
ens
ion)
an
d f
u
nc
t
i
o
n
e
v
alua
t
i
o
n
M
ap c
o
n
t
rol v
a
riab
les
f
r
o
m
ea
ch
g
r
ou
p o
f
mo
th i
n
t
o
p
o
w
e
rf
l
o
w
da
t
a
Ev
alua
t
i
o
n
(
O
b
t
ain
O
b
j
e
ct
iv
e f
u
n
c
t
i
on
fr
o
m
P
F
)
St
ore th
e
be
st resu
l
t
int
o
F s
t
or
e
and
po
sit
i
on
Up
d
a
ta
po
si
ti
o
n
s
us
i
n
g
Eq
.(21
)
f
o
r
eac
h
g
r
ou
p
o
f
m
o
th
V
a
riable
s
ou
t
of
b
oun
d?
Iter
a
ti
on
max
?
T
a
g
g
i
ng
at
t
h
e
bo
und
ar
i
e
s
Ch
oo
se
th
e
b
e
st
re
su
l
t
Se
t t
h
e nu
m
b
er
o
f
M
o
t
h
an
d Fl
am u
s
i
n
g
E
q
.(2
2
)
an
d se
t
t
he M
a
x
it
er
at
ion
No
Y
es
No
Ye
s
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
Optimal Power Flow us
ing the Moth Flam Opti
miz
e
r: A c
a
s
e
s
t
udy
of the Algerian ... (Bac
hir B.)
437
Figure 4. Single line diag
ram of the Algerian p
r
o
d
u
c
tion and tra
n
smissi
on net
work 5
9
-bu
s
system [24]
The MFO me
thod ha
s bee
n applie
d to solve the OPF probl
em for the followi
ng case
s:
Ca
se 0: The
basi
c
case
Ca
se 1: Mini
mization of g
eneration fuel
cost.
Ca
se 2: Mini
mization of to
tal emissi
on.
Ca
se 3: Voltage profile im
p
r
oveme
n
t.
Ca
se 4: Voltage profile im
p
r
oveme
n
t with fuel co
st minimizatio
n
.
Ca
se 5: Mini
mization of g
enerat
ion fuel
cost an
d emi
ssi
on.
Ca
se 6: Mini
mization of g
eneration fuel
co
st with
con
s
ide
r
ing valve
point effect.
The valu
e
of
the voltage
p
r
ofile i
s
co
nst
antly maintai
ned
within
th
e allo
wa
ble
o
peratin
g
limits by ad
di
ng the
penalt
y
factor. Th
e
prop
osed wo
rk
was
imple
m
ented and comp
uted un
der
Intel(R) Core(TM), 2.40 GHz
co
mpute
r
with 8 GB RAM.
4.1. Case 1
:
Minimization
of Gener
a
ti
on Fuel Cos
t
In this
c
a
s
e
,
we are interes
t
ed in solv
ing the pro
b
lem of OPF
while minim
i
zing the
corre
s
p
ondin
g
fuel co
st produ
ction. The
nature an
d fo
rm of the obj
ective functio
n
in this ca
se
is:
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
1, No. 3, March 20
16 : 431 – 445
438
1
(23
)
whe
r
e
: Total numbe
r of g
enerators.
: active po
wer
g
enerated by t
h
e unit i.
,
,
are the
fuel co
st coef
ficients of the
ith
generato
r
The redu
cti
on of the obj
ective functio
n
can b
e
ach
i
eved by finding the optim
al set of
control p
a
ra
meters
whi
c
h
is
a mi
nimu
m produ
ction
co
st. Th
e
result
s of th
e
optimal
cont
rol
variable
s
o
b
tained in thi
s
ca
se a
r
e
sho
w
n in T
able 1
.
These val
u
e
s
give u
s
the
best
solution
in
prod
uctio
n
co
st (minimum
co
st of
fuel).
Additionally, it can be s
e
e
n
that the optimal power flow
probl
em le
d
eco
nomi
c
di
spatch
to
cont
rol the
a
c
tive po
wer while
con
s
id
erin
g fl
exible fun
c
tio
n
al
con
s
trai
nts for influen
cin
g
in the optimization p
r
o
c
ed
ure, it ca
n be noted that all the contro
l
variable
s
rem
a
ined i
n
their
permi
ssible li
mits. The va
l
ues
of fuel co
st for the Alg
e
rian
59
-bu
s
t
e
st
system
are 1
693,61
93
($/
h
r), th
e MF
O
is
co
nsid
ere
d
a
s
12.8%
l
e
ss tha
n
the
base
ca
se. T
h
e
voltage di
agram sho
w
n i
n
Figure 5
illust
rates that
M
F
O violated
th
e up
per bo
un
darie
s
i
n
a
fe
w
buses.
Figure 6
sho
w
s th
e vari
ation of the fu
el co
st
ba
sed
on the n
u
mb
er of iteration
s
for th
e
prop
osed alg
o
rithm
s
MFO.
As observed
,
from and on
wards 6
0
iterations the
r
e i
s
no chan
ge
in
the fuel
co
st functio
n
value
.
That si
gnifie
s
t
he
optimal
solutio
n
for th
e problem
ca
n be
obtain
e
d
within
60 ite
r
ations.
Thi
s
can b
e
ju
stifie
d by p
e
rfo
r
m
ance of th
e p
r
opo
se
d m
e
thod to
explo
r
e the
sea
r
ch sp
ace and minim
i
ze the rat
e
of c
onvergen
ce. The MF
O manag
e
s
to find a goo
d
prod
uctio
n
co
st with greate
r
conve
r
g
e
n
c
e rate
whi
c
h i
s
mea
s
u
r
ed b
y
the number
of generation
s
.
Figure 5. Voltage dia
g
ra
m for ca
se 1
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
Optimal Power Flow us
ing the Moth Flam Opti
miz
e
r: A c
a
s
e
s
t
udy
of the Algerian ... (Bac
hir B.)
439
-
2
0
0
20
40
60
80
1
0
0
1
2
0
1
4
0
160
16
00
18
00
20
00
22
00
24
00
26
00
28
00
30
00
32
00
34
00
C
o
s
t
(
$
/h
)
I
t
erat
i
o
ns
Cost
Figure 6. Con
v
ergen
ce fo
r ca
se 1
Table 1.
Opti
mal setting
s of control vari
able
s
for different ca
se
s
Control variable
Case 0
Case 1
Case 2
Case 3
Case 4
Case 5
Case 6
V
G1
1,06
1,0999
1,1
1,099
1,1
1,1
0,96727
V
G2
1,04
1,0874
1,0975
1,0066
1,003
1,1
1,1
V
G3
1,05
1,098
1,098
0,9406
0,94
1,099
1,0872
V
G4
1,0283
1,0914
1,0896
1,0616
1,0392
1,0177
0,9976
V
G5
1
1,0994
1,0879
0,98048
1,0005
1,1
1,0264
V
G6
1,0266
1,0907
1,0895
1,04244
1,0393
1,018
1,0199
V
G7
1,0273
1,1
1,1
1,0308
1,03
1,1
1,093
V
G8
1,0966
1,09996
1,097051
1,0133
1,0207
1,097
1,0409
V
G9
1,034
1,1
1,097
1,0593
1,0684
1,0999
1,1
V
G1
0
1
1,09972
1,093
1,1
1,1
1,0986
1,06342
P
G1
8,0436
56,599
9,065
71,9711
72
28,38
60,8856
P
G2
70
23,5344
69,994
67,2696
24,164
64,2715
51,6744
P
G3
70
104,349
90,7919
31,0485
120,895
101,730
149,479
P
G4
115
114,893
86,4097
121,649
114,357
111,062
76,8308
P
G5
0
0
0
0 0
0 0
P
G6
40
10
82,402
64,062
33,4847
10
99,2313
P
G7
30
51,471
58,383
40,7262
49,676
58,8279
10
P
G8
110
98,5932
72,129
118,764
79,055
85,7488
140
P
G9
70
145,321
90,9795
174,595
146,424
103,18
42,4206
P
G1
0
200
105,779
87,646
188,862
113,792
103,649
50,0607
Fuel cost ($/h)
1943,4
1693,61
1811,93
2165,57
1732,852
1739,18
1773,04
Emission
(ton/h)
0,5834
0,5786
0,3844
1,8907
0,5922
0,4333
0.6488
Ploss
(MW)
28.944
29.621
23.078
40.573
26.391
30.298
30.144
QPloss(MVar)
97.83
112.37
76.85
150.01
111.82
108.28
108.01
VD
1,48
2,77
2,79
1,335
1,435
2,508
2.0101
4.2. Case 2
:
Minimization
of Total Emission
The central
thermal p
o
wer gen
eratio
ns a
r
e a m
a
jor
sou
r
ce of gree
nhou
se ga
se
s
emission: nitrogen
oxide
s
(NO
X
), sulfu
r
dioxide (SO
2
) and
ca
rb
on
dioxide
(CO2
). The
fun
c
tio
n
of
emission
s i
n
clud
es two
term
s, a
poly
nomia
l te
rm
and
an exp
o
nential te
rm.
The
analyti
c
al
expres
s
i
on for this
func
tion is
as
follows
:
2
(24
)
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
1, No. 3, March 20
16 : 431 – 445
440
with
,
,
,
are the
emissi
on fact
ors fo
r unit
i
.
T
h
e
o
p
t
ima
l
va
lu
e
s
o
f
th
e c
o
n
t
r
o
l va
r
i
ab
le
s o
b
t
a
i
n
e
d
b
y
min
i
miz
i
ng
e
m
iss
i
o
n
s
th
r
o
ugh
the algorith
m
are given in
Table 1. Fro
m
this re
sult, it is clear th
at emission
s
are redu
ce
d to
0,3844 to
ns/
hour,
whi
c
h
redu
ce
s emi
s
sion
s ove
r
3
4
.5% com
p
a
r
ed to the
ba
se ca
se. Fi
gu
re 7
sho
w
s the variation of th
e emission d
epen
ding
on
the numbe
r of iterations f
o
r the pro
p
o
s
ed
method. The
same
rema
rks ca
n be de
d
u
cted a
s
befo
r
e.
-
2
0
0
20
40
60
8
0
100
1
2
0
140
160
0,
0
0,
5
1,
0
1,
5
2,
0
2,
5
3,
0
Emissi
on (t
on
/
h
)
I
t
er
at
io
n
s
Em
issi
on
Figure 7. Con
v
ergen
ce fo
r ca
se 2
4.3. Case 3
:
Voltage Profile Impro
v
e
m
ent
For imp
r
ovin
g the voltage
profile, a ta
rget re
p
r
e
s
e
n
ting the redu
ction in the ga
p voltage
load bu
se
s compa
r
ed to the unit (1 pu
) is inclu
ded in
the OPF, this can be
writte
n as follo
ws:
3
|
1
|
(25
)
whe
r
e
voltage profil to loa
d
buses (pu
)
,
Total load bu
se
s.
We mu
st also minimize the deviation of
the
voltage profile of all buses in the
netwo
rk.
The optim
um
values
of the 20
cont
rol
variable
s
o
b
tained i
n
this
ca
se a
r
e
sho
w
n in
Table
1.
Noting that, the voltage p
r
ofile of a few buses
i
s
forced to 1
pu
Figure8 de
pi
cts the volta
ge
diagram of the Algeria
n 59
-bu
s
test sy
stem.
Evaluation Warning : The document was created with Spire.PDF for Python.