TELKOM
NIKA Indonesia
n
Journal of
Electrical En
gineering
Vol. 15, No. 1, July 2015, p
p
. 1 ~ 13
DOI: 10.115
9
1
/telkomni
ka.
v
15i1.808
3
1
Re
cei
v
ed Ma
rch 2, 2
015;
Re
vised
Ma
y 14, 2015; Accepted Ma
y 30
, 2015
A Novel Method
Based on Biogeography-Based
Optimization for DG Planning in Distribution System
Mohammad
Sedagha
t*, Esmaeel Ro
kr
ok, Mohammad Bak
h
ship
our
Dep
a
rtement o
f
Electrical Eng
i
ne
erin
g, Lores
tan Univ
ersit
y
,
Dan
e
shg
ah Str
eet, 712
34-9
8
6
53, Khorram
a
b
ad, Loresta
n, Iran
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: mohamma
ds
eda
gh
at74@
g
m
ail.com
A
b
st
r
a
ct
T
h
is pa
per
pr
opos
ed
a n
o
v
e
l tech
ni
que
base
d
o
n
b
i
o
geo
grap
hy-b
as
ed o
p
ti
mi
z
a
t
i
o
n
(BBO)
alg
o
rith
m in or
der to opti
m
al
plac
e
m
ent an
d
si
z
i
n
g
of
d
i
stin
ct types
of Dist
r
ibute
d
Ge
nera
t
ion (
D
G) u
n
its
i
n
the distri
butio
n
netw
o
rks w
h
ich is ap
pl
i
ed to
improve v
o
lta
ge pr
ofile
as
the
mai
n
factor
for pow
er qu
a
lity
improve
m
ent
a
nd r
educ
e
po
w
e
r losses. In
ord
e
r to
pro
m
ote the
i
n
vesti
gatio
n to
be
c
apa
ble
i
n
pr
ac
tical
terms, the
lo
ad
s are
lin
ear
ly v
a
rie
d
i
n
s
m
a
ll s
t
eps of
1% fro
m
50%
to
150
% of th
e b
a
se
valu
e. T
he
opti
m
a
l
si
z
e
a
nd
loc
a
ti
on of
distinct t
y
pes
of
DGs a
r
e foun
d
out i
n
eac
h l
o
a
d
st
ep. T
h
is w
i
l
l
ai
d the
distrib
u
ti
o
n
netw
o
rk oper
ators (DNOs) to have
a lo
ng te
rm sch
edu
li
ng
for the opti
m
a
l
ma
nag
e
m
e
n
t of DG units an
d
achi
eve the
maxi
mu
m p
e
rfor
ma
nce. T
o
veri
fy the efficienc
y of pr
opos
ed
meth
od, it has
bee
n con
ducte
d to
IEEE 33-bus radi
al distri
butio
n system. Also
, simul
a
tion
res
u
lts are co
mpa
r
ed w
i
th the analytical a
ppr
oa
c
h
and HPSO a
l
gorith
m
(
m
ix
e
d
bin
a
ry an
d typical p
a
rt
icle
sw
arm opti
m
i
z
a
t
i
on a
l
g
o
rith
m). T
he obta
i
ne
d
simulati
on res
u
lts demonstrat
e the better per
form
a
n
ce a
nd
effectiveness
o
f
the propos
ed
meth
od.
Ke
y
w
ord:
distrib
u
ted ge
nerati
on,
l
o
n
g
-
t
er
m sc
hed
ul
i
ng, l
oad
vari
a
t
ions, vo
ltage
profil
e, p
o
w
e
r loss,
bio
geo
gra
phy-
base
d
opti
m
i
z
a
t
ion (BBO)
Copy
right
©
2015 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
The definition of the di
stri
bu
ted generati
on i
s
a
gen
eration of power by
facilities that
are
adeq
uately smaller th
an
central
gen
era
t
ing plant
s a
nd can be a
d
joine
d
at
n
e
arly
any poin
t
in
power
syste
m
[1, 2]. Due to the con
s
i
dera
b
le p
r
og
ression in
sev
e
ral g
ene
rati
on tech
nolo
g
i
e
s,
power
syste
m
s de
reg
u
lati
on, enviro
n
m
ental effect
s and fab
r
icatio
n issue
s
of n
e
w tra
n
smission
lines, th
e p
e
netration
lev
e
l of
DG
s in
po
wer net
work have
be
en d
e
velopin
g
du
ring
the
last
decade [3, 4]. In addition DG may result in variou
s ad
vantage
s su
ch as control o
f
voltage profile,
ancill
ary se
rv
ice
s
, improvi
ng in po
wer quality and
reliability ch
a
r
acte
ri
stics, loss de
cre
m
e
n
t,
energy savin
g
s an
d dist
ri
bution
capa
ci
ty deferral [5
-11]. Lat
ely, nume
r
ou
s pa
pers have b
e
e
n
pre
s
ente
d
to
study the p
r
oblem
s of op
timal allo
cation an
d si
zin
g
in vario
u
s
con
d
ition. Using
analytical
me
thod, the
po
wer lo
ss mi
nimizatio
n
of
syste
m
wa
s preform
ed
by suita
b
le
DG
allocation [12
]. An approach base
d
on m
u
lti-obje
c
tive
index whi
c
h was utilize
d
to redu
ce voltag
e
drop
a
nd
po
wer l
o
ss
wa
s
sugge
sted
in [
13]. In
o
r
de
r t
o
optimi
z
e
co
rre
ctive a
c
tio
n
s, pl
anni
ng
and
operation of
distrib
u
tion n
e
twor
k, an al
gorithm b
a
se
d on multi-ob
jective GA was recomme
n
ded
in [14, 15].
From th
e met
hodol
ogy poi
nt of view, se
veral alg
o
rith
ms h
a
ve be
e
n
utilize
d
for
suitabl
e
DG
allo
cation
su
ch
a
s
im
p
r
oved
PSO t
e
ch
niqu
e [16]
, hybrid
GA
a
nd
simulate
d
anne
aling
[17
],
combi
ned
G
A
and PSO [
18], tabu
sea
r
ch
[19],
non
-linear and
dy
namic
prog
ra
mming [20,
2
1
],
differential
evolution
algo
rithm [22], a
r
tificial
bee
colo
ny algo
rithm
(ABC) [23]
h
a
rmo
n
y search
algorith
m
[24]. This stu
d
y pro
p
o
s
e
s
a
novel
a
pproach ba
se
d
on BBO
al
gorithm
whi
c
h is
investigate
d
to asce
rtain t
he optim
al DG allo
ca
tion
and
sizi
ng to
improve volta
ge p
r
ofile a
s
the
main fa
ctor
for p
o
we
r
q
uality improvement a
n
d
redu
cing
po
wer lo
sse
s
of
the di
stri
but
ion
netwo
rk.
Also
, from
50% to
150%, th
e n
e
twork load
i
s
cha
nge
d to
ma
ke th
e inv
e
stigatio
n mo
re
pra
c
tical BB
O ha
s the
a
d
vantage
s of
both we
ll known alg
o
rit
h
ms
GA an
d PSO. Sha
r
ing
informatio
n b
e
twee
n solutions i
s
on
e of
the GA's
fe
a
t
ures. In PS
O. from ea
ch
iteration to t
h
e
next, solution
s a
r
e
save
d
but ea
ch
sav
ed
solution
is ca
pable
to l
earn
from
its neigh
bo
rs
a
nd
simultan
eou
sl
y with the p
r
og
re
ssi
on of
the algo
rith
m, adopt itself [25]. so
contai
ning th
ese
feature
s
sim
u
ltaneou
sly, ca
use
s
the sup
e
rio
r
perfo
rm
ance of BBO algorith
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 15, No. 1, July 201
5 : 1 – 13
2
In this p
ape
r
with a
pen
alty function
wh
ich e
n
tails two pen
alties with flexible im
pact
s
, in
each loa
d
lev
e
l four
sp
ecifi
c
bu
se
s a
r
e
sele
ct
ed
as t
he candi
date
s
. A pen
alty for g
a
ining
m
o
re
loss red
u
ctio
n and the oth
e
r one fo
r obt
aining b
e
tter voltage profil
e, have been
con
s
id
ere
d
. With
this strat
egy without the p
enetratio
n
of volt
age prof
ile as an ind
epen
dent obj
ective in mai
n
obje
c
tive fun
c
tion, the
ap
prop
riate volt
age p
r
ofile i
s
a
c
cessibl
e
. This te
ch
ni
que h
e
lp
s th
e
algorith
m
s to
perf
o
rm
mo
re effective
se
arch a
nd i
n
e
a
ch
iteratio
n
find the
be
st
buses to in
st
al
DG and also
the
conve
r
g
ence spee
d of
the
alg
o
rit
h
ms
wo
uld b
e
incre
a
sed.
But this met
hod
need
s
som
e
algorith
m
s which
search
for the
solutio
n
s i
n
bi
nary
manne
r,
so
with som
e
h
e
u
r
isti
c
approa
che
s
li
ke PSO, this
techni
que
co
uld not b
e
im
plemente
d
. T
herefo
r
e i
n
th
is inve
stigatio
n
the PSO technique
whi
c
h
compared wit
h
BBO appr
oach i
s
the combination of
PSO and BPSO
(bina
r
y PSO algorithm
) named h
e
re
after as HPSO (hybrid
PSO). As
mentione
d b
e
fore,
becau
se of
h
a
ving the fe
a
t
ures of PSO
and
GA,
BBO is capabl
e
to search
in
bina
ry way and
doe
s not nee
d to be modified like PSO a
nd this is o
n
e
of the main advantage
s of this app
roa
c
h
.
To clarify
th
e
efficien
cy of
the
pre
s
e
n
ted a
p
p
r
oa
ch,
the
re
sults
are
compa
r
e
d
with
an
alytical
approa
ch an
d
HPSO algo
ri
thm. All the simulation
s are
carried o
u
t in MATLAB software.T
he rest
of the pape
r
is organi
ze
d as follo
ws:
section 2
high
lights DG types an
d probl
em formul
ation.
Section
3 re
pre
s
ent
s the
pro
p
o
s
ed B
B
O algo
rithm
for o
p
timal
DG
sitting a
nd si
zin
g
. T
he
simulation results
are illustrat
ed and di
scussed in
section
4 an
d finally
concluding remarks
are
dra
w
n in secti
on 5.
2. Problem
Formulation
2.1. T
y
pes of a DG
Based o
n
DG units term
inal cha
r
a
c
te
ristics in terms of active
and rea
c
tive powe
r
delivering capability, those can
be
categorized into three ma
jor types as follows [26]:
1)
Type 1: This
type o
f
DG
has capability of inje
cting
only
P, such as fuel cells, photovol
taic
systems and
micro turbin
es. This type of DG
unit is maximized their MWh benefit, From
DNO
s point of view. However, it may
cause
reducti
on in voltage support with respect to
distribution system charac
t
e
ristics in pro
v
iding the
needed reactive power [27].
2
)
Type 2: This type of DG has ca
pability of in
jecting
both P and Q. This grou
p of DG units
includes
synchronou
s machine and VSC based
DG
units. For instance, adjusting the power
angle and modulation index in VSI-based PV array
can be result
ed in controlling the output
active and reactive power independently [28].
3)
Type 3: This type of DG have capabili
ty of
injectin
g P but usu
a
lly absorbin
g
Q, such a
s
induction generators utilized in wind far
m
s.
2.2. Po
w
e
r Fl
o
w
M
e
th
od
Due to seve
ral advantage
s of the forwa
r
d/ba
ckwa
rd
sweep techni
que su
ch a
s
. Needin
g
low mem
o
ry,
high comput
ational
pe
rformance, simpl
e
stru
ctur
e, high conve
r
g
ence ca
pabili
ty,
and ap
plicabil
i
ty to utilization in unb
alan
ced syste
m
s, t
h
is po
we
r flo
w
metho
d
ha
s bee
n sele
cted
in this
s
t
udy [29-31].
2.3. Objectiv
e Functio
n
In this study,
the objective
function is described for real
power losse
s minimization:
mi
n
L
O
b
jectiv
e
F
u
n
ct
io
n
P
(1)
Whi
c
h the ex
act real p
o
we
r losse
s
are o
b
tained by th
e followin
g
eq
uation:
11
[(
)
(
)
]
bb
NN
L
i
j
ij
i
j
i
j
ij
j
i
P
aP
P
Q
Q
b
Q
P
Q
P
(2)
Whe
r
e,
co
s(
)
ij
ij
i
j
ij
R
a
VV
And
si
n
(
)
ij
ij
i
j
ij
X
b
VV
i
j
ij
ij
Z
Rj
X
are the
comp
onent
s of impedan
ce mat
r
i
x
and N
b
is the numbe
r of buses [32].
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
A Nov
e
l Method
Base
d on
Biogeog
rap
h
y
-Ba
s
e
d
Opti
m
i
zation for DG…
(Moh
a
m
m
ad Sedag
hat)
3
2.4. Cons
trai
ns
The ope
ratin
g
rest
rictio
ns
are de
scri
bed
as follows:
1) The Limitation of Vo
ltage
mi
n
m
a
x
i
VV
V
(3)
Where
mi
n
V
and
ma
x
V
indicate the minimum and maximum permissible vo
ltage (±5%)
and
i
V
is the
voltage at bus i.
2) Power bala
n
ce con
s
train
t
s
/
11
g
DG
N
N
g
wD
G
d
d
L
gd
Pg
Pg
P
P
(4)
Where N
g
an
d N
DG
are the whole numb
e
r of traditional generation
unit and whol
e number
of DGs, Pg
gw
/D
G
is the amo
unt of active
power of
traditional power generation un
it g with
introducing of
DG, P
gd
is the amount of
active power
of DG unit d,
P
d
is the who
l
e load
demand and P
L
is
the whole loss of active power.
3) Active and
reactive powe
r
constraints [33]:
22
2
,
g
ig
i
g
i
m
a
x
PQ
S
(5)
Where
Q
gi
an
d S
gi,m
a
x
repre
s
ents the am
ouns of reacti
ve
and appa
rent power of
the ith
DG.
3. Biogeog
rap
h
y
Theor
y
Biogeog
rap
h
y
Based
Opt
i
mization
(B
BO) meth
od
whi
c
h i
s
ba
sed
on bi
og
eography
theory, ha
s
b
een
pro
p
o
s
e
d
in
200
8 by
Dan
Simon
[3
4]. The
pro
c
e
dure
of BBO
i
s
a
n
exam
ple
of
natural process that can be ut
ilized to solve general problem
s
of
optimization.
In BBO, each
individual is
assume
d as an island (or a habi
tat), and the feature
s
su
bscription thorou
gh
individual
s a
r
e depi
cted
a
s
emig
ratio
n
and immi
grat
ion (Fi
gure 1
)
. Each
sol
u
tion prope
rty is
named
a su
itability index variable
(SIV). Geog
ra
p
h
ical
regio
n
s that are a
ppro
p
ri
ated
as
resi
dences f
o
r biol
ogi
cal
types
are said to have a high habita
t suitability index (HSI). T
he
meanin
g
of a
high
HSI of a
habitats i
s
proper pe
rform
ance on
the o
p
timization
problem
wh
ere
a
s
a low HSI sh
ows imprope
r perform
an
ce
on the opt
imization p
r
obl
e
m
. Heuri
s
tic
algorith
m
s
so
lve
the optimization problem
u
s
ing Inten
s
ification
the po
p
u
lation. In BBO gen
eratin
g next gene
rati
on
perfo
rmed
by
immig
r
ating
solutio
n
p
r
op
erties to
th
e
other isl
and
s,
and
giving
solution
prope
rties
by emigration
from the oth
e
r isl
and
s. Th
en muta
tion i
s
do
ne for
all
the pop
ulatio
n. This m
u
tation
pro
c
ed
ure is
simila
r to GA algorith
m
's m
u
tation.
Figure 1. Emmigratio
n of speci
e
s a
nd n
e
w isl
and
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 15, No. 1, July 201
5 : 1 – 13
4
In BBO, each individual
has its o
w
n i
mmigratio
n rate, depicte
d
by
λ
, and emigratio
n
rate, d
epi
cte
d
by
μ
. A
proper
solution
ha
s
highe
r
μ
; Therefor, i
t
has a
very
high
pro
babili
ty of
borro
wing p
r
o
pertie
s
from o
t
her sol
u
tion
s, helpi
ng it to improve for th
e next genera
t
ion illustrate
d
in Figure 2.
Figure 2. Specie
s model of
a single h
abi
tat
The fa
ct that in BBO, emi
g
ration
doe
s
not
expre
s
s t
hat the emig
rating isl
and l
o
se
s a
prop
erty
sho
u
ld b
e
con
s
id
ered. Emi
g
ration a
nd i
mmi
gration
can
b
e
mathe
m
atically investigat
ed
by a prob
abil
i
stic mo
del. In addition a
s
sume th
at, consi
der the
p
r
oba
bility P
s
that the habit
a
t
inclu
d
e
s
exactly S species
at
t
. varies
from time
t
to tim
e
tt
as
follows
:
11
1
1
1
ss
s
s
ss
s
s
P
tt
P
t
t
t
P
tP
t
(6)
If
0
t
, from Equa
tion (6) it ca
n be written a
s
follows:
11
11
1
1
m
a
x
11
m
a
x
()
,
0
()
,
1
1
()
,
ss
ss
s
ss
s
s
s
s
s
ss
s
s
s
PP
S
PP
P
P
S
S
PP
S
S
(
7
)
Figure 1 illust
rates these
relationships,
as
straight l
i
nes
but, gener
ally, they
might be m
o
re
compli
cate
d grap
hs. Th
e amount
s of emigratio
n
and
immigration
rates are obtai
ned a
s
:
k
E
k
n
(
8
)
1
k
k
I
n
(
9
)
Whe
r
e the m
a
ximum po
ssible immigration rate is
I; the maximum
possibl
e emi
g
ration
rate is E;
K is the number of kind
s
of the k-th individual
and
n is the num
ber of kind
s. No
w, assume
the
c
e
rtain c
a
s
e
E=
I (Figure 3). In this
c
a
s
e
:
kk
E
(
1
0
)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
A Nov
e
l Method
Base
d on
Biogeog
rap
h
y
-Ba
s
e
d
Opti
m
i
zation for DG…
(Moh
a
m
m
ad Sedag
hat)
5
Figure 3. Illustration of two
can
d
idate
sol
u
tions to som
e
probl
em
3.1. Biogeog
raph
y
-
Based
Optimizatio
n
Assu
me that
there i
s
a
probl
em a
n
d
a pop
ulatio
n of ca
ndid
a
t
e solutio
n
s
that are
ascertai
ned
a
s
vecto
r
s. In
addition
su
p
pose that th
ere
are
som
e
way
s
of
determi
ning
the
efficiency of the solutions. Proper
sol
u
tions
are
si
milar to i
s
lands
with a high island suitability
index (ISI), and imprope
r solution
s are
similar to isla
n
d
s with a lo
w
ISI.
Figure 4. The
migration o
p
e
rato
r in BBO
Figure 5. The
mutation ope
rator in BBO
Con
s
id
er that
ISI is like “fitness” in oth
e
r optimi
z
atio
n algorith
m
s
whi
c
h are ba
sed o
n
popul
ation. B
B
O spe
c
ially
wo
rks ba
se
d on
the
two st
ru
cture
s
,
migration
a
nd m
u
tation
as
sho
w
e
d
in Figure
s
4, 5.
3.1.1. Migration
With prob
abi
lity
P
mod
which is called
habitat modification proba
bility,
each solutio
n
can be
corrected ba
se
d
on other solution
s. If
a given sol
u
tion S
i
i
s
c
h
o
s
e
n
t
o
b
e
corre
c
ted, then its immigration rate
is perfo
rmed
to probabili
st
ically decid
e
whether or
not to corre
ct each suita
b
ility index variabl
e (S
IV) in that solution. After choosi
ng the SIV
for c
o
rrec
tion, the rates
of emigration
of other
solutions are
utilized to
choose whi
c
h
solutio
n
s thro
ugh th
e
pop
ulation
group
will
mi
grat
e
rand
omly
selecte
d
SIVs to th
e
ch
ose
n
s
o
lution S
i
.
3.1.2. Mutati
on
In BBO, utilizing the spe
c
i
e
s
count p
r
o
babilitie
s, the mutation rat
e
s a
r
e d
e
termined. As
rema
rked in Equation (7
),
the probabili
ties of
each
spe
c
ie
s co
un
t can be evaluated usi
ng the
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 15, No. 1, July 201
5 : 1 – 13
6
differential
eq
uation. Ea
ch
membe
r
of p
opulatio
n ha
s a rel
a
ted
pro
bability, whi
c
h dete
r
mine
s
the
prob
ability that it exists a
s
a sol
u
tion fo
r a giv
en p
r
obl
em. If the like
lihood
of a ce
rtain
solution
is
very low then that s
o
lution s
i
milar to mutate to
so
m
e
othe
r soluti
on. Likewi
se
if the som
e
o
t
her
solution probability is
greater then
that
solution
set has very
sm
all
chance to m
u
tate. Mutati
on
rate of
each
set of
solution can be
computed
i
n
terms of ki
nds
coun
t probability utilizing t
h
e
expre
ssi
on:
1
s
ma
x
ma
x
P
mS
m
P
(
1
1
)
Whe
r
e m
max
is a user defin
ed paramete
r
.
3.2. Propose
d
Method S
t
eps
This stu
d
y propo
sed a ne
w app
roa
c
h b
a
se
d on BBO algorithm wh
ich is inve
stig
ated to
determi
ne th
e optimal l
o
cation an
d ca
pacity of diffe
rent type
s DGs
whi
c
h i
s
applie
d to im
prove
voltage p
r
ofil
e as the m
a
in
factor for
po
wer qu
a
lity improvem
ent
and
red
u
ce p
o
we
r lo
sse
s
of the
distrib
u
tion
n
e
twork. Al
so i
n
this inve
stig
ation
fro
m
5
0
%
to 150%, t
he
system
lo
ad i
s
cha
nge
d to
make
the i
n
vestigatio
n mo
re p
r
a
c
tical.
With d
e
fining
two
pen
alty functio
n
s rel
a
ted to volta
g
e
profile an
d p
o
we
r loss red
u
ction, searching proc
edu
re of pro
p
o
s
e
d
algorith
m
h
a
s be
cam
e
m
o
re
fast and effective. The prop
ose
d
algo
rith
m step
s are p
e
rform
ed a
s
follow:
Step 1:
Ente
r the
loa
d
d
a
ta of th
e n
e
twork a
nd
run
p
o
we
r flo
w
fo
r
each
step
s
of load.
Chang
e
the load
s of the network a
s
follows:
,,
0.
5
,
1
,
,
in
e
w
i
n
e
w
i
i
L
LL
L
ai
N
PQ
a
P
j
Q
(12)
Whe
r
e
a
is the load co
efficie
n
t, which vari
es bet
wee
n
0
.
5 and 1.5.
Step 2:
Initiali
ze
a
sam
p
le
popul
ation
an
d DG
parame
t
ers an
d d
e
fine p
enalty fu
nction
s i
n
o
r
d
e
r
to obtaining t
he be
st voltage profile a
n
d
more lo
ss
re
ductio
n
, simul
t
aneou
sly.
Step 3: Dete
ct four be
st
buses
fo
r
DG
installatio
n
consi
deri
ng p
e
nalty function
s, in e
a
ch lo
ad
step.
Step 4:
Initialize the BBO
para
m
eters i
n
clu
d
ing m
a
ximum spe
c
ie
s count, max
i
mum mig
r
ati
on
rates, a
nd ma
ximum mutation rate an
d a
nelitism pa
ra
meter.
Step 5: Initialize ha
bitats
depe
nding u
pon ha
bitat si
ze withi
n
feasible regio
n
. Set the iteration
cou
n
t
e
r m =
0.
Step 6: Add
the
cou
n
ter
b
y
1. Ch
eck
whether it is le
ss than
t
he
maximum ite
r
ation limit. If not,
print the outp
u
t result
s.
Step 7: If not
, cal
c
ulate
th
e HSI val
ue f
o
r th
e given
μ
&
λ
an
d S
e
lect th
e o
p
timum
HSI val
u
e
based on eliti
s
m pa
ramete
rs.
Step 8: Modify each no
n-el
ite habitat usi
ng immigratio
n & emigratio
n
rate.
Step 9: Check for conceivability.
If yes,
HSI is com
p
uted.
Step 10: Specie
s co
unt probability is
up
dated an
d re
calcul
ated the
HSI.
Step 11:
Go t
o
ste
p
6
for the n
e
xt iterati
on.
Thi
s
p
r
o
c
edure
can
be
finish
ed
after a
con
c
eiva
bl
e
probl
em solut
i
on ha
s bee
n found.
The ab
ove
mentione
d m
e
thod
shoul
d
be re
peate
d
for all loa
d
ing level
s
(1% load
variation
s
).
The follo
win
g
BBO pa
ra
meters
have
been
used,
popul
ation
size=20, Ha
bitat
Modification Probability=1,
Immi
gration Probability
bound
s per
gene= [0,
1], elitism parameter
=
4, step
si
ze
for numeri
cal
integr
ation
of probabiliti
e
s=1, maxim
u
m
λ
an
d
μ
rates
for eac
h
islan
d
=1 and
Mutation Pro
bability=0.0
5
4.
Simulation Results a
nd Discussion
In order to inves
t
igate th
e performance of the propos
ed approac
h
, the IEEE 33-bus
radial di
stri
bution test syst
em is utilized i
n
this paper.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
A Nov
e
l Method
Base
d on
Biogeog
rap
h
y
-Ba
s
e
d
Opti
m
i
zation for DG…
(Moh
a
m
m
ad Sedag
hat)
7
Figure 6. Single line diag
ram of
33-b
u
s
dist
rib
u
t
i
on t
e
st
sy
st
em
Figure 6 sho
w
s th
e si
ngle
line diag
ram
of t
he test system. The t
o
tal amou
nts of the
active and
re
active load
s
of the system
are 3.
7
15 M
W
an
d 2.3 M
VAr, respe
c
tively. In addition,
as m
ention
e
d
in [3
5], the
initial am
ou
nt of t
he
acti
ve and
re
acti
ve po
wer lo
sse
s
b
e
fore
DG
allocation a
r
e
210.8
4
kW
a
nd 1
43.114
kVAr, re
spe
c
ti
vely. As ment
ioned
befo
r
e,
there
a
r
e th
ree
types of DG
s. In this investigatio
n the first tw
o typ
e
s are discu
s
sed. In the first ca
se stu
d
y,
without in
stall
a
tion any type of DG
unit
s
, the system load
s are va
ri
ed linea
rly fro
m
50% to 15
0
%
of base
ca
se
with 1% ste
p
s. In the se
con
d
ca
se
an
d third
ca
se the DG type 1 and type 2
are
investigate
d
resp
ectively.
4.1. Withou
t using DG
The results o
f
simulation t
e
st for va
riation in lo
sses
and minim
u
m
value of voltage a
r
e
obtaine
d for three di
stin
ct con
d
ition
s
: base lo
ad values, in
cre
a
se
d by 50% and decrea
s
e
d
by
50% are me
n
t
ioned in Tabl
e 1.
Table 1. Re
sults of Variati
ons (Witho
ut Usi
ng DGs)
IEEE 33
Decrease 50
%
Base case
Increase 50%
P
los
s
(kW)
48.7566
210.84
519.3936
Q
lo
ss
(kVAr)
33.0471
143.114
353.1554
Vmin(pu)@b
us 0.9540@18
0.9039@18
0.8483@18
The
50 %
increa
se
in l
oad
values ha
s l
e
d to
wo
rst vol
t
age p
r
ofile.
The mi
nimum
voltage
in this
co
nditi
on is expe
rie
n
ce
d at b
u
s 1
8
whi
c
h
is
eq
ual to 0.8
483.
On the
othe
r
hand, afte
r 5
0
%
enha
ncement
in lo
ad
value
s
, the
voltage
profile i
s
in
creased
and
th
e minim
u
m v
o
ltage l
e
vel i
s
at
bus 1
8
with the value of 0
.
9540. As sh
own in
Fig
u
re 7, load increase ca
uses
a negative ef
fect
on the voltag
e profile. On t
he othe
r h
a
n
d
, becau
se
of
increme
n
t in
the load, the
voltage profile is
enha
nced. O
n
ce th
e loa
d
i
s
de
crea
sed,
a re
du
cti
on in
the slo
pe of t
he lo
ss
cu
rve
coul
d be
se
e
n
,
as well.
Figure 7. illustrates the volt
age prof
ile un
der di
stinct lo
ad co
ndition
s
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 15, No. 1, July 201
5 : 1 – 13
8
For
example,
wh
en the
lo
ad i
s
en
han
ced a
bout
5
0
%
of its b
a
se
value, the
a
c
tive an
d
rea
c
tive po
wer lo
sses
are
redu
ce
d by 1
45.36–
145.
8
6
%
, resp
ective
ly. Neverthel
ess, as th
e lo
ad
is de
cre
a
sed
by 50% of its ba
se value
,
the active and rea
c
tive p
o
we
r losse
s
are redu
ce
d by
76.67–
79.71
%, resp
ective
ly. Figure
8 il
lustrate
s th
e
amount
s of t
he a
c
tive an
d re
active p
o
w
er
losse
s
und
er
distin
ct loadin
g
con
d
ition
s
.
Figure 8. Loss variation
s
u
nder diffe
rent
loading level
s
(without inst
alling DG)
4.2. Installati
on of t
y
pe1-DG
In this
ca
se, t
he o
p
timal pl
acem
ent a
nd
size of th
e
sin
g
le
DG
unit,
whi
c
h i
s
sche
duled
to
provide
only
active po
we
r (P), a
r
e eva
l
uated. To
m
a
ke it
com
p
a
r
able
wit
h
t
h
e re
sult
s of
l
a
st
sub
s
e
c
tion, the feede
r loa
d
s are ch
ang
ed in the sam
e
way.
After variou
s simul
a
tion
s i
n
diverse
co
ndition
s in
clu
d
ing lo
ad
ch
angin
g
, som
e
nota
b
le
points h
a
ve b
een carried o
u
t which are
as follo
w:
1) Fou
r
buse
s
are
cho
s
en
to instal the DG. This se
lection is b
a
sed on having
prope
r
voltage profil
e and mo
re redu
ction in p
o
we
r losse
s
simultaneo
usly
. The four ch
oice
s a
s
be
st bus
can
d
idate
s
are 6, 7, 26, and 27.
2)
With
re
ga
rd to
voltage
profile a
nd
volt
age
stabil
i
ty indice
s, t
he b
e
st
bu
s for
DG
installation is 7 , while with considering on power
loss redu
ction, the prop
er bu
s
to instal the DG
is 6. But
in th
is p
ape
r the
focu
s
on l
o
ss
redu
ction
is
more
than
vol
t
age p
r
ofile
so finally the
b
e
st
bus to in
stal
DG
is 6. Th
e num
eri
c
al
resu
lts
whi
c
h
prove
d
the
above m
enti
oned
point
s
are
sho
w
n in T
a
ble 2. This p
o
int also fo
r installatio
n
of type-2 DG i
s
true b
u
t in orde
r to avoid
repetition, in this inve
stigati
on only the re
sults
of type-1 DG pla
c
em
ent in the two
bus candi
dat
es
are exp
r
e
s
se
d and compa
r
ed.
Table 2. Co
m
pari
s
on of DG Installation
With the Sam
e
Size on Bu
s 6, 7 Und
e
r
Load Va
riatio
n
Res
u
lted By BBO Algorithm
IEEE 33
P
los
s
(k
W
)
(bus 6)
P
los
s
(k
W
)
(bus 7)
Q
lo
ss
(kVAr)
(bus 6)
Q
lo
ss
(kVAr)
( bus 7)
Vmin(pu)@b
us
(bus 6)
Vmin(pu)@b
us
(bus 7)
Load Decrease b
y
50
%
26.4559
26.9239
19.459
20.7015
0.9719@18
0.9739@18
Base Case
Load Increase
by
5
0
%
110.834
261.187
111.90
264.04
81.693
192.573
84.6439
199.50
0.9425@18
0.9122@18
0.9448@18
0.9170@18
Figure 9 de
monst
r
ate
s
the optimal
si
ze of
the DG unit asse
ssed by
HPSO method,
Analytical a
p
p
roa
c
h
[26]
a
nd p
r
o
posed
BBO app
ro
ach.
As
sh
own i
n
Fig
u
re
9, th
e optim
al
size
of
the DG unit varie
s
linea
rly by the chan
gi
ng in the feed
er load.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
A Nov
e
l Method
Base
d on
Biogeog
rap
h
y
-Ba
s
e
d
Opti
m
i
zation for DG…
(Moh
a
m
m
ad Sedag
hat)
9
Figure 9. Optimal size of type1-DG u
n
it unde
r differe
nt loading lev
e
ls
The loa
d
flow analysi
s
de
monst
r
ate
s
that
the percen
t
age of loss redu
ction in th
e BBO-
based ap
pro
a
c
h is
slightly greate
r
than t
hat
of the PSO method an
d analytical a
ppro
a
ch.
In Table 3, the re
sults of th
e pro
p
o
s
ed a
ppro
a
ch for t
h
ree
state
s
of loads a
r
e giv
en and
also
co
mpa
r
ed with
the
obtaine
d results of
H
PSO algo
rithm
and An
alytical app
ro
ach i
n
the
same con
d
it
ion.
Table 3. Co
m
pari
s
on
Re
sul
t
s of the Load
Chan
ging in
Presen
ce of Type1-DG, E
v
aluated by
HPSO Algorit
hm, Analytica
l
Approa
ch
a
nd Prop
osed
BBO Approa
ch
IEEE 33
Decrease 50
%
Base case
Increase 50%
HPSO algorithm
P
los
s
(kW)=26.45
61
Q
lo
ss
(kVAr)= 19.
487
Vmin(pu)@b
us=0.9718@18
DG Size= 1331 k
W
P
los
s
(kW)=111.0
30
Q
lo
ss
(kVAr)= 81.
911
Vmin(pu)@b
us=0.9424@18
DG Size=2712 k
W
P
los
s
(kW)=262.3
15
Q
lo
ss
(kVAr)= 192
.921
Vmin(pu)@b
us=0.9121@18
DG Size=4016 k
W
Anal
y
t
ical approch
P
los
s
(kW)=27.63
2
Q
lo
ss
(kVAr)= 20.
332
Vmin(pu)@b
us=0.9712@18
P
los
s
(kW)=111.9
21
Q
lo
ss
(kVAr)= 82.
321
Vmin(pu)@b
us=0.9719@18
P
los
s
(kW)=268.2
14
Q
lo
ss
(kVAr)= 196
.018
Vmin(pu)@b
us=0.9703@18
Proposed BBO
a
pproch
DG Size= 1235 k
W
P
los
s
(kW)=26.45
59
Q
lo
ss
(kVAr)= 19.
459
Vmin(pu)@b
us=0.9719@18
DG Size= 1272k
W
DG Size= 2501 k
W
P
los
s
(kW)=110.8
34
Q
lo
ss
(kVAr)= 81.
693
Vmin(pu)@b
us=0.9425@18
DG Size= 2598 k
W
DG Size= 3785 k
W
P
los
s
(kW)=261.1
87
Q
lo
ss
(kVAr)= 192
.573
Vmin(pu)@b
us=0.912@18
DG Size= 4012 k
W
Figure 1
0
d
e
m
onst
r
ate
s
t
he voltag
e p
r
ofile un
der di
fferent loa
d
in
g level
s
. Accordin
g to
the re
sults
of Figure 7 an
d
Figure
10, it
can
be n
o
ted
that appli
c
ati
on of DG in t
he sy
stem ha
s
amend
ed the
voltage profil
e effectively.
In the loa
d
g
r
owth
ca
se, th
e minimu
m voltage m
agnit
ude h
a
s o
c
cu
rre
d at b
u
s 1
8
, whi
c
h
is 0.9
122
pu.
For 50% l
o
a
d
in
cre
a
se.
On the
ot
he
r
side,
as the l
oad i
s
red
u
ced, the mi
ni
mum
voltage magn
itude is 0.97
1
9
pu. at bus 1
8
for a 50% d
e
crea
se.
Figure 10. Voltage profile u
nder diffe
rent
loading
level
s
after in
stallation of type1-DG (by BBO
approa
ch
)
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 15, No. 1, July 201
5 : 1 – 13
10
Figure 11. Lo
ss va
riation
s
unde
r differe
nt l
oading lev
e
ls after in
sta
llation of type1-DG
Figure 1
1
illu
strate
s th
e a
c
tive and
re
act
i
ve po
we
r lo
sse
s
u
nde
r
different
conditio
n
s
after
establi
s
hm
en
t of type1-DG
and utilizin
g the biog
eog
ra
phy based op
timization (B
BO) algo
rith
m.
Acco
rdi
ng to
the re
sults
of Figure 8 an
d
Figur
e 11, it can
be note
d
that the act
i
ve and
rea
c
tive po
wer lo
sse
s
a
r
e
detra
cted f
o
r
all loa
d
level
s
after i
n
stallat
i
on of type
1-DG. T
he val
u
es
of active and
rea
c
tive po
wer lo
sses
are
redu
ce
d
by 4
9
.25– 44.8
3
%
, resp
ective
ly. In the case of
50% redu
ctio
n in th
e lo
ad,
the a
c
tive an
d reacti
ve
po
wer lo
sses a
r
e de
crea
sed
by 45.5–
40.8
2
%,
r
e
spec
tively.
4.3. Installati
on of t
y
pe2-DG
In this case, the DG
unit
can
pro
d
u
c
e both P an
d Q. The
re
sults
of three
different
loadin
g
condi
tion in the
prese
n
ce of typ
e2-DG
unit a
nd resulted b
y
BBO app
ro
ach
are given
in
Table 4 and
also com
p
a
r
ed with the
obtained
re
sults of HP
SO algorith
m
and Anal
ytical
approa
ch in the sam
e
co
n
d
ition.
Table 4. Co
m
pari
s
on
Re
sul
t
s of the Load
Chan
ging in
Presen
ce of Type2-DG, E
v
aluated by
HPSO Algorit
hm, Analytica
l
Approa
ch
a
nd Prop
osed
BBO Approa
ch
IEEE 33
Decrease 50
%
Base case
Increase 50%
HPSO algorithm
P
los
s
(kW)=16.44
38
Q
lo
ss
(kVAr)=13.0
303
Vmin(pu)@b
us=0.9719@18
DG Size= 1496
KVA
P
los
s
(kW)=65.93
82
Q
lo
ss
(kVAr)= 53.
2140
Vmin(pu)@b
us=0.9588@18
DG Size= 3137
KVA
P
los
s
(kW)=156.2
214
Q
lo
ss
(kVAr)= 126
.1621
Vmin(pu)@b
us=0.9345@18
DG Size= 4778
KVA
Anal
y
t
ical approch
P
los
s
(kW)=16.21
23
Q
lo
ss
(kVAr)= 12.
6303
Vmin(pu)@b
us=0.9719@18
P
los
s
(kW)=66.33
21
Q
lo
ss
(kVAr)= 53.
8721
Vmin(pu)@b
us=0.9573@18
P
los
s
(kW)=157.1
235
Q
lo
ss
(kVAr)= 126
.754
Vmin(pu)@b
us=0.9341@18
Proposed BBO
a
pproch
DG Size= 1482
KVA
P
los
s
(kW)=16.54
33
Q
lo
ss
(kVAr)= 13.
3303
Vmin(pu)@b
us=0.9785@18
DG Size= 1334
KVA
DG Size= 3040
KVA
P
los
s
(kW)=67.94
48
Q
lo
ss
(kVAr)= 54.
8304
Vmin(pu)@b
us=0.9568@18
DG Size= 2925
KVA
DG Size= 4599
KVA
P
los
s
(kW)=157.4
616
Q
lo
ss
(kVAr)= 127
.1075
Vmin(pu)@b
us=0.9340@18
DG Size= 4587
KVA
Figure 12. Op
timal size of t
y
pe2-DG unit
under diffe
re
nt loading lev
e
ls
Evaluation Warning : The document was created with Spire.PDF for Python.