Indonesian J
ournal of Ele
c
trical Engin
eering and
Computer Sci
e
nce
Vol. 2, No. 2,
May 2016, pp
. 241 ~ 247
DOI: 10.115
9
1
/ijeecs.v2.i2.pp24
1-2
4
7
241
Re
cei
v
ed O
c
t
ober 1
6
, 201
5; Revi
se
d April 13, 201
6; Acce
pted April 24, 2016
Reconfiguration of Distribution Networks with Presence
of DGs to Improving the Reliability
Amir Sabba
gh Alv
a
ni, Se
y
e
d Mehdi
Mahaei*
Irania
n
Organi
zation for En
gi
neer
ing Ord
e
r of Build
ing
Pr
o
v
ince East Aza
r
ba
yj
an Abras
a
n,
T
abriz, Iran
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: me.mahae
i@
gmail.c
o
m
A
b
st
r
a
ct
In this p
a
p
e
r, the n
e
tw
ork rec
onfig
uratio
n i
n
t
he pr
esenc
e o
f
distribut
ed
ge
nerati
on
units
w
i
th th
e
ai
m of improvi
ng the reli
ab
ilit
y of t
he netw
o
rk is studied. For this purp
o
se
four reliab
ility
para
m
eters in th
e
obj
ective fu
nction
are c
ons
id
ered, w
h
ic
h is
avera
ge
en
er
gy not s
u
p
p
li
e
d
syste
m
av
er
age
interr
uptio
n
freque
ncy in
d
e
x, system a
v
erag
e interr
u
p
tion d
u
rati
on
index a
nd
mo
mentary av
erag
e interr
up
tion
freque
ncy i
n
d
e
x
. T
he n
e
w
me
thod w
i
l
l
b
e
no
rma
l
i
z
e
d
obj
ective fu
nction.
A
nother
su
ggest
i
on
of this
p
a
p
e
r
are co
nsi
deri
n
g the
differe
nt fault rat
e
s, loc
a
ting ti
me of
fa
ult
s
type a
nd
pri
o
riti
z
a
t
i
o
n
of cus
t
omers
bas
ed
on
their i
m
p
o
rtanc
e. T
h
is nonl
in
e
a
r prob
le
m has
optimi
z
e
d
by p
a
rticle sw
arm
o
p
timi
z
a
ti
on (PS
O
) algorith
m
.
Ke
y
w
ords
: Re
config
uratio
n, DG, reliab
ility, fault rate, loca
ting time
Copy
right
©
2016 In
stitu
t
e o
f
Ad
van
ced
En
g
i
n
eerin
g and
Scien
ce. All
rig
h
t
s reser
ve
d
.
1. Introduc
tion
Distri
bution
Networks
are
last p
a
rt of
the po
we
r
sy
stem a
nd fe
d vario
u
s co
nsum
ers
dire
ctly. This
pat of
syst
em
ha
s diffe
rent
chall
enge
s.
O
ne of th
es
e
challen
ges i
s
reliability. In th
is
netwo
rk,
dive
rsity of
equi
p
m
ent a
nd
direct
comm
uni
cation
with
co
nsum
ers
ha
s
cau
s
e
d
the
le
vel
of reliability is low. Variou
s
solutio
n
s h
a
ve been
p
r
op
o
s
ed to imp
r
ov
e the reliabilit
y of distribution
Network. But the reconfig
uration
of n
e
twork i
s
one
of the best m
e
thod
s of improving
relia
b
ility,
bec
au
se ha
s
v
e
ry
low co
st
.
Re
config
urati
ons can b
e
d
e
fined a
s
"th
e
proc
ess
of cha
ngin
g
the
co
nfiguration
of the
power
syste
m
by cha
ngi
ng the sw
itches
situation
to satisfy
the
ope
ration constraints." Whe
n
faced
with reco
nfiguratio
n, syst
em o
perato
r
s nee
d to chan
ge
the status
of the swit
ch
es to
minimize faults
effec
t
s
of
network loads
. In fac
t, in
reco
nfiguratio
n path f
r
om t
he
sou
r
ce to
the
load
cha
nge
so that th
e
netwo
rk is radial
and
system
relia
bility is improved. Ope
r
a
t
ion
con
s
trai
nts can be a
s
follo
ws:
•
Radi
ality of th
e netwo
rk to
be maintain
e
d
•
The new net
work will fed
all busses.
•
Load
s are not
more than n
e
twork
capa
ci
ty and produ
ction
•
Busses volta
ge and n
e
two
r
k eq
uipm
ent are withi
n
the
allowa
ble ra
nge.
•
Curre
n
t lines
and eq
uipme
n
t are within t
he allo
wable
rang
e.
By con
s
ide
r
i
ng the im
po
rtance
of net
work
re
confi
guratio
n, ma
ny studie
d
h
a
s b
een
publi
s
hed
in
this fiel
d. Publ
ishe
d
studie
d
have
bee
n
cl
assificatio
n
in
five cate
go
ri
es: evol
ution
a
ry
techni
que
s, p
a
rticle intelli
g
ence, innovat
iv
e, combinati
onal an
d anal
ytical-p
rob
abi
lity.
One of the g
eneral metho
d
s of artifici
a
l
in
telligen
ce
is evolution
a
ry tech
niqu
es. Thi
s
techni
que i
s
prop
osed
by Darwin
u
s
ing
the fund
am
ental
con
c
ept
of evolution
prop
osed. Th
is
techni
que a
r
e ran
domly g
enerated an i
n
itial popul
at
ion and the
n
usin
g the sev
e
ral
stage (e.g.,
mutation, intera
ction, etc.
) extra
c
t the
optim
um re
spo
n
se amo
ng them. Ge
netic al
gorith
m
s,
differential e
v
olution algo
rithm, taboo sea
r
ch
and
evolutiona
ry algorith
m
s in
cludi
ng meth
ods
based on ev
olutiona
ry techni
que
s tha
t
in publishe
d pape
r hav
e been p
r
op
ose
d
to solv
e
reconfigu
r
atio
n probl
em on
distrib
u
tion n
e
twork [1-10].
Particle
intell
igen
ce i
s
o
ne of
other in
telligen
ce
method
s th
at after evol
utionary
techni
que
s, is the general optimiz
atio
n method
s. The
s
e techniqu
e
s
are b
a
se on
trying creatu
r
es
like fish
es, a
n
ts an
d bee
s to live in a grou
p with th
e aim of find
ing food o
r
i
mmigratio
n [11-
18].The inn
o
v
ative techni
que
s with u
n
ique a
nd
n
e
w meth
od
s that have d
r
awn often b
a
si
c
con
c
e
p
ts
solv
e the
co
mple
x-nonlin
ear p
r
oble
m
s [
19-24]. Each te
chniqu
e h
a
s some a
d
vanta
ges
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
2, No. 2, May 2016 : 241 –
247
242
and di
sa
dvan
tages.
Re
sea
r
ch
ers b
enefi
t
from ca
pabi
lities of different algo
rithm
s
by combi
n
ing
two or mo
re i
n
telligent techniqu
e [25-2
8
].
In [29] is proposed a new method for impr
oving reliability by re
configurat
ion using
Interval a
naly
s
is techniq
u
e
s
with
reg
a
rd
to u
n
certainl
y to maximi
ze reliability i
m
provem
ent
and
pow
er
los
s
e
s
red
u
ct
io
n.
Ca
se
st
ud
ies
sh
ow the effici
en
cy of propo
sed
meth
od
for
reconfigu
r
atio
n. In [30], a
new
probabili
ty based
met
hod i
s
p
r
e
s
e
n
ted for th
e reco
nfiguratio
n to
redu
ce th
e tot
a
l co
st of swit
ch a
nd lo
sse
s
co
sts.
With
regard to time
-varying l
oad
s, the p
r
op
osed
method i
s
ab
le to achi
eve
an optimum
balan
ce
b
e
tween the n
u
m
ber of switchi
ng and l
o
sse
s
.
Several exp
e
r
iment
s
sho
w
the supe
rio
r
ity of the p
r
op
ose
d
meth
od
and th
e results a
r
e
com
p
a
r
ed
with ce
rtain
method
s in several state
s
.
Ho
wever, th
e
s
e m
e
thod
s
have di
sadva
n
tage
s
a
nd advantag
es with
respe
c
t to
ea
ch
other, but ex
perience has shown
that
methods based
on parti
c
le intelligen
ce techni
que is
approp
riate
compa
r
ed to
other te
chni
q
ues. O
ne of
t
he mo
st wi
d
e
ly use
d
opti
m
ization
met
hod
based on part
icle intelligence is PSO algorithm t
hat has advantages over
other al
gorithm
s [31].
In this p
ape
r, the re
confi
guratio
n of t
he
di
strib
u
tio
n
network i
s
done
with
DG
s for
improvem
ent
of distributi
on network reliabilit
y usi
ng the PSO algorithm.
Of course,
by
con
s
id
erin
g this su
bje
c
t that the distributi
on networks have v
a
riou
s
con
s
u
m
ers that their
sup
p
lying ha
ve not sam
e
importa
nce
and they
sh
o
u
ld be
prio
ritize fro
m
reli
a
b
ility viewpoi
nt.
Therefore,
a
n
impo
rtant i
s
sue
in th
e
netwo
rk re
co
nfiguratio
n
is prio
ritizatio
n
con
s
u
m
ers
and
applying th
e
importa
nce o
f
the con
s
um
ers in th
e
re
config
uratio
n. Also, the
fa
ult rate
ch
an
ges
durin
g network se
ction
-
by-se
c
tion shoul
d be co
ns
i
dered that in this paper i
s
stud
ied.
2. Objectiv
e
Function
The main ch
alleng
e in this step is the i
n
trodu
ction of
objective fun
c
tion. By con
s
ide
r
ing
that define
d
relia
bility indexes and
power l
o
sse
s
in th
e o
b
j
e
ctive fun
c
ti
on h
a
ve different
amount
s, no
rmali
z
ation t
e
ch
niqu
es u
s
ed to in
co
rporate th
ese
para
m
eters in the obje
c
tive
function.
Thu
s
the
value
s
of the o
b
je
ctive func
tio
n
te
rms a
r
e
divided to
before
placement va
lues.
With this tech
nique, ea
ch p
a
ram
e
ter is n
o
rmali
z
e
d
ba
sed o
n
logi
cal
and scientific amounts.
1
00
0
0
0
ny
kk
k
k
k
k
SA
I
D
I
S
A
I
F
I
A
E
N
S
M
A
I
F
I
L
o
s
s
OF
S
A
ID
I
S
A
I
F
I
A
E
N
S
M
A
IF
I
L
o
s
s
(1)
Whe
r
e,
k
and
0 in
dices are
the valu
es
before
and
aft
e
r th
e reconfi
guratio
n, resp
ectively.
In som
e
p
a
p
e
rs,
su
ch
pro
b
lems a
r
e
sol
v
ed by we
igh
t
ing coefficie
n
ts a
nd the
s
e coefficie
n
ts are
set by the
user (th
e
sum
o
f
the co
efficie
n
ts eq
ual to
1
)
. The
s
e m
e
thod
s a
r
e n
o
t suitabl
e meth
ods
for solvin
g these p
r
o
b
lem
s
and
actu
all
y
effect
of param
eters wi
th low value
s
decrea
s
e
s
on
obje
c
tive function. Whil
e in the norm
a
li
zation te
chni
que
s, the impact of each p
a
ram
e
ter is
same
on obje
c
tive functio
n
.
3.
The Con
s
tr
aints of O
p
tim
i
zation
Problem
con
s
traint
s a
r
e
consi
s
ts of two pa
rts.
T
he
first pa
rt of t
he
DG
con
s
traints
are
con
s
i
s
ts
of th
e num
be
r, a
c
tive and
rea
c
t
i
ve po
wer an
y sou
r
ce. Oth
e
r p
r
ovi
s
ion
s
con
s
trai
nts
are
the allo
wable
bus volta
ge
so that d
u
ri
n
g
the
isl
and
s,
the voltage
on the lo
ad
should
not exceed
limits.
3.1. The Con
v
ergence Co
ndition of Po
w
e
r
Flo
w
Corre
c
tive po
wer flo
w
i
s
the first ste
p
in
the placeme
n
t and dete
r
mines th
e ca
pacity of
the DG
s. Wh
ile the powe
r
system load
flow probl
e
m
s se
em
s is simple, but it is important
on
probl
em re
sul
t
s. Equation (2) and
(3)
sh
ow the
a
c
tive and re
active
power flow
re
lationship
s
.
n
1
j
ij
j
i
ij
j
i
di
gi
0
)
cos(
Y
V
V
P
P
(2)
n
1
j
ij
j
i
ij
j
i
di
gi
0
)
sin(
Y
V
V
Q
Q
(3)
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
Re
config
urati
on of Dist
ribu
tion Netw
orks with Presen
ce of DGs to
…
(Se
y
e
d
Me
hdi Maha
ei)
243
3.2. The Bala
nce of Po
w
e
r
Produ
ce
d po
wer
on Slack bus an
d distributed ge
ne
ration unit
s
should b
e
equ
al with
sum of po
we
r losses a
nd total load
s accordin
g equ
ation (4
).
L
N
1
i
Di
N
1
i
DGi
Slack
P
P
P
P
(4)
3.3. Range o
f
Produce
d
Activ
e
and Reactiv
e Po
w
e
r Distrib
u
te
d Gener
a
tio
n
Units
The pro
d
u
c
e
d
active and reactive po
we
r dist
rib
u
ted g
eneration unit
s
don’t must
be more
than ca
pa
city of these unit
s
.
max
DGi
DGi
min
DGi
max
DGi
DGi
min
DGi
P
P
P
Q
Q
Q
(5)
3.4. Range o
f
Ne
t
w
o
r
k Lo
sses
If you add DG in non-opti
m
al point increase po
we
r transmi
ssion lo
sses thu
s
call
will not
be acce
pted.
)
withoutDG
(
Loss
)
withDG
(
Loss
k
k
(6)
3.5. Range o
f
Bus Voltag
e
Installation of
distribute
d
generation uni
ts
shoul
d not
incre
a
se a bus voltage
greate
r
than (1.05 p
u
)
or redu
ce le
ss tha
n
(0.95
pu).
max
i
i
min
i
V
V
V
(7)
3.6. Range o
f
Curr
ent Flo
w
throug
h Line
The propo
sal
to install distri
buted ge
nera
t
ion uni
ts sho
u
ld not increa
se the current
flow
throug
h line
s
more tha
n
no
minal value, in
fac
t, thes
e limits
s
h
ows
current limits
.
max
i
i
I
I
(8)
In the above equatio
ns
V
i
: voltage of
i
th
bus
P
ij
ac
tive power flow from bus
i to j
P
gi
, Q
gi
: Production of a
c
tive and re
activ
e
power at bu
s i
P
di
, Q
di
: active and re
activ
e
load
s at bu
s i
V'
s
,
δ
'
s
: amount and angl
es of bus voltag
e
Y
ij
: admittance matrix
4.
The Optimization Algori
t
hm
The o
p
timizat
i
on alg
o
rithm
used in
this
pape
r i
s
PSO
algo
rithm th
at ca
n be
ex
pre
s
sed
with bello
w st
eps [32]:
4.1. Random
Amount o
f
a
Particle in Societ
y
w
i
th D Dimension
a
l Search Sp
ace
For each pa
rticle
Initiali
z
e
par
t
icle
End
Algorithm PS
O is po
pulati
on-b
a
sed
alg
o
rithm,
whi
c
h
mean
s that
many pa
rticl
e
s try to
find optimal point. The first step is p
o
pulation
of random p
opul
ation that is called p
r
im
ary
popul
ation, re
spe
c
tively. Usually
the nu
mbers of pri
m
ary pa
rticle
s are betwee
n
10 up to 40
, but
for mo
st of the pro
b
lem
s
,
10 pa
rticle
s a
r
e suffici
ent. To
solve specific
a
nd
com
p
lex problem
s
, it
can
be
100
or 2
00 p
a
rti
c
l
e
s. Th
e al
go
rithm
shoul
d
be written
so
that pa
rticle
s a
r
e
within t
he
rang
e of the sea
r
ch sp
ace
.
To initialize a
particle be
tween two ra
nge
s, the followin
g
equati
o
n
sho
u
ld ap
ply:
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
2, No. 2, May 2016 : 241 –
247
244
0,
1
ui
i
R
an
d
b
b
b
(9)
Whe
r
e, Ra
nd
(0, 1) sho
w
s the random
numbe
r between 0 and 1.
b
u
is the upper bou
nd of the
rang
e and b
i
is the lowe
r bound of the rang
e. Not
e
the size of the populati
on don’t ch
a
nge
durin
g the opt
imization p
r
o
c
e
ss.
4.2. Asse
ss
ment of the
Particles Fitness
Do
For e
a
ch parti
cle
C
a
lculate fitness value
If
the fitness value is better than the
best fitness valu
e in history
S
e
t current value
as the new pe
rsonal best
End
The pu
rp
ose of the fitness
is creatin
g a
si
gnifi
cant, m
easura
b
le a
n
d
com
p
a
r
able
amount
for qu
ality asse
ssm
ent. O
p
timization
re
sults
sh
ow th
at the u
s
ed
p
a
rticle
is
ho
w much go
od
or
bad. After cre
a
ting pop
ulation, amount of
assessme
nt must be
calculated for ea
ch particl
e. Each
particl
e h
a
s
a propo
rtion
that it is
call
ed the
"be
s
t part".
Thi
s
p
a
rticle
is
the
best point of
the
same p
a
rti
c
le
untie now. After the calcul
ation of fi
tness, it's com
pared with
be
st particl
e fitness.
If current fitness i
s
better, it will create the new particl
e
.
4.3. Recor
d
the Bes
t
Po
int of Each Particle, p
best,i
, a
nd Ov
erall B
est Point, g
b
est
Cho
o
se pa
rticle
with be
st fitness val
ue
of
all parti
cle
as the gl
obal
best Parti
c
le
swarm
optimizatio
n,
the overall o
p
timum loo
k
i
ng ste
m
s.
In fac
t, the bes
t fit of all has
been the
bes
t
overall value.
Thus all pa
rti
c
le
s are a
b
le
to move smo
o
thly to the best neig
hbo
r.
4.4. Upda
te the Velocit
y
Vector a
nd
the Vecto
r
Position of Eac
h
Particle
For each pa
rticle
Calculate part
i
cle velo
city
Update pa
rticle position
End
This step
is necessa
ry
fo
r
eve
r
y pa
rticle an
d it i
s
consi
s
ted
of t
w
o
part
s
, sp
eed
and
positio
n. Each parti
cle u
pdate the speed a
nd
it'
s
po
sition b
a
se
d on giv
e
s the follo
wing
equatio
ns:
1
11
2
2
k
k
kk
kk
id
id
i
i
d
i
id
v
w
v
c
r
pbe
st
x
c
r
gbe
st
x
(10
)
11
kk
k
id
id
id
x
xv
Whe
r
e:
W: weig
ht of inertia
C
1
, C
2
: accel
e
ration fa
ctors
r
1
, r
2
: two ran
dom num
ber i
n
the rang
e [0,1]
p
best,i,k
: The p
o
sition of i
th
particle at
k
th
iteration
g
best,k
: The overall situ
ation
at k
th
iteration
4.5. Repea
t
2
up to 4 Step
s to Satis
f
y
Stopping Criterion
Algorithm u
n
til a stop
ping
certai
n condit
i
on is
sati
sfie
d co
ntinue
s.
This
con
d
itio
n ca
n be
one of the followin
g
:
• Achieve the highe
st numb
e
r of rep
eat
• Achieve the highe
st numb
e
r of rep
eat a
fter the latest update
s
g
best
• Determin
e a
predefin
ed a
m
ount of fitness
• Update velo
city near
zero
Maximum nu
mber of iterati
ons to run the
algorit
hm i
s
usu
a
lly simpl
e
st stop
ping
crite
r
ion.
For e
a
ch parti
cle
Initiali
z
e
pa
rticle
End
Do
For each p
a
rticle
Calculate fitness value
If the fitness value is better than the best fitness value in history
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
Re
config
urati
on of Dist
ribu
tion Netw
orks with Presen
ce of DGs to
…
(Se
y
e
d
Me
hdi Maha
ei)
245
Set curre
n
t
value as the new personal best
End
Choose particle with best fitness val
ue of all particle as the global best
End
For each pa
rticle
Calculate par
ticle velo
city
Update p
a
rticle position
End
W
h
ile m
a
xim
u
m
iteration
5
.
C
a
se
St
u
d
i
es
For
ca
se
st
u
d
ies,
6
9
b
u
s
s
e
s
net
wo
rk
is
u
s
ed. T
h
e
simul
a
tion
wa
s pe
rform
ed u
s
ing
MATLAB software. Value
s
of PSO algorithm, W, C
1
and C
2
, are resp
ectively, 4, 1 and 4. Four
scena
rio
s
are
desig
ned for
prop
erly an
alyze the re
sult
s:
• Scenari
o
1: different fault rates a
nd customers prio
riti
zation
• Scenari
o
2: the same
rela
tive fault rate
• Scenari
o
3: rega
rdl
e
ss of cu
stome
r
prio
ritization
• Scena
rio 4:
relative fault
rate i
s
the
same
re
gardl
ess of the
custome
r
p
r
io
ritizatio
n
5.1. Recon
f
i
guration
w
i
t
hout th
e Dis
t
ributed G
e
n
e
ration
Units
Four sce
nari
o
s
applie
d o
n
propo
se
d
69 b
u
sse
s
n
e
twork
witho
u
t DG.
The
re
sults i
n
Table 1 a
r
e li
sted.
Table 1. Re
sults of the re
configuration without
DG
SAIFI
SAIDI
MAIFI
AENS
P
l
o
ss
OF
Scenario
40.23
103.99
9.76
48.96
119.99
4.38
1
40.15
104.26
9.82
49.53
120.55
4.40
2
40.19
104.19
9.68
49.42
120.52
4.38
3
40.55
104.82
9.84
49.35
120.95
4.41
4
Acco
rdi
ng to
the re
sults sh
own i
n
table
(1) In g
ene
ral,
the first a
nd f
ourth
scena
ri
os m
a
y
provide
the b
e
st an
d
worst
re
spo
n
se, re
spe
c
tively
. After the first
scenari
o
, the thi
r
d
scena
rio i
s
a
better resp
on
se. It also
ca
n be argue
d
that, second,
third, first an
d fourth
scen
ario
s have b
e
st
results f
r
om
point SAIFI index, re
spec
tively. In SAIDI,
res
p
ec
tively
firs
t, third,
s
e
c
o
nd and fourth
scena
rio
s
sh
ow a
bette
r respon
se. Th
e
sce
nari
o
s
3, 1, 2 a
nd
4 are b
e
st from MAIFI in
dex
viewpoi
nt. AENS and lo
sse
s
ca
n hav
e a simila
r situation with
SAIDI. Finally, the objective
function
is p
r
ioriti
zed
such a
s
o
ne, th
ree, tw
o a
n
d
four
scen
ari
o
s, respe
c
tively. Table
2
is
provide
d
the swit
ch code
s
in the
absen
ce of distribute
d
gene
rato
rs.
Table 2. Swit
ch code
s of reco
nfiguratio
n without DG
Sw
itch codes
Scenario
69 61 13 12
57
1
13 10 18 61
56
2
14 9 61 56 7
0
3
62 19 10 57
13
4
5.2. Recon
f
i
guration
w
i
t
h
DG
In this case, DG e
n
ters th
e re
config
ura
t
i
on pro
c
e
s
s. A DG is
appli
ed on n
e
two
r
k an
d its
effect on the reliability para
m
eters and th
e obje
c
tive
function
simulta
neou
sly with reco
nfiguratio
n
are stu
d
ied. T
able 3 list
s
the results of th
e study.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
2, No. 2, May 2016 : 241 –
247
246
Table 3. re
sul
t
s of reconfig
urat
ion in the
pre
s
en
ce of a
DG
SAIFI
SAIDI
MAIFI
AENS
P
l
o
ss
OF
Scenario
36.14
93.05
8.28
45.14
108.31
3.90
1
38.21
94.37
8.75
46.11
114.44
4.06
2
37.18
93.26
8.52
46.10
112.60
3.99
3
39.18
92.17
8.91
47.12
113.17
4.09
4
Acco
rdi
ng to
Table 3, it can be cl
aime
d that
the losses
can b
e
signifi
cantly redu
ced
comp
ared to before.
Ho
we
ver, still, first and f
ourth
scen
ario
s may
provide the
best an
d wo
rst
respon
se, re
spectively
b
u
t
differen
c
e
s
fo
urth scena
rio
s
a
nd
late
r scenari
o
(the seco
nd scena
rio)
decli
ned. It is clea
r th
at scenari
o
s 1, 3,
2 and
4,
respectively, ha
ve the be
st a
n
swer for SA
IFI
index. For S
A
IDI stran
ge
thing occu
rre
d
and
sc
ena
rio 4 ha
s the
best an
d sce
nario
2 ha
s the
worst an
swe
r
. MAIFI is
simila
r to the SAIFI.
A
bout AENS, priority is simi
lar to SAIFI
but
differen
c
e
se
con
d
an
d thi
r
d
scena
rio
s
are l
o
wer.
Scena
rio
s
first, third, fou
r
th and
se
co
nd,
r
e
spec
tively, dis
p
lays
the low
e
s
t
power
los
s
es
. T
he re
sult
s of
the five parameters of t
he
obje
c
tive function a
r
e
sho
w
n that the
succe
ssi
on
scenari
o
s fo
r th
e obje
c
tive function a
r
e
1, 3, 2
and 4. Lo
cati
on and
ca
pa
city of DG u
n
its a
s
we
ll
as
swit
ch co
des from the
applied
a DG is
s
h
ow
n
in
T
able
(
4
)
.
Table 4. switch cod
e
s a
nd
DG of re
co
nfiguratio
n in the pre
s
en
ce of
DG
place (capacity)
of DGs
sw
itch codes
Scenario
(400)2
0
69 13 12 61
52
1
(500)1
3
57 62 69 12
19
2
(450)1
1
55 13 18 61
10
3
(600)2
1
69 62 19 14
57
4
6. Conclu
sion
In this p
ape
r,
re
config
urati
on of di
stri
bu
t
ed net
works with p
r
e
s
e
n
ce of DGs to i
m
prove
the relia
bility and po
we
r
loss ha
s be
e
n
studi
ed. F
o
r this
pu
rpo
s
e, four i
ndi
ce
s of reli
abi
lity
indices ha
s
been
con
s
id
ered
in
obje
c
tive fun
c
tion
co
nsi
s
ts of:
System
average
inte
rru
p
t
ion
freque
ncy in
dex (SAIFI), System averag
e inte
rru
p
tion du
ratio
n
index (SA
I
DI), Momen
t
ary
averag
e inte
rruption f
r
eq
u
ency in
dex (MAIFI),
avera
ge en
ergy n
o
t
suppli
ed (A
ENS). It has
been
optimize
d
with PSO al
gorithm. Simulation h
a
s bee
n do
ne
on
6
9
bu
sse
s
n
e
t
work
with fo
ur
scena
rio
s
. T
he
simul
a
tio
n
s
re
sult
s h
a
ve sho
w
n
that relative f
ault rate a
n
d the
pri
o
rit
y
of
customers are effective on re
liability and relative costs.
Referen
ces
[1]
Gupta N, S
w
a
r
nkar A, Niazi
KR. Distributi
o
n net
w
o
rk rec
o
nfigur
ation for
po
w
e
r qua
lit
y
and re
lia
bil
i
t
y
improvem
ent u
s
ing Gen
e
tic Al
gorithms.
Elect
r
ical Pow
e
r an
d Energy Syste
m
s
. 201
4; 54: 664-
671.
[2]
Mend
oza J, L
ó
pez R
an
d Mor
a
les
D. Enri
qu
e
Ló
pez, P
h
il
ip
pe D
e
ssa
nte, a
nd R
o
g
e
r Mora
ga, Min
i
m
a
l
loss rec
onfi
gur
ation
usi
n
g
ge
n
e
tic a
l
gor
ithms
w
i
t
h
re
stricte
d
pop
ulati
o
n
an
d
ad
dresse
d op
erators:
re
a
l
app
licati
on.
IEEE Transactions on Power System
s
. 20
06; 21(2): 94
8-9
5
6
.
[3]
Din DR, Chiu YSh. A genetic
alg
o
rithm
for solvin
g vi
rtual top
o
lo
g
y
reconfig
urati
o
n pro
b
lem i
n
surviva
b
le W
D
M net
w
o
rks
w
i
th reco
nfig
ura
t
ion co
nstrai
nt.
Co
mp
uter C
o
mmu
n
icati
ons
.
200
8; 3
1
:
252
0-25
33.
[4]
Chio
u JP, Cha
ng ChF
a
nd Su
ChT
.
Variable
Scali
ng H
y
br
id
Differenti
a
l Evo
l
utio
n for Solvi
ng Net
w
o
r
k
Reco
nfigur
atio
n of Dis
trib
utio
n S
y
stems.
IE
EE T
r
ansactio
n
s on Pow
e
r
Systems
. 20
05
; 20(2): 66
8
-
674.
[5]
Jazeb
i
S a
nd
Vahi
di B. R
e
c
onfig
uratio
n of
di
strib
u
tio
n
n
e
t
w
o
r
ks to m
i
tigate
utiliti
e
s p
o
w
e
r
qu
ali
t
y
disturb
ances.
Electric Power System
s Res
e
arch
. 201
2; 91:
9-17.
[6]
Jazeb
i
S, Hos
s
eini
an S
H
an
d Vah
i
di B. D
S
T
A
T
C
OM allo
cation i
n
d
i
stributio
n net
w
o
rk
s consid
eri
n
g
reconfi
gurati
o
n
using d
i
fferen
t
ial evo
l
utio
n a
l
gorit
hm.
Ener
gy Conv
ersio
n
and Man
a
g
e
me
nt
. 201
1;
52(7): 27
77-
27
83.
[7]
Abde
laziz AY, Moham
ed F
M
,
Mekham
er SF
and Ba
dr
MAL. Distributio
n s
y
stem
reconfig
u
r
ation usi
n
g
a modifi
ed T
abu Search a
l
g
o
ri
thm.
Electric Power System
s
Research
. 20
1
0
; 80: 943-
953.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
Re
config
urati
on of Dist
ribu
tion Netw
orks with Presen
ce of DGs to
…
(Se
y
e
d
Me
hdi Maha
ei)
247
[8]
Z
hang D, F
u
Z
h
and Z
h
ang
L
.
An improved
T
S
al
gorithm f
o
r loss-min
i
mu
m reconfig
urati
on in l
a
rg
e-
scale distribution s
y
stems.
Electric Pow
e
r Systems Res
ear
ch
. 2007; 7
7
: 685-6
94.
[9]
Rese
nde B
a
rb
osa CHN, So
ares Men
des
MH
and Antô
nio d
e
Vasco
ncel
os J. Rob
u
st feede
r
reconfi
gurati
o
n
in radia
l
distri
butio
n net
w
o
rk
s.
Electrical Po
w
e
r and Energ
y
Systems
. 20
14; 54: 619-
630.
[10]
Ebrah
i
mi Mil
a
n
i
A, Hagh
ifam
MR. An evol
uti
onar
y a
ppr
oac
h for optima
l
ti
me interva
l
det
erminati
on i
n
distrib
u
tion
net
w
o
rk reco
nfigu
r
ation u
n
d
e
r varia
b
le l
o
a
d
.
Mathe
m
atic
al
and C
o
mp
uter
Model
lin
g
.
201
3; 57: 68-7
7
.
[11]
Su ChT
,
Chang ChF
and C
h
iou JP. Distrib
ution
n
e
t
w
ork
reconfi
gurati
o
n
for loss reduc
tion b
y
an
t
colo
n
y
s
earch
alg
o
rithm.
Elec
tric Power System
s Research
.
2005; 7
5
: 190-
199.
[12]
S
w
a
r
nk
ar A,
Gupta N and N
i
az
i KR. Ada
p
ted ant col
o
n
y
optimiz
ati
on for
efficient recon
f
igurati
–on o
f
bal
ance
d
and
un
bal
anc
ed
distrib
u
tion
s
ystems for l
o
s
s
minim
i
zatio
n
.
Sw
arm a
n
d
Evol
ution
a
r
y
Co
mp
utation
. 2
011; 1(3): 1
29-
137.
[13]
Carp
aneto E,
Chicco G. Dis
tributio
n s
y
ste
m
minimum l
o
ss reconfig
urat
ion i
n
the h
y
p
e
r-cub
e ant
colo
n
y
optimiz
ation
frame
w
o
r
k.
Electric Power System
s Research
. 20
08;
78(1
2
): 203
7-2
045.
[14]
Olamae
i J, Nik
nam T
,
Badal
i
S an
d Arefi F
.
Distributi
o
n
F
e
eder
Rec
onfig
u
r
ation f
o
r L
o
ss
Minimiz
a
tio
n
Based o
n
Mod
i
fied Ho
ne
y Be
e Mating Optim
i
zatio
n
Algor
ith
m
.
Energy Pro
c
edi
a
. 201
2; 14: 304-3
11.
[15]
Rese
nde B
a
rb
osa CHN, So
ares Men
des
MH
and Antô
nio d
e
Vasco
ncel
os J. Rob
u
st feede
r
reconfi
gurati
o
n
in radia
l
distri
butio
n net
w
o
rk
s.
Electrical Po
w
e
r and Energ
y
Systems
. 20
14; 54: 619-
630.
[16]
Niknam
T
,
Kavous
i F
a
rd
A an
d S
e
ifi
A. Distr
ib
ution
feed
er rec
o
n
f
igurati
on c
o
n
s
ideri
n
g
fue
l
cell/
w
i
nd/p
hoto
v
oltaic p
o
w
e
r p
l
ants.
Ren
e
w
able En
ergy
. 20
12; 37: 21
3-22
5.
[17]
Sathish K
u
mar
K and Ja
ya
b
a
rathi T
.
Po
w
e
r s
y
stem rec
onfig
uratio
n a
nd loss m
i
nim
i
zation for a
n
distrib
u
tion
s
ystems usin
g b
a
cteria
l fora
gin
g
optim
izati
o
n
alg
o
rithm.
El
ectrical P
o
w
e
r and
En
er
g
y
System
s
. 20
12
: 36: 13-17.
[18]
Kavous
i-F
a
rd
A and Nik
nam
T
.
Multi-objec
tive
stochastic
Distributi
on F
eed
er Rec
onfi
gurati
on fro
m
the reli
abi
lit
y
p
o
int of vie
w
.
En
ergy
. 201
4; 64:
342-3
54.
[19]
Gil Men
a
AJ
and M
a
rtin G
a
rcia JA. A
n
e
w
h
eur
istic
appr
oach
for
optima
l
rec
onf
igur
ation
i
n
distrib
u
tion s
y
s
t
ems.
Electric Pow
e
r Systems Researc
h
. 20
12; 83: 26
4-26
5.
[20]
Niknam T
,
Kavousi F
a
r
d
A an
d Bazi
ar A. Mu
lti-
obj
ective sto
c
hastic d
i
stribu
tion fee
der rec
onfig
urati
o
n
prob
lem cons
i
deri
ng h
y
dr
og
en an
d therm
a
l en
er
gy
p
r
od
u
c
ti
on
by
fu
el
ce
l
l
po
w
e
r pla
n
t
s.
Energy
.
201
2; 42: 563-
573.
[21]
Pfitscher L
L
, B
e
rnar
don
DP,
Can
ha
LN, M
o
ntagn
er VF
, G
a
rcia
VJ a
n
d
A
bai
de A
R
. Intel
lige
n
t s
y
stem
for automatic r
e
confi
gur
ati
on of
distrib
u
tion
net
w
o
rk in
re
al
time.
Electric
Power Systems Research
.
201
3; 97: 84-9
2
.
[22]
Z
i
dan A, S
haa
ban MF
, El-Sa
ada
n
y
EF
. Lon
g-term mu
lti-
ob
jective distrib
u
tion net
w
o
rk
pl
a
nni
ng b
y
D
G
alloc
a
tio
n
an
d feed
ers’ reco
nfi
gurati
on.
Electr
ic Pow
e
r Systems Res
earc
h
. 201
3; 105: 95-
104.
[23]
JS Rosseti G, J de Oliveira
E, W
de Oliv
e
i
ra L,
C Silv
a
Jr I and
Peres
W
.
Optimal al
locati
on
o
f
distrib
u
ted ge
n
e
ratio
n
w
i
t
h
re
config
urat
io
n i
n
el
ectric d
i
stri
butio
n s
y
stem
s.
Electric Power Systems
Research
. 20
1
3
; 103: 17
8-18
3.
[24]
José d
e
Olive
i
ra E, José R
o
sseti G, W
ille
r
de Olive
i
ra
L
,
Vanders
on
Gomes F, Peres W
.
Ne
w
alg
o
rithm for reconfi
gurati
on
and o
perati
n
g
proc
ed
ures i
n
electric distr
i
buti
on s
y
stem
s.
Electrical
Pow
e
r and En
ergy Syste
m
s
. 201
4; 57: 129-
134.
[25]
Niknam T
,
Azadfarsan
i
E and
Jabbar
i M. A
ne
w
h
y
b
r
id ev
oluti
onar
y a
l
g
o
r
ithm base
d
o
n
ne
w
fuzz
y
ada
ptive P
S
O
and
NM
alg
o
rit
h
ms for D
i
strib
u
tion
F
eed
er
Reco
nfigur
atio
n.
En
e
r
g
y
Co
nve
r
si
on
and
Mana
ge
me
nt
. 201
2; 54: 7-16.
[26]
Shari
a
tkha
h M
H
, Hag
h
ifam
MR, Sale
hi J
and M
o
ser A.
Durati
on b
a
se
d reco
nfig
urati
on of
electri
c
distrib
u
tion net
w
o
rks
usi
ng d
y
n
a
mic progr
a
mming and ha
rmon
y
se
arch alg
o
rithm.
Elec
trical Pow
e
r
and En
ergy Sy
stems
. 20
12; 4
1
: 1-10.
[27]
Hoos
hman
d R
and S
o
lta
n
i SH
.
Simultan
eo
us
optimiz
ation
o
f
phase
ba
lanc
ing
and
reco
nfi
gurati
on i
n
distrib
u
tion n
e
t
w
o
r
ks usi
ng B
F
-NM algorith
m
.
Electrical Pow
e
r and En
er
gy Systems
. 2
012; 41: 7
6
-86.
[28]
T
o
moiaga B,
M Chin
dris, Su
m
per A, Vill
afa
f
ila-Ro
b
l
e
s R a
nd Su
dr
ia-A
nd
reu A. Distrib
u
t
ion s
y
stem
reconfi
gurati
o
n
usin
g g
e
n
e
ti
c alg
o
rithm
b
a
sed
on c
o
n
n
e
cted
grap
hs.
Electric P
o
wer System
s
Research
. 20
1
3
; 104: 21
6-22
5.
[29]
Z
hang
P, L
i
W
and
W
a
n
g
Sh. Re
lia
bil
i
t
y
-
o
rie
n
ted distri
b
u
tion
n
e
t
w
ork reconfi
gurati
o
n
cons
ider
in
g
uncerta
inties
of data b
y
interv
al an
al
ysis.
Ele
c
trical Pow
e
r a
nd Ener
gy Systems
. 2
012; 3
4
: 138-1
44.
[30]
Milan
i
AE
and
Hag
h
ifam MR.
A ne
w
pro
b
a
b
ilisti
c
appr
oac
h
for distrib
u
tio
n
net
w
o
rk
reco
n
f
igurati
on:
Appl
icab
ilit
y to real n
e
t
w
orks.
Mathe
m
atic
al a
nd Co
mputer
Mode
lli
ng
. 20
1
3
; 57: 169-
179.
[3
1
]
Al
Ra
sh
id
i MR
, El
-Ha
w
a
r
y
ME.
A surve
y
of
p
a
rticle
s
w
arm
o
p
timizati
on A
p
plicati
ons
i
n
e
l
ectric p
o
w
e
r
s
y
stems.
IEEE Transactions on Evolutionary Com
p
utation
. 200
6; 13(4): 91
3-91
8.
[32]
Eberh
a
rt RC, Shi Y.
Particle sw
arm o
p
t
imi
z
at
io
n: De
vel
o
p
m
ents, Applic
atio
ns an
d resourc
e
s
.
Procee
din
g
s of
the 200
1 Con
g
resso
n Evolut
ion
a
r
y
Comp
utation. 20
01; 1: 81-8
6
.
Evaluation Warning : The document was created with Spire.PDF for Python.