TELKOM
NIKA
, Vol.9, No.1, April 2011,
pp. 1~8
ISSN: 1693-6
930
accredited by D
G
HE (DIKTI
), Decree No: 51/Dikti/Kep/2010
¢
1
Re
cei
v
ed Jan
18
th
, 2011; Revi
sed Ma
r 9
th
, 2011; Acce
pted Apr 1
st
, 2011
Generator Contribution Based Congestion Management
using Multiobjective Genetic Algorithm
Sa
w
a
n Sen*
1
, Pri
y
anka Ro
y
2
, Abhijit
Chakrabarti
3
, Samarjit Sengupta
4
1,2
EE Department,
T
e
chno In
dia, Kolk
ata, India
3
EE Departme
n
t, Bengal E
ngi
neer
ing a
nd Sc
ienc
e Univ
ersit
y
, Shi
bpur, Ind
i
a
4
Departme
n
t of Appli
ed Ph
ysi
cs,
Kolkata, W
e
st Benga
l, Ind
i
a
e-mail: se
n_sa
w
a
n
@r
ed
iffmai
l.com*
1
, roy
_
priy
an@r
ediffmail.com
2
,
a_chakr
a
b
o
rti55@
ya
ho
o.co
m
3
, samarsgp@rediffmai
l.co
m
4
Abs
t
rak
Manaje
m
e
n
kon
ges
ti
adal
ah
sa
la
h
satu
fungs
i
utama
op
era
t
or
sistem
p
a
da
in
dustri
d
a
ya
terstrukturisasi
sela
ma ko
nti
gens
i tak terd
uga. Maka
la
h ini
me
ngus
ulk
an seb
u
a
h
metoda
ma
na
je
me
n
kong
esti men
g
gun
aka
n
alg
o
ri
thma
gen
etik
muti-
obj
ek
ber
basis ko
ntribus
i pe
mb
angk
it. Pada a
l
g
o
ritma
ini,
rugi-ru
gi r
e
a
l
dan
reaktif
di
opti
m
as
i
me
n
ggu
nak
an
mo
del
al
iran
day
a o
p
ti
mal
da
n ko
ntribus
i d
a
r
i
pe
mb
angk
it ya
ng ru
gi-ru
gi
ny
a di
opti
m
alka
n
ada
la
h d
i
kalk
ulasi. P
a
d
a
l
e
v
e
l ke
du
a, jal
u
r
yang
men
g
a
l
ami
kong
esti d
iid
en
tifikasi d
e
n
g
a
n
ind
e
ks b
e
b
an
berl
ebi
h ya
ng
dius
ulka
n se
la
ma
konti
g
e
n
si,
da
n j
a
lur
terse
but
dilo
ng
gark
an d
eng
an kontri
b
u
si bar
u dari
pe
mb
angk
it, yang
mer
upak
a
n
kelu
aran d
a
r
i algor
itma ya
ng
dike
mban
gka
n
. Metoda yan
g
dire
ncan
aka
n
me
ng
ga
mb
ar
k
an infor
m
asi te
rkait ma
na
je
men kon
gesti u
n
tuk
me
mini
malka
n
biay
a inv
e
stasi
,
tanpa i
n
stala
s
i pira
nti ekste
r
nal d
an
untuk
me
maksi
ma
lk
an kes
e
ja
htera
a
n
konsu
m
en d
e
n
gan
meng
hi
nd
ari p
e
mbatas
a
n
be
ban t
anp
a men
gaki
batk
an pr
ofil teg
a
nga
n dar
i sist
e
m
sede
mikia
n
h
i
n
gga r
ugi
siste
m
total tero
pti
m
a
s
i.
Siste
m
bus
IEEE 30
dig
u
n
a
k
an u
n
tuk
me
n
d
e
m
o
n
strasika
n
keefektifan
dari
meto
de yan
g
dius
ulka
n.
Ka
ta k
unc
i
: al
gorit
ma
ge
neti
k
, ind
e
ks
beb
a
n
b
e
rle
b
i
h
, jar
i
nga
n
daya
ter
egu
lasi, k
onti
g
ensi,
mana
je
men
kong
esti
A
b
st
r
a
ct
Cong
estion
ma
nag
e
m
e
n
t
is
one
of
t
he key functio
n
s of system operator
in
the
r
e
structured
p
o
w
er
ind
u
stry duri
n
g
unex
pected c
ontin
ge
n
cy. T
h
is pap
er pro
p
o
s
es a metho
d
for gen
erator c
ontrib
u
tion
bas
e
d
cong
estio
n
ma
nag
e
m
ent us
in
g multio
bj
ectiv
e
ge
netic
a
l
g
o
r
ithm. In the
al
gorith
m
, b
o
th r
eal
and r
eactiv
e
losses
h
a
ve b
een opti
m
is
ed usin
g
opti
m
al pow
er
flow
mo
del a
nd th
e co
ntributi
ons of t
he g
ener
ators
w
i
th
those
opti
m
is
e
d
l
o
sses
are
ca
lculat
ed. On
se
cond
lev
e
l, th
e
cong
ested
li
ne
s are
id
entifi
e
d
by th
e
prop
os
e
d
overl
oad
in
g in
dex (OI) durin
g cont
in
ge
ncy
and thos
e li
n
e
s are rel
i
ev
e
d
w
i
th the ne
w
contributio
n
of
gen
erators, w
h
ich is
the
o
u
tco
m
e
of th
e
dev
e
l
op
ed
al
gorith
m
. T
h
e p
l
a
nne
d
meth
od
de
pi
cts the i
n
for
m
a
t
io
n
relate
d to cong
estion
ma
nag
e
m
e
n
t to min
i
mi
z
e
the i
n
ve
st
ment cost, w
i
thout installi
ng a
n
y external d
e
vi
ces
and to
maxi
mi
se the cons
u
m
er w
e
lfar
e b
y
avoid
i
ng
a
n
y
load curta
i
l
m
ent w
i
t
hout affecting the v
o
lt
age
profile of the system
as
well as the optim
is
ed total sy
stem
loss. IEEE 30 bus system
is used to
de
mo
nstrate th
e effectiven
ess
of the meth
od.
Ke
y
w
ords
: cong
estio
n
ma
nag
e
m
ent, co
nting
ency, der
egu
late
d pow
er netw
o
rk, genetic a
l
g
o
rith
m,
overl
oad
in
g ind
e
x
1. Introduc
tion
The b
a
si
c
req
u
irem
ent of p
o
we
r n
e
two
r
k is
to m
eet th
e dem
and
even in
co
nting
ent stat
e
of the
system
. But the t
r
an
smissio
n
lo
ss incurs
rou
ghl
y 3% to 5%
o
f
the total p
o
w
er ge
ne
ratio
n
,
whi
c
h m
a
y b
e
co
nsi
d
e
r
ed
as
one
of th
e majo
r fa
cto
r
s i
n
de
re
gul
ated po
we
r
system, i.e. lo
ss
allocation m
a
y co
nsi
dera
b
ly affect th
e compet
itive po
sition
of
the GE
NCOs i
n
the
p
o
we
r
netwo
rk.
Nev
e
rthele
s
s, it
seem
s th
at
most of th
e
electri
c
al
ma
rket
s h
a
rdly ever
refle
c
t the
transmissio
n loss in their spot pri
c
in
g due to
the complicated a
s
pe
cts of loss allocation [1].
Earlier researche
r
s have
d
i
scusse
d diff
erent
mod
e
ls
for l
o
ss
allo
cation in
cl
assical
method
s.
It
can
be m
o
d
e
led by in
cre
m
ental tra
n
smissi
on lo
ss (ITL)
co
efficient [2], physical flo
w
-
ba
sed
approa
ch [3]. Bus impe
da
nce m
a
trix [4] or ba
si
c
ci
rcuit theorie
s [5
] have been
use
d
to tra
c
e
the
power flo
w
o
f
the network durin
g catte
ring the
dem
a
nd. Again efficient comp
utation algo
rith
m
[6], load flow method [7], on line optimal
powe
r
flow (OPF) tools [8
] had been al
so intro
d
u
c
ed
for
Evaluation Warning : The document was created with Spire.PDF for Python.
¢
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 9, No. 1, April 2011 : 1 – 8
2
loss allocatio
n
. In other literat
u
r
e [9] genetic alg
o
rit
h
m (GA) te
chniqu
e is also used fo
r loss
tracin
g. Although lo
ss al
locatio
n
ca
n
be en
sured
but still no
bodie
s
can
guarantee t
hat
unexpe
cted
situations
su
ch as g
ene
rat
o
r fault, li
ne
fault or trippi
ng wo
uld n
o
t happe
n. In this
unprecede
nte
d
state, th
e
loss allo
catio
n
may
cha
n
ge a
nd
som
e
ne
w te
chni
que
has to
be
implemented
to maintain sys
tem s
e
c
u
rity.
For the devel
opment of the prop
osed tech
niqu
e,
co
ntingen
cy like
tripping of line has to
be con
s
id
ere
d
. In case of
any kind of line faul
t of the power net
work, the po
wer, which had
flowed th
rou
gh the tripp
e
d
line, sho
u
l
d
flow th
ro
ug
h other exi
s
ting line
s
to meet co
nsu
m
er’
s
expectatio
n
[
10]. Thi
s
cau
s
e
s
the
po
ssibility of othe
r lin
e b
e
com
e
s
overl
oade
d o
r
con
g
e
s
ted.
Thus tran
smi
ssi
on lo
ss du
ring po
we
r transmi
ssi
on a
nd line co
ng
estion a
r
e th
e most impo
rtant
issue
s
for d
e
reg
u
lated
p
o
we
r environ
ment, whe
r
e
spot p
r
ice is the mai
n
con
s
id
eratio
n
for
con
s
um
er
we
lfare. He
nce, a tran
spa
r
e
n
t met
hod fo
r co
nge
stion
relief du
ring
contin
gen
cy an
d
allocating tra
n
smi
ssi
on lo
ss bet
wee
n
all
of the in
terested partie
s
in
an eq
uitable
and fair man
ner
is
requi
re
d. T
here
a
r
e
sev
e
ral
re
se
arch
es i
n
expl
oitin
g
the
rel
e
a
s
e
e
of lin
e ove
r
f
l
ow.
Con
g
e
s
tion
of line ca
n b
e
mana
ged
by different
way e.g.
loa
d
cu
rtailment
, econ
omic l
oad ma
nag
e
m
ent,
VAR s
u
pport
[11].
At this step, an esse
ntial and challen
g
i
ng ta
sk is to develop a
sof
t
computing
method,
whi
c
h optimi
z
e the total system lo
ss
as well
as g
enerate po
wer sche
dule
to minimize the
investment cost with
out
i
n
sta
lling any
external
devi
c
es and to
m
a
ximise the
consumer welfare
by avoiding a
n
y load cu
rtai
lment for co
n
ges
tio
n
mana
gement un
de
r deregul
ated
environm
ent.
In this pap
er,
the basi
c
co
nce
p
t of GA based lo
ss
o
p
timization i
s
laid und
er t
he OPF
model
whe
r
e
loss i
s
fun
c
tion of B-coef
ficients fo
r a
c
tive po
we
r
and
C-coeffi
cient for
rea
c
ti
ve
power. With
the optimise
d
loss value
s
, the
gene
rator co
ntrib
u
tion also has been found
out
throug
h thi
s
p
r
opo
se
d m
o
d
e
l. In the
pa
p
e
r, a
n
ove
r
lo
a
d
ing i
ndex
(O
I) ha
s
also
be
en p
r
o
p
o
s
ed
to
find out the conge
sted l
i
nes for a
n
y type of
contingen
cy. The gene
ratio
n
schedul
e, with
optimize
d
lo
ss al
so
ha
s b
een u
s
e
d
to
relief line
ove
r
flow
witho
u
t load
cu
rtailm
ent even
duri
ng
contin
gen
cy. Thro
ugh
cl
assical an
alysis, volt
ag
e profile
h
a
s b
een che
c
ked
with
p
r
opo
sed
gene
ration schedul
e,
whi
c
h
yields satisf
actory re
su
lt. Actually GA
analysi
s
hel
p
s
to find o
u
t the
most effective gene
ration
sch
edule, which h
a
s
be
e
n
use
d
for conge
stion m
anag
ement a
l
ong
with lo
ss
opti
m
isation
of the overall
syst
em. It also
a
s
sist
s to come
to a de
cisi
on
that no external
comp
en
satio
n
, load cu
rtail
m
ent is re
quired for c
onge
stion manag
e
m
ent up to a certai
n limit in a
dere
gulate
d
p
o
we
r environ
ment.
2. Proposed
M
e
thod
In this pape
r, the flow of prop
osed me
thod is
d
e
vided in two
ste
p
s. In the first step,
gene
rato
r co
ntribution
ca
n
be dete
r
min
ed ba
se
d on
loss optimi
z
ation u
s
ing
G
A
and the
s
e
re-
contri
bution
s
of gen
erators have
bee
n
use
d
to
reli
ef the
co
nge
st
ed tran
smissi
on lin
es du
ri
ng
contin
gen
cy in the next step.
2.1. Loss Op
timization us
ing GA
GA is a glo
b
a
l adaptive
search te
chni
q
ue ba
sed
on
the mechani
cs of natu
r
al g
enetics
[9]. It is applied to optimi
z
e existing
sol
u
tions
by
usi
ng biol
ogi
cal
evolution b
a
sed metho
d
s.It
has
many appli
c
a
t
ions in certai
n types of problem
s
that yield better results than the
comm
only used
method
s with
out
any co
m
p
licate
d
cla
s
sical cal
c
ul
at
ion. To
solve a specifi
c
problem with GA,
a
function
kn
o
w
n, a
s
obj
ect
i
ve function
need
s to b
e
con
s
tru
c
ted
whi
c
h all
o
ws different p
o
ssible
solutions to
be evaluated. The al
gorit
hm will th
en
take those solutions, whi
c
h
seem to
show
s
o
me ac
tivity towards
a
work
ing s
o
lution.
2.1.1.
Problem for
m
ulation co
nsidering O
P
F
The obje
c
tive
function for
convention
a
l cost optimization is as follows
Minimize
1
n
NG
n
F
C
=
=
∑
$/hr
(1)
whe
r
e,
(
)
2
00
ng
i
g
i
CA
P
B
P
C
=+
+
(2)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
¢
Con
g
e
s
tion
m
anagem
ent usin
g m
u
ltiobjective GA (S
awa
n
Sen)
3
But in the pro
posed meth
o
d
, the obje
c
ti
ve function
consi
deri
ng th
e total active
and
rea
c
tive l
o
ss
can b
e
formul
ated as follo
ws
()
/
F
x
=minimi
z
e
()
11
nm
PP
B
P
ij
L
Gi
Gj
ij
=
∑∑
==
00
0
11
1
nn
m
BB
P
P
B
P
ij
iG
i
G
i
G
j
ii
j
=+
+
∑∑
∑
==
=
(3)
()
//
F
x
=minimi
z
e
()
11
nm
QQ
C
Q
ij
L
Gi
Gj
ij
=
∑∑
==
00
0
11
1
nn
m
CC
Q
Q
C
Q
ij
iG
i
G
i
G
j
ii
j
=+
+
∑∑
∑
==
=
(4)
P
L
and Q
L,
the
active and
re
active loss te
rms can b
e
expre
s
sed u
s
in
g B and C-co
efficient [12] as
follows
:
11
nm
PP
B
P
ij
L
Gi
Gj
ij
=
∑∑
==
00
0
11
1
nn
m
BB
P
P
B
P
ij
iG
i
G
i
G
j
ii
j
=+
+
∑∑
∑
==
=
(5)
w
h
er
e
cos
(
)
cos
c
os
|
|
|
|
ij
i
j
ij
ij
i
j
R
B
VV
θ
θ
ϕϕ
−
=
,
()
0
1
m
BB
B
P
ij
j
i
Dj
i
j
=−
+
∑
=
and
00
11
nm
BP
B
P
ij
Di
Dj
ij
=
∑∑
==
(6)
11
nm
QQ
C
Q
ij
L
Gi
Gj
ij
=
∑∑
==
00
0
11
1
nn
m
CC
Q
Q
C
Q
ij
iG
i
G
i
G
j
ii
j
=+
+
∑∑
∑
==
=
(7)
cos
(
)
cos
c
os
|
|
|
|
X
ij
i
j
C
ij
VV
ij
i
j
θθ
ϕϕ
−
=
,
()
0
1
m
CC
C
Q
ij
j
i
Dj
i
j
=−
+
∑
=
and
00
11
nm
CQ
C
Q
ij
Di
D
j
ij
=
∑∑
==
(8)
whe
r
e
j
jj
θ
δϕ
=−
and
j
jj
θ
δϕ
=−
The ineq
ualit
y or gene
rato
r output const
r
aints
mi
n
0
m
a
x
g
ig
i
g
i
P
PP
≤≤
(9)
mi
n
0
ma
x
g
ig
i
g
i
QQ
Q
≤≤
(10
)
mi
n
m
a
x
j
gi
gi
gi
PP
P
Δ≤
Δ
≤
Δ
(11
)
mi
n
m
a
x
j
gi
gi
g
i
QQ
Q
Δ≤
Δ
≤
Δ
(12
)
Voltage co
nst
r
aint:
mi
n
m
ax
j
ii
i
VV
V
≤≤
(13
)
2.1.2.
Problem Encoding
Each
control
variable is
called a ge
ne,
while all co
ntrol varia
b
le
s integ
r
ated i
n
to one
vector i
s
call
ed a
chromo
some. T
he
GA alway
s
d
eals
with a
set of ch
romo
some
s
calle
d
a
popul
ation. T
r
an
sformi
ng
chromo
som
e
s from
a pop
ul
ation, a ne
w po
pulatio
n
is obtain
ed,
i.e.,
next gene
rati
on is fo
rme
d
. It need
s thre
e gen
etic o
p
e
rato
rs:
sele
ction, crossov
e
r, an
d mutat
i
on
for this
purpos
e.
Evaluation Warning : The document was created with Spire.PDF for Python.
¢
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 9, No. 1, April 2011 : 1 – 8
4
2.1.3. Initializ
ation
Usually, at th
e be
ginni
ng
of the
GA opt
imizat
ion
p
r
o
c
e
ss,
ea
ch v
a
riabl
e g
e
ts
a rando
m
value from its pre
define
d
domain. Th
e
gener
ator p
o
we
r output
s have well-d
e
fined lo
wer
and
uppe
r limits, and the initiali
zation p
r
o
c
ed
ur
e comme
nces with the
s
e
limits given b
y
mi
n
m
ax
PP
P
GG
G
i
ii
≤≤
and
mi
n
m
ax
QQ
Q
GG
G
ii
i
≤≤
(14
)
2.1.4.
Cons
train
t
function
s an
d paren
t
sel
ection
Implementati
on of a
prob
lem in a
ge
n
e
tic al
g
o
rith
m is
reali
z
e
d
within th
e constraint
function. T
h
e propo
se
d
approa
ch
uses th
e
conv
entional
po
wer b
a
lan
c
e
equatio
n a
s
its
con
s
trai
nt whi
c
h can be
wri
tten as
1
1
n
PP
P
DL
G
i
i
ε
=−
−
∑
=
and
2
1
n
QQ
Q
DL
G
i
i
ε
=−
−
∑
=
(15
)
The conve
r
g
ence
is obtai
ned whe
n
1
ε
f
o
r
ac
tive loss
and
2
ε
f
o
r re
act
i
v
e
lo
ss l
e
ss t
h
an a
toleran
c
e. Im
provem
ent of the averag
e
fitness
of the popul
ation is
achi
eved thro
ugh sele
ction
of
individual
s a
s
pare
n
ts from
the co
mplete
d pop
ulati
on.
The
sele
ction
is pe
rform
e
d
in su
ch
a wa
y,
that chromo
somes h
a
ving
highe
r fitness are mo
re like
l
y to be select
ed as p
a
re
nts.
2.1.5.
Cros
sov
e
r and Muta
tion
After the sele
ction, GA ap
plies a rand
o
m
gene
ration
to cut the string
s at any positio
n
(the
crossove
r poi
nt) a
nd
excha
nge
s th
e sub
s
tr
ing
s
betwe
en
the
two chromo
somes. On
ce
t
he
cro
s
sove
r is
perfo
rmed, th
e new
chrom
o
som
e
s a
r
e
adde
d to the new p
opul
ation set. Mutat
i
on
being
an
othe
r pa
ram
e
ter,
it involves
ra
ndomly
sele
cting
ge
ne
s within
the ch
ro
moso
me
s
a
n
d
assigni
ng the
m
rand
om va
lues
within th
e co
rre
sp
ond
ing predefine
d
interval. Th
e pro
bability of
mutation
i
s
n
o
rmally ke
pt very
lo
w,
a
s
high
mutation
rate
s co
uld degrade
the
evolving
p
r
o
c
ess
into a rand
om
search p
r
o
c
e
ss.
2.2. Conge
stion Manage
ment
w
i
th
Re- Co
ntribu
tion of Gen
e
r
a
tors
With the re-contributio
n of
gene
rato
rs
u
s
i
ng equ
ation
(3) and (4
),
con
g
e
s
ted
lin
es can
be relieve
d during
contin
g
ency. To find the cont
ing
e
n
t
lines durin
g contin
gen
cy, an Overlo
adi
ng
Index can b
e
defined
as ch
ange i
n
p
o
we
r flow thro
ugh
a tra
n
smi
s
si
on line
du
ring
co
ntingen
cy
of
other line
s
. M
a
thematically it
can be expressed a
s
follo
ws
mn
m
n
mn
mn
PP
P
μ
−
=
(16
)
whe
r
e,
mn
P
and
mn
P
are the
active
power flo
w
th
roug
h the lin
e
m-n afte
r co
ntingen
cy an
d befo
r
e
contin
gen
cy resp
ectively. High
er value
of this
index indicate
s th
e more
con
g
e
sted lin
e in the
power net
wo
rk.
3. Results a
nd Discu
ssi
on
The fe
asibilit
y and
effectivene
ss of the
prop
osed
me
thod h
a
s be
e
n
dem
on
strat
ed in
the
IEEE 30 bus tes
t
s
y
s
t
em as sk
et
c
h
ed in Figur
e 1.
The tes
t
s
y
s
t
em
and produc
t
ion
unit
s
’
prop
ertie
s
a
r
e given in
Ta
bles
1 an
d 2.
For th
e
enti
r
e sim
u
lation,
logic
program
in GA h
a
s
b
een
employed to
formulate a
c
po
we
r flow model. The
stand
ard
pa
ramete
rs
sett
ings fo
r all the
simulatio
n
s of
the adopted
GA have bee
n depi
cted in
Table 3.
Thro
ugh
pro
posed
optimi
z
ation
meth
od, GA
, the
optimized
values of
sche
dule
d
gene
ration
for all
GE
NCOs
ha
s b
e
e
n
dete
r
min
e
d
con
s
ide
r
in
g all
equ
ality and i
neq
u
a
lity
con
s
trai
nts
of optimal
po
wer flo
w
a
s
m
entione
d
in
(5) to
(1
3)
an
d by ta
king
both a
c
tive a
nd
reactive losses as obj
ective func
tions. Table 4 illustra
tes a compari
s
on
of the
sol
u
tions
obtained
by conventio
nal co
st opti
m
ization m
e
thod who
s
e fi
tness fun
c
tion
has b
een d
e
scribe
d in (1) and
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
¢
Con
g
e
s
tion
m
anagem
ent usin
g m
u
ltiobjective GA (S
awa
n
Sen)
5
prop
osed m
u
l
t
iobjective o
p
t
imization m
e
thod mai
n
tain
ing the
re
al a
nd rea
c
tive lo
ss ((3) an
d (4
))
for a fixed act
i
ve demand o
f
283.6 MW a
nd rea
c
tive d
e
mand of 12
6
.
2 MVAR.
Figure 1. Single line diag
ram (SLD) of
IEEE 30 bus
tes
t
s
y
s
t
em
Table 1. Te
st system p
r
op
e
r
ties
Name of param
e
ters
Value
Number of
buses
30
Number of
gener
ator units
6
Number of
branc
hes
43
Number of
tie lines
6
Total po
w
e
r dem
and in MW
283.6
Table 2. Prod
uction u
n
its’ p
r
ope
rtie
s
Gene
rator
no
P
ma
x
MW
P
mi
n
MW
Q
ma
x
MV
AR
Q
mi
n
MV
AR
1 150
50
-2
-5
2 70
50
-0.3
-0.9
3 40
10
30
10
4 50
10
30
10
5 30
10
30
10
6 30
10
40
10
Table 3. Parameter
settin
g
of GA base
d
optimizatio
n
Name of the p
a
r
a
meters
Value
Population size
20
Selection
stochastic
unifor
m
Mutation adaptive
feasible
Cr
ossover scatter
ed
Table 4. Co
m
pari
s
on of ge
nerato
r
s co
ntribution
s
obta
i
ned from
con
v
entional cost
optimization
method (m
ethod 1) a
nd p
r
opo
sed multi
obj
e
c
tive optimization m
e
thod (m
ethod
2)
Contribution of
Generato
r
s
Method 1
Method 2
GE
NC
O 1
G
P
(p. u.)
1.384
1.235
G
Q
(p. u.)
-0.185
-0.02
GE
NC
O 2
G
P
(p. u.)
0.575
0.682
G
Q
(p. u.)
-0.0056
-0.0065
GE
NC
O 3
G
P
(p. u.)
0.245
0.339
G
Q
(p. u.)
0.212
0.204
GE
NC
O 4
G
P
(p. u.)
0.35 0.334
G
Q
(p. u.)
0.267
0.254
GE
NC
O 5
G
P
(p. u.).
0.179
0.105
G
Q
(p. u.)
0.241
0.247
GE
NC
O 6
G
P
(p. u.)
0.169
0.207
G
Q
(p. u.)
0.317
0.318
sy
stem loss
L
P
(p.u.)
0.074
0.067
The
cha
nge
s in re
al an
d reactive p
o
we
r contrib
u
tion
for all GE
NCOs
(Figu
r
e
2) are
with in th
eir
spe
c
ified limit
as d
e
scri
bed
in (11
)
to (1
2). W
hen a
line is tri
ppe
d
by a su
dde
n
fault, there is a
possibility of anothe
r line
overflow b
e
cause t
he po
wer,
whi
c
h h
ad flowe
d
through the tri
p
ped
line, should
flow
elsewh
ere. In the
s
e
circum
st
an
ce
s, a
rem
edia
l
actio
n
ha
s
to be ta
ke
n
to
maint
a
in
t
he sy
st
em se
curi
t
y
.
Evaluation Warning : The document was created with Spire.PDF for Python.
¢
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 9, No. 1, April 2011 : 1 – 8
6
Figure 2. Cha
nge in Real a
nd Peactive
Powe
r in
p.u. for Multiobje
c
tive optimization method
The seve
rity of the line
fault depend
s on
the am
ount of power, whi
c
h ha
d flowed
throug
h the t
r
ippe
d line. I
n
this te
st, five di
fferent li
nes
have b
e
en tripp
ed th
at we
re
cho
s
en
according to
the amou
nt of line flow co
nsid
ere
d
. Ta
ble 5 shows
the simul
a
tio
n
re
sults. Fo
r a
particula
r line
fault, five most cong
este
d line
s
have
been fo
und
with the hel
p
of the pro
p
o
s
ed
overloa
d
ing i
ndex. With t
he calculate
d
co
nt
ributio
n of GENCOs
(Tabl
e 4
)
usi
ng p
r
op
ose
d
optimizatio
n; co
nge
sted
l
i
nes can
be
relieve
d fro
m
overl
oadin
g
(T
able
5).
A rem
a
rka
b
le
redu
ction
in
li
ne
con
g
e
s
tio
n
ha
s bee
n
o
b
se
rved
with
prop
osed
ge
neratio
n a
s
compa
r
ed
with
the
gene
ration
o
b
tained from
conve
n
tion
al co
st opt
i
m
ization te
chniqu
e. This redu
ction
of
overloa
d
ing
may redu
ce t
he co
st of co
nge
stion whi
c
h is
an inte
gral p
a
rt of lo
cation
al marg
inal
price (LMP
) [13] in dere
gul
ated
enviro
n
m
ent of powe
r
system.
Table 5. Co
n
gestio
n
mana
gement with
new
sched
ule
of generato
r
s co
ntributio
n
Line
fault
F
i
ve most
congested
line
Power flo
w
in p.u
.
Before
Fault
(A)
After
fault w
i
th
generation
using
Method 1
(B)
After fault
w
i
th
generation
using
Method 2
(C)
%
Overload
*1
0
0
BA
A
−
⎛⎞
⎜⎟
⎝⎠
% Ove
r
load
w
i
th
proposed
generation
*100
CA
A
−
⎛⎞
⎜⎟
⎝⎠
Case 1
2-4
2-6
0.3802
0.5277
0.4099
38.79
7.80
3-4
0.4481
0.5552
0.4563
23.91
1.82
1-3
0.4816
0.5937
0.5039
23.27
4.63
2-5
0.5802
0.6379
0.5946
9.94
2.48
10-17
0.6627
0.7176
0.6794
8.28
2.51
Case 2
2-5
2-6
0.3803
0.6666
0.4036
75.30
6.12
2-4
0.2911
0.5043
0.3186
73.22
9.44
12-16
0.0594
0.0837
0.0604
40.90
1.68
24-25
0.0280
0.0316
0.0290
12.91
3.57
4-12
0.2532
0.2767
0.2657
9.292
4.93
Case 3
6-7
8-6
0.0155
0.0255
0.0157
63.97
1.29
6-9
0.2076
0.2489
0.2153
19.88
3.70
9-10
0.1734
0.1973
0.1872
13.74
7.95
6-10
0.1096
0.1214
0.1132
10.70
3.28
1-2
0.9073
1.0009
0.9145
10.31
0.79
Case 4
12-15
14-15
0.0535
0.1032
0.0557
92.98
4.11
12-16
0.0594
0.1000
0.0692
68.19
16.4
22-24
0.0943
0.1237
0.1074
31.16
13.8
4-6
0.4027
0.4416
0.4116
9.66
2.21
1-2
0.9073
0.9141
0.9136
0.73
0.69
Case 5
4-12
6-10
0.1091
0.1689
0.1126
67.10
3.20
4-6
0.4027
0.6208
0.4100
54.17
1.81
12-13
0.1091
0.1689
0.1126
14.19
3.20
2-6
0.3802
0.4201
0.3993
10.47
5.02
2-5
0.5802
0.5938
0.5874
2.34
1.24
The othe
r important a
s
pe
ct of this pro
pos
ed metho
d
is the redu
ction of the syste
m
operating lo
sse
s
du
ring
continge
ncy.
Table 6
co
m
pare
s
the l
o
sse
s
with
orig
inal contrib
u
tion
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
¢
Con
g
e
s
tion
m
anagem
ent usin
g m
u
ltiobjective GA (S
awa
n
Sen)
7
(metho
d 1
)
a
nd re-co
n
trib
ution of
GENCO
s
(metho
d
2)
du
ring
co
ntingen
cie
s
.
As sho
w
n, th
e
system o
p
e
r
ating losse
s
have de
cre
a
se
d by
a con
s
id
era
b
le
amount with
re
-contrib
u
t
ion
sched
ule
of
GENCO
s
. He
nce, it
ca
n b
e
state
d
t
hat
along
with
th
e redu
ction
o
f
con
getion
cost
this propo
sed
method can
lowe
r sy
stem
operating lo
ose
s
an
d the
r
eby the spot
price of ene
rg
y
in dere
gulate
d
environ
men
t
.
The ab
ove a
d
vantage
s of
this re
-contri
b
u
tion sch
edul
e rem
a
in
s le
ss con
s
eq
uent
unle
ss
it has lea
s
t e
ffect on the o
peratin
g con
d
itions
of the
system. The
voltage profil
es sho
w
n in t
h
e
Figures 3 to
5 for the test
cases,
stren
g
then
s
the competen
cy of
the re-co
n
tri
bution sch
e
d
u
le
with re
spe
c
t
t
o
the no
rmal gene
ration
schedul
e.
Durin
g
conting
e
n
cy the voltage
profile
rem
a
in
s
least affecte
d
with the imp
o
se
d sche
dul
e. It implie
s that without af
fecting the vo
ltage, this ne
w
sched
ule ca
n
offer significant benefits li
ke minimi
za
ti
on of system
losse
s
and lo
cation
al marg
inal
prices (L
MP)
in term
s of
co
nge
stion ma
n
ageme
n
t
du
ri
ng conting
e
n
c
y and i
n
no
rmal condition
of
the system. T
he othe
r imp
o
rtant
adva
n
tage of calcul
ated re
-contri
bution sch
e
d
u
le that ha
s b
een
prep
ared by
GA by optimi
z
ing th
e a
c
tive and
rea
c
tive
losse
s
in
d
e
reg
u
lated
el
ectri
c
ity market is
that it does n
o
t threaten th
e eco
nomi
c
d
i
spat
ch.
Table 6. Co
m
pari
s
on of rea
l
losses
with orig
in
al co
ntri
bution an
d re
-co
n
trib
ution
sched
ule
of GENCOs d
u
ring
contin
g
ency
Cases P
L
in p.u.
w
i
th
% reduction in a
c
tive loss w
i
th
Re-contribution o
f
GENC
Os
Original contribut
ion of
GE
NC
Os
Re-contribution o
f
GE
NC
Os
Case 1
0.0798
0.0733
8.14
Case 2
0.1427
0.1236
13.38
Case 3
0.0891
0.0775
13.01
Case 4
0.0809
0.0742
8.28
Case 5
0.0813
0.0719
11.48
Figure 3. Co
mpari
s
o
n
of voltage du
ring
contin
gen
cy with origi
nal a
nd ne
w gen
eration
for Ca
se 1 an
d 2
Figure 4. Co
mpari
s
o
n
of voltage du
ring
contin
gen
cy with origi
nal a
nd ne
w gen
eration
for Ca
se 3 an
d 4
Evaluation Warning : The document was created with Spire.PDF for Python.
¢
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 9, No. 1, April 2011 : 1 – 8
8
Figure 5. Co
mpari
s
o
n
of voltage du
ring
contin
gen
cy with origi
nal a
nd ne
w gen
eration
for Cas
e
5
4. Conclusio
n
This p
ape
r p
r
opo
se
s a ne
w metho
d
, which
not only
optimize
s
rea
l
and rea
c
tive power
loss usin
g G
A
, but also redu
ce
s con
g
e
stion of
tran
smissio
n
line
s
duri
ng cont
ingen
cy by using
new
sche
dul
e of gen
erators contrib
u
tions. T
he ov
e
r
loadi
ng in
de
x, propo
se
d i
n
this p
ape
r
can
efficiently tra
c
e th
e
co
nge
sted li
ne
s in
contin
gen
cy
so th
at a
re
medial
a
c
tion
ca
n
be ta
ke
n to
relief the line
from con
g
e
s
tion. In this paper
, the
new g
ene
rati
on sche
dule
obtaine
d by th
e
prop
osed lo
ss optimisatio
n GA model has be
en
taken as a corrective mea
s
u
r
e for co
nge
stion
manag
eme
n
t. This l
o
ss o
p
t
imisation b
a
s
ed
gen
erati
on sch
edul
e
has
bee
n co
upled
with lo
ad
flow to
ch
eck the
voltag
e profile of t
he
sy
stem al
ong with co
n
gestio
n
ma
n
ageme
n
t
of
t
h
e
transmissio
n
line
s
du
rin
g
continge
ncy. The te
st
re
sults sho
w
that th
e
new sch
edul
e of
gene
ration i
s
a po
werful to
ol for cong
est
i
on ma
n
agem
ent schem
e o
v
er the othe
r
scheme
s
su
ch
as loa
d
cu
rtai
lment and FA
CTS device inclu
s
io
n.
Referen
ces
[1]
K
y
u
ng M, S
a
n
g
-H
ye
on H, S
u
-W
on L, Yo
un
g-H
y
u
n
M. T
r
a
n
smissio
n
l
o
ss
all
o
catio
n
a
l
g
o
rithm us
in
g
path-i
n
tegr
al b
a
sed o
n
transa
c
tion strateg
y
.
IEEE Transaction on Power
System
s
. 20
10
; 25(1):195
-
205.
[2]
Sch
w
e
ppe F
,
Carama
nis
R, T
a
rbor R, Boh
n
R. Spot
pr
ici
ng of El
ectricit
y. Klu
w
er Ac
ade
mic Publ
ish
e
r.
198
8.
[3]
Gross G, T
ao S. A ph
ys
ical-f
lo
w
-
b
a
se
d a
p
p
r
oach
to a
lloc
a
ting tra
n
smissi
on l
o
sses
in
a
transacti
o
n
frame
w
ork.
IEEE Transaction on Power System
s
. 20
00; 1
5
(2): 631
–6
37.
[4]
Cha
ng Y-C,
Lu C-N. A
n
electricit
y traci
ng
meth
od
with ap
plic
atio
n
t
o po
w
e
r loss
all
o
catio
n
.
Internatio
na
l Journ
a
l of Electr
ical Pow
e
r Ene
r
gy Systems
. 2
001; 23(
1): 13-
17.
[5]
Abde
lkar S.
Efficient
comp
utat
on
alg
o
rithm
fo
r calcu
l
ati
ng
lo
ad c
ontrib
u
tio
n
s
to l
i
ne
flo
w
s
and
loss
es
.
IEE proc. Of
Gener
ation, T
r
an
smiss
i
on
and
Distributi
o
n
. 20
06; 153(
4): 391
-398.
[6]
Abed
lkar
S. T
r
ansmissi
on
los
s
al
locati
on
in
a d
e
re
gul
ated
electric
al
en
er
g
y
mark
et.
Electrical P
o
wer
System a
nd R
e
searc
h
. 200
6; 76(11): 96
2-9
67.
[7]
Nep
o
muce
no L
,
Santos Jr. A.
Equiv
a
le
nt opti
m
izat
io
n mode
l
for loss minimi
zation: A suita
b
le a
nal
ys
is
appr
oach
. IEE
E
T
r
ansaction
on Pow
e
r Systems
. 1
997; 1
2
(
4
):140
3-1
412.
[8]
Jen-H
ao T
.
Po
w
e
r fl
o
w
an
d l
o
ss al
locati
on
fo
r der
egu
late
d t
r
ansmissi
on
s
ystem.
Internati
ona
l Jo
urn
a
l
of Eletrical Po
w
e
r and Ener
g
y
System.
200
5; 27 (1): 327-
333.
[9]
Mohd H
e
r
w
a
n
S, Mohd W
a
zir
M, Omar A.
Trans
missi
on l
o
ss and lo
ad flo
w
allocatio
n
s via Geneti
c
Algorit
h
m
tech
niq
u
e
. IEEE
T
encon 2009.
[10]
Parna
ndi
S,
Schod
er K, F
e
liac
h
i
A.
Po
w
e
r Market A
nalysis
too
l
fo
r Con
gesti
on
ma
na
ge
me
nt.
Internatio
na
l C
onfere
n
ce o
n
Electrical E
ngi
ne
erin
g Desi
gn a
nd T
e
chnol
ogi
es. 2007: 4-
6.
[11]
H
w
a-S, S
eun
g
M.
A N
e
w
Operatio
n Stra
tegy of
Seri
es
Co
mpens
atin
g D
e
vice
u
n
d
e
r li
ne
flow
cong
estio
n
usi
ng li
near
i
z
e
d
l
i
ne flow
sensiti
v
it
y
.
Po
w
e
r En
gin
eeri
ng Soc
e
it
y
W
i
nter Me
eting IEE
E
200
1; 2: 416-4
20.
[12]
Chakr
abarti A,
Hald
er S. Po
w
e
r S
y
st
em Ana
l
ysis, Op
eratio
n
and
Contro
l. F
ourth Ed
itio
n P
H
I Lear
nin
g
Pvt. Ltd, India. 201
0.
[13]
Yong F
,
Z
u
yi
L.
Different M
ode
ls an
d Pro
perties o
n
LM
P Calcu
l
ati
ons
.
Po
w
e
r En
gi
ne
erin
g Soce
it
y
Genera
l
Meeti
ng IEEE. 2006.
Evaluation Warning : The document was created with Spire.PDF for Python.