TELKOM
NIKA
, Vol.11, No
.11, Novemb
er 201
3, pp. 6870
~6
878
e-ISSN: 2087
-278X
6870
Re
cei
v
ed Ma
y 13, 201
3; Revi
sed
Jul
y
2
4
, 2013; Acce
pted Augu
st 7, 2013
A Fault Area Location Method in Distribution Network
with DG
Zhongjian Kang*, Aina Ti
an, Zhe Bai
Dep
a
rtment of the Electrica
l
Engi
neer
in
g of Chin
a Univ
ersit
y
of petro
leum,
Qingd
ao, Ch
in
a, 0532
86
98
34
50
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: kangzjz
h@1
63.com*, tian
ai
na2
19
9@
ya
ho
o.cn, kingbz
2
@
sina.c
n
A
b
st
r
a
ct
T
o
precise
l
y lo
cate the gro
u
n
ded fa
ult are
a
i
n
distrib
u
tio
n
n
e
tw
ork w
i
th D
G, a new
grou
nde
d faul
t
area
loc
a
tion
a
l
gorit
hm b
a
sed
on thr
ee-
phas
e i
m
p
e
d
ance
mo
de
l a
nd fa
ul
t features i
n
d
i
stributio
n n
e
tw
ork
w
i
th DG is propos
ed i
n
thi
s
paper. T
h
is
meth
od esta
blish
e
s three-
phas
e i
m
pe
da
nce mod
e
l if th
e
distrib
u
tion
net
w
o
rk w
i
th DG, ana
ly
z
e
s
and
e
x
tracts the fau
l
t charact
e
ristics
on th
e b
a
sis
of
the thre
e p
has
e
impe
danc
e mo
del of distri
buti
on netw
o
rk w
i
th DG. Accordin
g to the charac
teristic of
the feature val
ue i
n
th
e
fault p
o
int, th
is
meth
od
travers
e
s a
ll th
e n
o
d
e
s
an
d g
e
ts the
fault ass
o
ciat
e
d
n
odes. T
h
e
meth
od
pro
pos
ed
in this paper was tested in
a
sixty nodal distribution syst
em with six
DG
’
s
,
the res
u
lt
of
the m
e
thod
verifies
that the met
h
o
d
can prec
isely
lo
cate the gr
o
und
ed fau
l
t are
a
in distri
butio
n
netw
o
rk w
i
th
DG.
Ke
y
w
ords
:
dis
t
ributio
n netw
o
rk w
i
th DG, fa
ult sectio
n lo
c
a
tion, thre
e ph
ase i
m
p
e
n
den
ce mode
l, en
e
r
g
y
character
i
stic
Copy
right
©
2013 Un
ive
r
sita
s Ah
mad
Dah
l
an
. All rig
h
t
s r
ese
rved
.
1. Introduc
tion
Distri
bution n
e
twork i
s
an
importa
nt part of
power
system an
d it is clo
s
ely a
s
sociate
d
with the
cu
st
omers. Th
e
secu
rity
of a p
o
we
r di
strib
u
tion net
wo
rk i
s
of g
r
e
a
t sig
n
ifican
ce to
the
long-te
rm st
able ope
rati
on of powe
r
system. Di
stributio
n net
work topol
og
ical structu
r
e
is
compl
e
x an
d
the e
n
viron
m
ent vari
es.
It is prone
to
failure.
The
r
efore, the
po
wer di
strib
u
tion
netwo
rk fault
diagn
osi
s
ha
s bee
n the focu
s study poi
nt [1-7]. Fault area lo
cation
is the basi
s
of
the fault locat
i
on.
For the
di
stri
buted n
e
two
r
k with
di
strib
u
ted ge
ne
rat
o
rs, th
e fault
are
a
lo
catio
n
ha
s
become
a i
n
d
i
spe
n
sable
st
ep in
the
fault
location
of
di
stributio
n net
work. Literatu
re
[8] propo
sed
an ada
ptive matrix distrib
u
tion network fault locati
on
algorithm o
n
the basi
s
of FTU fault partit
i
on
matrix algorit
hm. Acco
rdi
n
g to the dire
ction of
the cu
rre
nt into the FTU
, it determine
s wheth
e
r
distrib
u
ted p
o
w
er is i
nput a
nd whethe
r th
ere
are
othe
r
circum
stan
ce
s. And the
n
p
r
elimin
ary fau
l
t
area i
s
d
e
termined a
nd l
o
cate
d. Then
on the
b
a
si
s of topol
og
y and FT
U
over-cu
r
rent, the
adaptive faul
t matrix formation wa
s formed to co
nfirm the fault
area. Su
ch
method
s re
q
u
ire
some
me
asu
r
e u
n
it (F
TU). The inve
st
ment cost i
s
high. Lite
ratu
re [9] p
r
op
osed a
n
imp
r
ov
ed
matrix algorit
hm based o
n
the former fault ar
ea locali
zatio
n
algorithm. The
method add
he
orientatio
n to
the pa
ram
e
ters of n
e
two
r
k
matr
ix a
n
d
information
matrix. And
it set o
n
ly
one
positive
direction for the
multiple power
supply
ne
twork. Lite
rat
u
re [1
0] imp
r
oved th
e fa
ult
judgme
n
t mat
r
ix a
c
cordi
ng
to rel
a
tionshi
p bet
wee
n
ov
er-cu
r
rent di
rection
in fa
ult nod
es an
d t
he
load flo
w
direction. T
he f
ault inform
ation matr
ix co
mbined with the
net
work stru
cture
mat
r
ix
finishe
d
the improvem
ent. The method
search
ed th
e pare
n
t nod
e to determin
e
the fault node.
This
kind
of a
l
gorithm
ca
n l
o
catio
n
the
smallest
are
a
, however it i
s
greatly
influe
nce
d
by sy
stem
topology
stru
cture
so it i
s
lack
of
ro
bu
stness. Lite
ratu
re [1
1] p
r
opo
sed
an
a
r
ea
l
o
catio
n
m
e
th
od
based on
wa
velet transfo
rm. The meth
od applie
d wave
let transfo
rm to analysi
s
the sin
gula
r
ity
variation of th
e origi
nal si
g
nal. It looked
for the f
ault chara
c
te
risti
c
that can
be u
s
ed to locate the
fault area
so
as to determine the fau
l
t point.
This kind of intel
ligent algo
rithm dep
end
s on
experie
nce, the sel
e
ctio
n of
param
eters not eno
ugh
accuracy.
In orde
r to so
lve the probl
ems p
r
eviou
s
ly
mentioned,
three-pha
se
impeda
nce m
odel of
the distri
buti
on net
work
wa
s e
s
tabli
s
hed in thi
s
p
aper. T
he fa
ult feature of
the fault ph
ase
curre
n
t wa
s
extracted
wh
en fault occurred. Finally
a new fa
ult area l
o
cation method
was
prop
osed
by
usin
g the
faul
t cu
rre
nt
cha
r
acteri
stic.
Thi
s
m
e
thod
ha
s bee
n ve
rified
in a
sixty no
dal
dist
rib
u
t
i
on sy
st
em wit
h
six
DG’
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
e-ISSN:
2087
-278X
A Fault Area Location Met
hod in Di
strib
u
tion Net
w
ork with DG (Z
h
ongjia
n Kang
)
6871
2. The Impedance Mod
e
l
1) The m
a
in power supply
and DG thev
enin eq
uivale
nt model
From the th
evenin theo
rem, the mai
n
power
sup
p
ly and DG
equivalent im
peda
nce
model can re
spe
c
tively be
equivalent to
an ideal voltage source
with internal re
sista
n
ce. Each
pha
se cu
rren
t and voltage can be sim
u
ltaneou
sly m
easure
d
in the main feede
r and ea
ch DG
acce
ss
point
s (thi
s i
s
fea
s
ible
und
er t
he conditi
o
n
of existing te
chn
o
logy). T
hen the
po
sitive
seq
uen
ce, n
egative
sequ
ence a
nd
ze
ro sequ
enc
e curre
n
t
an
d voltage can be cal
c
ulated
.
By
adoptin
g the
po
wer eq
ui
valent mod
e
l
in literat
ure
[12], the p
o
sitive
seq
u
ence, ne
gati
v
e
seq
uen
ce
an
d zero
sequ
ence e
quival
ent impe
dan
ce
of the
DG can
be
se
t according
to the
theory of the asymmet
r
y electri
c
ity network.
0
12
12
0
12
0
,
F
LF
F
SS
S
F
LF
F
V
VV
V
ZZ
Z
II
I
I
g
gg
g
gg
g
g
,
(1)
Her
e
,
1
S
Z
,
2
S
Z
and
0
S
Z
is positive
seque
nce
impedan
ce
, negative seq
uen
ce
impeda
nce a
nd zero sequ
ence impe
da
nce
of DG re
spe
c
tively.
L
V
g
,
L
I
g
is the
positive
seq
uen
ce
voltage and
curre
n
t of the bran
ch th
at DG acce
sse
s
befo
r
e the
fault.
1
F
V
g
,
2
F
V
g
,
0
F
V
g
and
1
F
I
g
,
2
F
I
g
,
0
F
I
g
is the
po
siti
ve se
que
nce, neg
ative seq
uen
ce
and
zero
se
que
nce
voltage
and
curre
n
t
values when
fault occu
rre
d
. The
thre
e-pha
se
as
ym
metrical imp
e
dan
ce m
odel
of the
DG i
s
as
follows
:
(0
)
(1
)
22
22
(2)
0
11
11
0
1
1
1
10
0
1
3
10
0
1
Sa
S
a
c
S
Sa
b
S
abc
Sb
a
S
b
S
bc
Sc
a
S
c
Sc
b
S
ZZ
Z
Z
Z
ZZ
Z
a
a
Z
a
a
ZZ
Z
a
a
aa
Z
(2)
There,
2/
3
j
ae
.
Acco
rdi
ng
to Equation (1) and
(2),
the pos
itive sequ
ence,
neg
ative
se
que
nce and zero
seq
uen
ce eq
uivalent impe
dan
ce of DG
can be
cal
c
ulated by a set of voltage
and cu
rrent d
a
ta
variation
of t
he
DG
mea
s
urem
ent p
o
in
t before a
n
d
after the fault
.
And then the as
ymmetrical
impeda
nce m
odel of DG i
s
acq
u
ire
d
.
2) The th
ree
-
pha
se impe
d
ance model o
f
the double
windi
ng tran
sformer
The prim
ary’s current and
voltage (expressed in p or ABC) can b
e
asso
ciated
with the
se
con
dary’
s curre
n
t and
voltage (exp
ressed in
s o
r
ABC) a
c
co
rding to the
doubl
e win
d
i
n
g
transfo
rme
r
’
s
conn
ectio
n
mode [13]. So the thre
e
-
p
hase admitta
nce m
a
trix of the transfo
rmer
can b
e
set.
=
pp
ps
p
p
sp
ss
s
s
YY
IV
YY
IV
(3)
Takin
g
g
Y
tran
sform
e
r
as e
x
ample,
Y
is th
e sta
nda
rd v
a
lue of th
e t
r
an
sform
e
r
admittance.
11
10
0
0
33
11
01
0
0
33
11
00
1
0
33
11
2
1
1
0
33
3
33
11
1
2
1
0
33
3
33
11
1
1
2
0
33
3
33
A
A
B
B
C
C
a
a
b
b
c
c
I
V
I
V
I
V
y
I
V
I
V
I
V
(4)
Evaluation Warning : The document was created with Spire.PDF for Python.
e-ISSN: 2
087-278X
TELKOM
NIKA
Vol. 11, No
. 11, Novemb
er 201
3: 687
0 – 6878
6872
3) The th
ree
-
pha
se impe
d
ance model o
f
the feeder line
For th
e
distri
bution
network fee
d
e
r
[14],
the lin
e l
engt
h is not
very l
ong. T
he fe
e
der line
s
adopt
s three
-
pha
se
equiv
a
lent model i
n
this article.
The asymm
e
tric impe
da
nce mo
del of
unit length lin
e is as follo
ws.
=
aa
a
b
ac
L
abc
ba
bb
bc
ca
cb
c
c
Z
ZZ
Z
ZZ
Z
Z
ZZ
(
5
)
,
aa
b
b
Z
Z
and
cc
Z
are re
spe
c
tively the self
-imp
edan
ce of
the ABC phases,
,
,,,
a
b
ba
bc
c
b
c
a
Z
Z
ZZZ
and
ac
Z
are
re
sp
ectively the mutual imped
a
n
ce b
e
twe
en
ABC pha
se
s.
4 ) The three-pha
se impe
d
ance model o
f
the load
This pap
er o
n
ly studi
es th
e si
mple
co
n
s
t
ant im
peda
nce
loa
d
a
n
d
co
nsta
nt po
wer loa
d
.
The admittan
c
e mat
r
ixes o
f
the load in e
a
ch n
ode
are
establi
s
he
d to refle
c
t the u
nbala
n
ced lo
ad
of node [14]. So each p
h
a
s
e’s loa
d
admi
ttance matrix
can b
e
expre
s
sed a
s
:
-1
,
2
||
ii
LD
i
i
i
Pj
Q
Y
V
(6)
Her
e
,
=,
,
i
ia
b
c
,
i
P
is each p
h
a
s
e’
s
active po
we
r (Mw) ,
i
Q
is ea
ch ph
ase’s
reactive
power of ea
ch pha
se (MV
A
R),
i
V
is ea
ch
pha
se’
s
voltage (kV
)
of the
load nod
e.
5) The fo
rmat
ion of system’
s
three
-
p
h
a
s
e
admittance
matrix
Based o
n
the system top
o
logy stru
ctu
r
e, line para
m
eters and l
oad pa
ramet
e
rs, the
three
-
ph
ase unbal
an
ced
node ad
mittance eq
uation
fo
r a distrib
u
tion netwo
rk with DG ca
n
be
built as Equat
ion (7
).
1
1
12
1n
1
1
21
22
2
2
2
n1
n2
3
3
31
31
=
n
nn
n
n
nn
n
n
YY
Y
V
I
YY
Y
V
I
YY
Y
V
I
L
L
MM
M
M
M
M
L
(7)
Here, n is the node numb
e
r
of the distrib
u
tion
system,
and m is the distrib
u
ted ge
nerato
r
numbe
r of the distrib
u
tion
system. The
node
a
d
mittance mat
r
ix of the system i
s
:
11
12
1
n
21
2
2
2
33
n1
n
2
33
=
n
ab
c
n
n
nn
nn
YY
Y
YY
Y
Y
YY
Y
L
L
MM
M
M
L
(8)
The u
nbal
an
ced
thre
e-p
h
a
se
nod
e im
peda
nce eq
u
a
tion of the
p
o
we
r
system
is
sho
w
n
as follo
wing.
11
12
1n
1
1
21
22
2
2
2
n1
n2
3
3
31
31
=
n
nn
n
n
nn
n
n
ZZ
Z
I
V
ZZ
Z
I
V
ZZ
Z
I
V
L
L
MM
M
M
M
M
L
(9)
The nod
e imp
edan
ce mat
r
i
x
of the system is:
11
12
1
n
21
2
2
2
33
n1
n
2
33
=
n
abc
n
n
nn
nn
ZZ
Z
ZZ
Z
Z
ZZ
Z
L
L
MM
M
M
L
(10)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
e-ISSN:
2087
-278X
A Fault Area Location Met
hod in Di
strib
u
tion Net
w
ork with DG (Z
h
ongjia
n Kang
)
6873
The nod
e imp
edan
ce mat
r
i
x
is the invers
e matrix of the node a
d
mittance matrix.
=
ia
ii
b
ic
V
VV
V
and
=
ia
ii
b
ic
I
I
I
I
in F
o
rm
ul
a (7) an
d F
o
rmula
(9
) a
r
e
resp
ectively th
e voltage
a
n
d
the injectio
n current at noa
d
i
.
Thus, the th
ree-p
h
a
s
e im
peda
nce mo
del of t
he distribution n
e
twork
with DG
has b
een
establi
s
h
ed.
The th
ree
-
ph
ase
impe
dan
ce m
odel
takes th
e a
s
ym
metry of the
distrib
u
tion
n
e
twork
in consider.
3. The Fault
Area L
o
ca
tio
n
Method in
Distribu
tion
Sy
stem
w
i
th DG
3.1. The Fa
ult Fea
t
ure
Analy
s
is of the
Distribu
tion Ne
t
w
o
r
k
w
i
th DG b
ased o
n
Thr
ee
Phase Impe
dance Mo
del
Assu
ming th
at the DGs are
located
in no
de
s
(1)
,
(
2
)
,
,
(
)
B
SB
S
B
S
m
L
, t
he falut
occurre
d
in n
oad
j
and th
e
voltage an
d
the inje
ction
cu
rre
nt of e
a
ch
po
we
r supply can b
e
synchro
nou
sl
y measu
r
ed.
The mea
s
u
r
ed volt
age
and current sign
als befo
r
e the fault
are
(1)
,
(
2
)
(
)
BS
BS
BS
VV
V
m
gg
g
K
and
(1)
,
(
2
)
(
)
BS
BS
BS
I
II
m
gg
g
K
.
The measured
v
o
ltage and curre
n
t
sign
als
wh
en
the fault o
c
curred a
r
e
''
'
(
1
),
(
2
),
,
(
)
B
S
BS
BS
VV
V
m
gg
g
K
and
''
'
(
1
),
(
2
),
,
(
)
B
S
BS
BS
I
II
m
gg
g
K
.
The voltag
e o
f
the fault poi
nt before faul
t, noted a
s
oc
j
V
g
, c
a
n b
e
c
a
lc
u
l
a
t
e
d
b
y
th
e sys
te
m n
ode
fault equation
.
=1
=(
,
)
(
)
m
oc
jB
S
k
VZ
j
k
I
k
g
g
(11)
From Equ
a
tio
n
(11
)
, when
sho
r
t circuit current
f
j
I
g
is inje
cted into a nod
e, it will generate
fault compo
n
ent voltage, noted as
V
g
, at each no
de a
s
shown in (12
)
.
1
11
1
1
1
1
0
=
-
0
in
ii
i
i
n
i
f
i
nn
i
n
n
n
V
ZZ
Z
ZZ
Z
V
I
ZZ
Z
V
g
g
g
g
LL
M
M
ML
M
L
M
LL
ML
M
L
M
M
M
LL
(12)
Whe
n
sho
r
t-circuit
curre
n
t
Ifj is inject
ed into
node
j
, the fault compon
ent vo
ltage,
gene
rated by
Ifj, at each po
wer m
e
a
s
ure
m
ent point is:
=-
(
=
1
,
2
,
,
)
ii
j
f
j
VZ
I
i
m
gg
L
(13)
The fault voltage compo
n
e
n
t at measu
r
ement point
i
can b
e
obtain
ed by the differen
c
e
betwe
en voltage value b
e
fore fault and t
he value in fa
ult.
'
=(
)
(
)
(
=
1
,
2
,
,
)
iB
S
B
S
VV
i
V
i
i
m
g
L
(14)
Whe
n
the fau
l
t occu
rred in node
j
, The fault current ca
n be cal
c
ulat
ed as Equ
a
tion (15
)
by the fault voltage compo
nent in
so
urce measureme
n
t point, noted as
f
j
I
.
Evaluation Warning : The document was created with Spire.PDF for Python.
e-ISSN: 2
087-278X
TELKOM
NIKA
Vol. 11, No
. 11, Novemb
er 201
3: 687
0 – 6878
6874
'
()
=
(
,
)
(
V
(
)
-
V
()
)
(
=
1
,
2
,
,
)
f
j
BS
BS
Ii
Y
i
j
i
i
i
m
gg
g
K
,
(15)
The fault cu
rrent gene
rate
d by all powe
r
sup
p
lie
s is:
-1
=(
Z
(
,
)
+
)
oc
f
jg
j
I
jj
R
V
(16)
g
R
is fault grou
n
d
ing re
si
stan
ce.
The erro
r betwee
n
the fault
current of the fault phase
ca
lculated b
y
each po
we
r supply
and the fault current of the fault phase calculat
ed
by all power supplie
s is
defined a
s
fault
cha
r
a
c
teri
stic value in this pape
r.
The fault cha
r
acteri
stic valu
e, not
ed as E
(
j), is defin
ed
as follo
wing:
=1
()
=
|
(
)
-
|
m
f
jf
j
i
Ej
I
i
I
(17)
Acco
rdi
ng to the electri
c
al
netwo
rk the
o
ry, t
he
fault curre
nt of the fa
ult phase
cal
c
ulate
d
by each p
o
wer supply is
equal to the
fault curr
ent
of the fault phase cal
c
ul
ated by all po
wer
sup
p
lie
s fault curre
n
t. Thus, the fault characte
ri
stic val
ue at the true
fault point is zero.
3.2. Fault Section Loc
ati
on
If the ne
utral
point of
the
p
o
we
r
system
is di
re
ctly gro
unde
d, the f
a
ult cu
rrent i
s
large.
The fault cu
rrent cha
r
a
c
teri
stics ca
n be
e
x
tracted to lo
cate the fault area.
The distri
buti
on
n
e
two
r
k with
n
eutral
point
di
re
ctly ground
ed
is take
n a
s
the
obje
c
t of
study in
this pap
er. T
h
e
dist
ribution
netwo
rk
with DG
is mo
d
e
led
by u
s
in
g three
pha
se
impeda
nce m
odel. The fau
l
t feature of the dist
ri
butio
n netwo
rk wit
h
DG b
a
sed
on thre
e pha
se
impeda
nce m
odel is a
nalyzed and extra
c
ted with
the a
r
ithmetic
rep
r
ese
n
ted in 3.
1.
The fault associate
d
nod
e
is the node
with t
he sm
a
llest fault feature. Thu
s
, the fault
area i
s
locate
d by the fault
asso
ciated n
ode
s.
Assu
me the fault occurred
in
B(1)
, the fault current in fault node is cal
c
ul
ate by usin
g
Equation (15) and Equatio
n (16
)
. A
nd the fault feature value of
n
ode 1 is calculated by usi
ng
Equation
(17
)
. Then
assu
ming fault o
c
curred i
n
(
2
),
(3
),
,
(
)
B
BB
n
K
, the fault characteri
sti
c
value
(1
)
,
(
2
)
,
,
(
)
E
EE
n
K
is re
sp
ectively cal
c
u
l
ated.
The no
de fau
l
t characte
rist
ic value
whi
c
h is cl
osest t
o
the actu
al fault locatio
n
on is the
smalle
st. By confirming th
e two mi
nim
u
m of the
e
rror, the fault
area
is j
u
st
b
e
twee
n the t
w
o
node
s. In this way, the fault are
a
is
locate
d.
Ho
wever, in som
e
ca
se
s, this pro
c
e
ss
ma
y
misjud
ge the
fault area. In
orde
r to avoi
d this
p
r
oble
m
, three fault
asso
ciated n
ode
s are u
s
e
d
to
locate the fau
l
t area in this
pape
r. The flow chart
of th
e fault area lo
cation i
s
sh
o
w
n in Figu
re
1.
4. Simulation and Analy
s
is
4.1. The Simulation Mod
e
l
The fault a
r
e
a
location m
e
thod b
a
sed
on the th
ree
pha
se imp
e
n
den
ce mat
r
ix and the
fault feature
is te
sted
in
a sixty-n
ode
distri
b
u
tion system with si
x
distri
buted
gene
rato
r.
T
h
e
system’
s
ne
u
t
ral point i
s
g
r
oun
ded
dire
ctly. The to
p
o
logy of the
system i
s
sh
own i
n
figure
2
.
Load
parame
t
er is
de
scrib
ed in the lite
r
ature [1
5].
System’s total l
oad i
s
2.5 M
VA. System’s line
Numb
ers an
d
DG unit ca
p
a
city and a
c
cess point ar
e
sho
w
n in lite
r
ature [15]. In this pape
r, th
e
three p
h
a
s
e
node ad
mittance m
a
trix establi
s
h
e
d
in this pap
er can capt
ure a
n
y possible
imbalan
ce
an
d the g
r
o
und
ing
capa
citan
c
e. An
d
cabl
es
and
overh
ead lin
es ca
n al
so u
s
e
the
matrix model.
The
voltage and current waveforms o
f
each DG substatio
n
po
wer su
pply and
the
terminal
are
reco
rde
d
. Th
e
wavefo
rm
was t
r
an
sferre
d to MAT
L
AB whe
n
fault
occurs and
was
transfo
rme
d
into synchro
n
ous p
h
a
s
or
b
y
full wave Fourie
r tran
sformation.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
e-ISSN:
2087
-278X
A Fault Area Location Met
hod in Di
strib
u
tion Net
w
ork with DG (Z
h
ongjia
n Kang
)
6875
()
33
abc
bu
s
n
n
Z
()
33
ab
c
bus
n
n
Y
Figure 1. The
Flow Ch
art o
f
the Fault Area Lo
cation b
a
se
d on the T
h
ree
-
p
h
a
s
e Impeda
nce
Model
Figure 2. The
Topology of the Sixty-node
Distri
b
u
tion
System with Six Distribute
d
Gene
rato
r
4.2. Simulation Example
s
based on T
h
ree Phas
e Impedanc
e M
odel
Whe
n
the ne
ural point
of
t
he system’
s
main
p
o
wer
source i
s
di
re
ct grou
nde
d, the
sho
r
t-
circuit
curre
n
t will be
large
whe
n
fault o
c
curs
. In this ca
se, the th
ree-p
h
a
s
e im
peda
nce mo
del
can
be
used
to locate
the
fault are
a
. Th
e sim
u
lati
on
example
s
in
different fault
con
d
ition
s
a
r
e
addresse
d as following.
Example 1. T
he on
e-p
h
a
s
e gro
und fa
ul
t happe
ns
on
line 1, the g
r
oundi
ng resi
stance is
10
0
. Acco
rdi
ng t
o
the d
e
finitio
n
of fault feat
ure va
l
ue, th
e fault featu
r
e value
s
of
al
l the no
de
s
are
sh
own in
Figu
re 3.
Th
e fault a
s
soci
ated n
ode
s
searche
d
a
r
e
1, 2 an
d 3.
T
he meth
od
can
locate the Fa
ult area a
c
curately.
Evaluation Warning : The document was created with Spire.PDF for Python.
e-ISSN: 2
087-278X
TELKOM
NIKA
Vol. 11, No
. 11, Novemb
er 201
3: 687
0 – 6878
6876
Example 2. The two-pha
se
groun
d fault happ
en
s on line 8, the gro
undin
g
re
sist
ance is
10
. Acco
rding t
o
the definitio
n of fault feature va
lu
e, the fault feature values
of a
ll the node
s
are sho
w
n in
Figure 4.
Figure 3. Fau
l
t Chara
c
te
ristic Value of each
Nod
e
wh
en Line 1 O
c
curs Singl
e-p
hase Gro
und
Fault
Figure 4. Fau
l
t Chara
c
te
ristic Value of each
Nod
e
wh
en Li
ne 8 Occu
rs Two
-
ph
ase G
r
oun
d
Fault
Figure 5. Fau
l
t Chara
c
te
ristic Value of each
Nod
e
wh
en line 39 O
c
curs Three
-
ph
ase
Grou
nd Fa
ult
The fault a
ssociate
d
no
de
s sea
r
ched
a
r
e 8, 9
an
d 1
0
. The m
e
tho
d
ca
n lo
cate t
he Fa
ult
area a
c
cu
rate
ly.
Example 3. The three
-
ph
a
s
e groun
d fault
happen
s o
n
line 39, the groun
ding re
sista
n
ce
is
50
. According
to the definition of fault feature va
lu
e, the fault featu
r
e value
s
of a
ll the node
s
are sho
w
n in
Figure 5.
The fault a
ssociate
d
no
de
s sea
r
ched
a
r
e 3
9
, 40 a
n
d
41. Th
e m
e
thod
can l
o
cate the
fault area a
c
curately.
Example 4. Different faul
t condition
s
were
set to all the lines.
The fault area wa
s
sea
r
ched
by t
he m
e
thod
propo
sed
in thi
s
p
ape
r.
Th
e
results are
sh
own
in
Table
1 (as spa
c
e i
s
limited, only parts of the lo
cating re
sults
were listed
)
.
Example 5. T
he nu
mbe
r
of
distri
buted
g
ener
ator re
du
ced, and sin
g
l
e
pha
se
grou
nd,
two
pha
se groun
d and three p
hase fault we
re set in
all bran
ch. The fa
ult area wa
s
sea
r
ched by the
method prop
ose
d
in this paper. The
se
ction deter
min
a
tion accu
ra
cy with different number of
DG
is sh
own in Table 2.
Thro
ugh
a large num
ber
of simulatio
n
e
x
amples
, it can be
see
n
that the propo
sed fa
ult
se
ction lo
cati
ng meth
od b
a
se
d on
thre
e pha
se
impe
dan
ce m
odel
for a di
stri
buti
on net
wo
rk
with
D
G
h
a
s
h
i
ghe
r
acc
u
r
a
c
y
. T
h
is
me
th
od w
a
s
ver
i
fie
d
efficient i
n
t
he n
eutral
di
rectly g
r
o
und
ed
dis
t
ribution network
with
DG.
0
10
20
30
40
50
60
0
1
2
3
4
5
6
7
8
x 1
0
-3
f
a
u
l
t
f
eat
u
r
e v
a
l
u
e o
f
ea
c
h
noa
d
n
oad
f
aul
t
f
e
a
t
ur
e
v
a
l
u
e
0
10
20
30
40
50
60
0
0.
005
0.
01
0.
015
f
a
u
l
t
f
e
at
ur
e v
a
l
u
e of
e
a
c
h no
a
d
no
ad
f
aul
t
f
e
a
t
ur
e v
a
l
u
e
0
10
20
30
40
50
60
0
0.
0
0
5
0.
01
0.
0
1
5
0.
02
0.
0
2
5
0.
03
0.
0
3
5
0.
04
0.
0
4
5
0.
05
f
aul
t
f
eat
u
r
e v
a
l
u
e
of
eac
h
no
ad
noa
d
f
a
ul
t
f
eat
ur
e
v
a
l
u
e
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
e-ISSN:
2087
-278X
A Fault Area Location Met
hod in Di
strib
u
tion Net
w
ork with DG (Z
h
ongjia
n Kang
)
6877
Table 1. The
Re
sult of Fau
l
t Area Location
Fault condition
Fault line
section
Associated nodes
Whether accurat
e
AG 2
2-3
2
、、
34
y
e
s
BG 16
16-17
16
、、
17
18
yes
CG 24
24-25
24
、、
25
26
yes
AG 31
27-31
30
、、
29
31
yes
BG 48
48-49
48
、、
49
51
yes
ABG 6
2-7
7
、、
23
y
e
s
BCG 12
12-13
12
、、
13
14
yes
ACG 17
17-18
18-17-
19
yes
ACG 21
20-22
22
、、
20
21
yes
BCG 26
26-27
26
、、
27
32
yes
ABCG 32
32-33
32
、、
33
34
yes
ABCG 40
39-41
39
、、
41
31
yes
ABCG 42
42-43
42
、、
43
44
yes
ABCG 50
49-51
49
、、
51
53
yes
ABCG 52
51-53
49
、、
51
53
yes
Table 2. Secti
on Dete
rmin
a
t
ion Accu
ra
cy
with Different
Numbe
r
of DG
DG num
ber
The accurac
y
u
n
der different
num
ber of D
G
1 93.33%
2 95%
3 98.33%
4 100%
5 100%
6 100%
5. Conclusio
n
Fault
lo
cat
i
on
in di
st
rib
u
t
i
o
n
sy
st
em
s
wi
t
h
larg
e a
m
o
unts
of DG
c
an g
r
eatly aff
e
ct the
operation of the powe
r
di
stributio
n system. The me
thod pro
p
o
s
e
d
in this pap
er is efficie
n
tly
solved the fa
ult area lo
cati
on problem.
The metho
d
use
s
the me
asu
r
ing te
ch
n
i
que that can
be
fulfilled at pre
s
ent. It ha
s a
solid th
eo
reti
cal fou
ndatio
n and
it is ve
rified in a
60 n
ode
s di
strib
u
tion
system with six
gen
erato
r
s.
The simul
a
tion
re
sult
s
sho
w
that th
e pro
p
o
s
ed f
ault are
a
lo
cation
method ba
se
d on three p
h
a
se imp
eda
n
c
e mod
e
l in t
he distri
butio
n netwo
rk
with DG is effe
ctive.
Even when t
he numb
e
r o
f
the DG is
small, the m
e
thod is al
so
accurate. In the simulati
on
example, the
accuracy of the fault are
a
locati
o
n
is ab
ove 93.33%. The mo
re DG
s are empl
oyed,
the method is more ro
bu
st. When
DG
s n
u
mbe
r
is
ab
o
v
e 4, the accura
cy of the method is e
q
ual
to 100%. In addition, the m
e
thod p
r
op
osed in this p
a
per i
s
not affected by the
variation
s
of the
system top
o
logy, system
oper
ational
states, fault time
and the gro
undin
g
re
sist
ance.
Ackn
o
w
l
e
dg
ment
This pa
pe
r is sp
on
sored
by the Chinese Nation
al Nature Scien
c
e Fu
nd
Project
(612
710
01).
The auth
o
rs
are g
r
ateful f
o
r all the
revi
ewe
r
s fo
r val
uable
su
gge
stions to imp
r
ove
the quality of this pap
er.
Referen
ces
[1]
Rodri
go
Hartstein S
a
lim, Mar
i
an
a Res
e
n
e
r, André D
a
rós
F
ilomen
a
, et a
l
.
Exte
nd
ed F
ault-L
o
catio
n
F
o
rmulati
on for
Po
w
e
r Distri
b
u
t
ion S
y
stems.
IEEE Transactions On Power Deliv
ery.
200
9; 24(2).
[2]
XU
Ha
o, MIAO Shih
on
g, JIAN
G Z
hen, Z
E
NG fei, Z
H
ANGlei,
LIUPei. A
Ne
w
F
a
ult L
o
cati
on Al
gorit
h
m
Based
on
F
a
u
l
t Comp
on
ent
from F
i
nite S
y
nchro
n
ize
d
Ph
asor Me
asur
e
m
ent Un
it.
Au
tomati
on of
Electric Power System
s
. 20
12
; 36: 1-6.
[3]
W
A
NG Lin-ch
u
an, LI Qin
g
-
x
in
, LIU Xin-
qua
n, Z
H
ANG W
e
i, PAN W
en-mi
n
g
. Distrib
utio
n
net
w
o
rk fa
ul
t
locati
on b
a
se
d
on the im
prov
ed a
n
t colo
n
y
alg
o
rithm.
Pow
e
r System Pr
o
t
ection a
nd C
o
ntrol.
20
08;
36(2
2
): 29-3
3
.
[4]
SU Li
ng-mi
n, PENG Min-fan
g
, W
A
NG Yao
-
nan. D
i
st
ributi
on n
e
t
w
ork fa
u
l
t locati
on
bas
ed o
n
d
a
ta o
f
sensor F
T
U.
Tr
ansducer and
Microsystem
Technologies
. 2012; 31(
11): 70
-73.
[5]
Z
hong
jia
n K
a
n
g
, Ain
a
T
i
an, Y
a
n
y
a
n
F
e
ng. F
ault
Detectio
n
i
n
C
o
mpl
e
x D
i
s
t
ributio
n N
e
t
w
o
r
k Base
d o
n
Hilb
ert-Hu
ang T
r
ansform.
TE
LKOMNIKA.
2013; 11(
3): 883
-890.
Evaluation Warning : The document was created with Spire.PDF for Python.
e-ISSN: 2
087-278X
TELKOM
NIKA
Vol. 11, No
. 11, Novemb
er 201
3: 687
0 – 6878
6878
[6]
GUO Z
huang-z
h
i, W
U
Jie-ka
n
g
.
Electro
m
a
g
n
e
tism-
like M
e
c
han
is
m Base
d
F
ault Sectio
n
Diag
nos
is for
Distributi
on N
e
tw
ork
. Proceedings of the CS
EE. 2010; 30(
1
3
): 34-40.
[7]
Hon
g
she
ng S
u
, Yunch
u
a
n
Z
hang. D
i
strib
u
tion Gri
d
F
a
ult Loc
atio
n A
ppl
yi
ng T
r
ansi
entZ
e
ro-mo
d
e
Current.
TELK
OMNIKA.
2012
; 10(5): 140
3-1
413.
[8]
W
u
Le-Pe
ng, Hua
ng C
hun,
Qi Yong, Z
h
a
o
W
e
i, Ji
ang H
o
ng-T
ao. A ne
w
ada
ptive matri
x
a
l
gor
ithm for
fault locati
on in
distributio
n ne
t
w
ork
w
i
t
h
distr
i
bute
d
gen
erati
on. Internatio
n
a
l Conf
erenc
e on Electric
a
l
and C
ontrol E
n
gin
eeri
ng. 20
1
1
: 499-5
04.
[9]
Z
hong C
hao,
Z
eng
Xia
ngj
un
, Deng F
eng,
Li Lin
g
, Ye Ji
amin
g.
A fault area loc
a
tion
algorit
hm i
n
distrib
u
tion n
e
t
w
ork w
i
th DG.
PROCEEDINGS OF
CUS-EPSA. 2008: 25
8
3
-25
85.
[10]
LUO Xi
ao, W
A
NG Jin-feng,
LI Xia
o
-
w
e
i
, Z
H
AO Jin-bao.
Identificatio
n of F
aul
t Area i
n
Distributi
o
n
Netw
ork Based
on Orientativ
e Layer
ing Mo
de
l
. Proceed
in
gs of the CSU-EP
SA. 2005; 17(
6
)
: 84-86.
[11]
Xu
zih
ua W
a
ng
ya
nso
ng.
Stud
y of fa
ult
regi
on
loc
a
ti
ng met
hod
fo
r distrib
u
tio
n
net
w
o
rks
w
i
th
d
i
stri
bu
te
d
ge
ne
ra
to
rs.
Pow
e
r System Protect
i
on a
n
d
C
ontrol
. 2011; 39(
24): 22-2
6
.
[12]
Sukumar
M. B
r
ahma. F
a
ult
Locati
o
n
in
Po
w
e
r D
i
stri
bu
ti
on
Sy
ste
m
Wi
th Pe
ne
tra
t
i
o
n
of D
i
s
tri
b
u
t
ed
Generati
on. IEEE
T
r
ansactio
n
s On Pow
e
r Delivery.
2
011; 0
885-
897
7: 1-9.
[13] George J
W
a
ki
leh. Po
w
e
r
S
y
s
t
ems Harmo
nic
s
F
undam
enta
l
s.
Analysis
an
d F
ilter D
e
sig
n
. 2001:
14
3-
175
[14]
Elham
B Makr
am, Adl
y
A Gi
rgis. A Ge
ner
alize
d
Comp
uter T
e
chniq
ue f
o
r the
Dev
e
lo
pment
of t
h
e
T
h
ree-Phase Imped
ance Mat
r
ix for Un
ba
lan
c
ed Po
w
e
r S
y
stems.
Electric Pow
e
r Systems Res
earc
h
.
198
8; 15: 41-5
0
[15]
Sukumar M
B
r
ahma, A
d
l
y
A
Girgis. D
e
vel
opm
e
n
t of Ad
aptive
Protecti
on Sc
heme
fo
r Distrib
utio
n
S
y
stems With
High P
e
netration
of Distribut
ed Generation. IEEE
T
r
ansactions On Po
w
e
r Delivery
.
200
4; 19(1): 56
-63
Evaluation Warning : The document was created with Spire.PDF for Python.