TELKOM
NIKA Indonesia
n
Journal of
Electrical En
gineering
Vol.12, No.5, May 2014, pp
. 3745 ~ 37
5
3
DOI: http://dx.doi.org/10.11591/telkomni
ka.v12i5.5109
3745
Re
cei
v
ed
No
vem
ber 1
1
, 2013; Re
vi
sed
De
cem
ber 2
5
,
2013; Accep
t
ed Jan
uary 8
,
2014
Comprehensive Evaluation of Reliability of Protection
System in Smart Substation
Jipu Gao
1
, Xu He
2
, Peichao Zhang
2
*, Chan
gbao
X
u
1
1
Guizhou R
e
se
arch Institute o
f
Electric Po
w
e
r Exp
e
rime
nt, Gui
y
an
g 55
000
2, Chin
a
2
Dept. Electric
al Eng
i
ne
eri
ng,
Shang
ha
i Jiao
T
ong Univ
ersit
y
, Sha
ngh
ai 2
0
024
0, Chi
n
a
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: pczhan
g@sjt
u
.edu.cn
A
b
st
r
a
ct
The reli
ab
ility
of smart su
bstations
has a
g
r
eat
sign
ifica
n
c
e
on the s
a
fe
t
y
and stab
ility
of smart
grid
operation.
Taking the pr
otecti
on system
in s
m
art substation as
an
example, this
paper c
onstructs
compre
hens
ive
reli
abi
lity mo
d
e
ls
to eval
uate
the
rel
i
a
b
il
it
y of smart substati
ons w
i
th d
i
ffere
nt architect
u
re
s.
The paper first
illustrat
e
s tw
o i
m
portant
aspects w
h
ich affec
t
the rel
i
abilit
y
of the pr
otection system
, namel
y
the
netw
o
rk
ar
chitecture an
d the ma
i
n
ten
a
n
c
e strategy. T
o
satisfy these t
w
o aspects, th
e pa
per th
en
a
dopt
the Monte C
a
rlo si
mul
a
tio
n
combi
ned w
i
th the Rel
i
ab
ility
Block Di
agra
m
meth
od to make qu
antitati
v
e
reliability analysis.
At last, reliability of f
our power transform
e
r
protec
tion system
s
applying differ
ent
ma
inte
nanc
e s
t
rategies
w
i
th
altern
ative
arc
h
itectures
are
eval
uate
d
. T
h
e
si
mul
a
tio
n
res
u
lts sh
ow
clea
rly
that adva
n
ced
mai
n
ten
anc
e
strategies su
ch as con
d
iti
ona
l maint
ena
nce w
ill pl
ay
a critical ro
le
in
enh
anci
ng the
relia
bi
lity and
a
v
aila
bi
lity of smart substation.
Ke
y
w
ords
:
smart substation,
protectio
n
, reli
abil
i
ty, Mont
e-
Carlo si
mul
a
tio
n
, conditi
on-
ba
sed mai
n
ten
a
n
c
e
Copy
right
©
2014 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
Chin
a ha
s b
e
com
e
the
country to put
the larg
est
numbe
r of smart sub
s
tations into
operation. Ne
w tech
nolo
g
i
e
s such a
s
Gigabit
Ethernet comm
uni
cation, syn
c
h
r
oni
zed
sam
p
ling
with micro
s
e
c
ond a
c
curacy
, nonconventi
onal tra
n
sdu
c
ers
are wid
e
l
y
used in
sm
art su
bstatio
n
s
.
These
new technologies bring tr
emendous changes
to
smart
subst
a
tion, whereas the reliabilit
y
issue ha
s al
so arou
se
d wi
despre
ad con
c
ern at the sa
me time.
Smart substation reliability should be analyzed fr
om t
w
o aspects. The
first aspect is the
netwo
rk a
r
chi
t
ecture of the
system. Because of
the LAN-b
a
sed fea
t
ure of the smart su
bstati
ons,
a variety of archite
c
tu
re
s h
a
ve been
de
sign
ed to me
et deifferent requireme
nts i
n
pra
c
tice. The
key differe
nce in the alternative
archite
c
ture
s i
s
that whethe
r t
he netwo
rk archi
t
ecture dep
e
nds
on Ethernet
swit
che
s
. Existing
relia
bility analyse
s
are
all focused on
a p
a
rticular net
wo
rk
architectu
re [
1
-3], therefore there i
s
a
lack
of ho
ri
zontal
com
p
arison of diff
erent n
e
two
r
k
architectu
re
s.
The
se
cond
asp
e
ct i
s
the
maintena
nce stra
tegy of
the syste
m
. A sma
r
t su
bstatio
n
system is a
repai
ra
ble sy
stem and th
e mainten
a
n
c
e st
rategy a
pplied ha
s a
great impa
ct
on
system
reli
ability. Currentl
y
t
he periodi
c mai
n
tenance
strategy
i
s
widely used for the power
system. M
o
re adva
n
ced
maintena
nce
strategie
s
such
a
s
con
d
i
t
ional
mai
n
te
nan
ce have been
studie
d
in
re
cent yea
r
s. Comp
ared to
conve
n
ti
onal
sub
s
tation
s,
the sy
stem
architectu
re
s of
smart substat
i
ons become
more
compl
e
x which adver
sely affect the system
reli
ability. However,
the po
ssi
bility to apply more
advan
ced m
a
inten
ance strategi
es in
sma
r
t
sub
s
tation
s will
definitely co
mpen
sate th
e
sh
ortcomin
g
s
in
term
s of
stru
ctural
co
mplexity, corresp
ondi
ngly their
reliability may
reach or even exceed that
of c
onventional substations. Ex
isting reliability studi
es
either
simplif
y smart
sub
s
tation system
s to non
-r
ep
airabl
e sy
ste
m
s [1-3], or
merely
con
s
i
der
perio
dic m
a
i
n
tenan
ce
strategy [4-5]. The study
whi
c
h con
s
id
ers a
nd co
mpares diffe
rent
maintena
nce strategi
es fo
r the sm
art substations is stil
l limited.
To satisfy the
s
e two a
s
pe
cts of reli
ability analysi
s
, reli
ability simulat
i
on metho
d
n
eed
s to
be
studie
d
. T
he
simulatio
n
method
ha
s to ad
apt to
the complexit
y
of t
he syst
em stru
cture of
sma
r
t sub
s
ta
tions, a
nd
h
andle
vario
u
s
m
a
intena
n
c
e
strategie
s
. Among th
e
existing
stu
d
ies,
reliability block diagram
(RBD) [1-3] and fault tr
ee analysis [6] met
hods
have been widely used,
but these met
hod
s are only
suitable for n
on-rep
a
ir
a
b
le
systems. Fo
r repai
rable system
s,
Marko
v
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 5, May 2014: 3745 – 37
53
3746
state
spa
c
e
[7] method
is wid
e
ly u
s
ed
. Ho
wever,
d
ue to th
e
structural
com
p
lexity of sma
r
t
sub
s
tation
s,
applying
this
method
is p
r
one to
le
adin
g
to
state
sp
ace
explo
s
io
n p
r
obl
em
which
decreases the availability of the method.
Focu
sin
g
o
n
the p
r
ote
c
tion sy
stem i
n
sm
art
sub
s
tation, thi
s
pape
r a
nalyzes
and
comp
ares th
e relia
bility and availa
bility of sma
r
t
substatio
n
s
wi
th typical alt
e
rnative n
e
t
w
ork
architectu
re
s
and different maint
ena
nce strategi
es. In
orde
r to meet
the analytica
l
requi
reme
nts
of smart su
bstations, RBD
method
an
d Monte-Ca
rlo
simulatio
n
are combi
ned to form a pra
c
tical
approa
ch. Th
e effectivene
ss of the ap
proa
ch i
s
de
monst
r
ated b
y
detailed example
s
, and
the
reliability of different protection syst
em
s
are evaluated comprehansi
v
ely.
2. Altern
ativ
e Archi
t
ec
tu
res of Smart
Substa
tions
T
h
e
ne
tw
or
k
o
f
s
m
ar
t su
bsta
tio
n
s
c
o
ns
ists
o
f
sub
s
tation-level
net
work an
d p
r
o
c
ess-level
netwo
rk.
As t
he p
r
o
c
e
s
s-l
e
vel net
work is
re
sp
on
sibl
e for the t
r
an
smissio
n
of
sampling
valu
es
and tri
ppin
g
sign
als which are of
critical im
po
rtan
ce to
p
r
ote
c
tion fun
c
tion,
this
pap
er
only
discu
s
ses th
e netwo
rk archite
c
ture
of pro
c
e
ss l
e
ve
l
.
Acco
rding t
o
the pra
c
tices in
Chin
a, this
pape
r discu
s
se
s the following altern
ative netwo
rk a
r
chitectures [8]:
1) "Netwo
rk-sampli
n
g
-
net
work-trippi
ng"
. The sampli
ng value
(SV) [9] and th
e
Gene
ric
Obje
ct Orien
t
ed Substati
on Events (GOOSE)
a
r
e
both transmitted throu
gh the Ethernet
swit
che
s
. Thi
s
archite
c
tu
re is con
s
iste
nt wi
th the IEC 618
50
standa
rd a
nd
has th
e sim
p
lest
network structure,
but its reliability has
alway
s
been questi
oned as it relies on
Ethernet swit
ches
and extern
al time sou
r
ce re
quire
d for syn
c
hroni
zed
sa
mpling.
2) "Di
r
e
c
t-sa
mpling
-
net
wo
rk-trippi
ng". The
opti
c
al
fibers
of SV use p
o
int-t
o
-poi
nt
con
n
e
c
tion
s, while th
e tra
n
smi
ssi
on of
GOOSE
still depe
nd
s on
Ethernet
switch
es. In t
h
is
architectu
re,
the voltage
a
nd
curre
n
t si
gnal
s fro
m
di
fferent me
rgi
ng u
n
its
are
time align
ed
via
resampli
ng te
chn
o
logy thu
s
eliminatin
g the dep
end
en
ce on externa
l
time source.
3) "Di
r
e
c
t-sa
mpling
-
direct
-tripping". Thi
s
archit
e
c
tu
re is si
milar t
o
that of con
v
entional
sub
s
tation
s e
x
cept for the
electri
c
al ca
bles a
r
e
re
pl
ace
d
by optical fibers. This archite
c
tu
re
eliminate
s
the need fo
r Ethern
e
t switch
es an
d extern
al time sou
r
ce, at the expense of
sacrificin
g
the simpli
city of network.
Figure 1. Structure
Diag
ra
m of a Powe
r Tran
sform
e
r
Protectio
n
System
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Com
p
rehensi
ve Eval
uation of Reliability
of Prot
ection
System
in Sm
art Substation (Ji
pu Gao)
3747
In ord
e
r to
ca
rry o
u
t a h
o
ri
zontal
contrast
with
conve
n
tional
sub
s
tati
ons, thi
s
p
a
p
e
r ta
ke
s
conve
n
tional
su
bstatio
n
s as the
ba
seline fo
r
co
mpari
s
o
n
. L
a
ter i
n
thi
s
pape
r, a
typica
l
transfo
rme
r
p
r
otectio
n
sy
stem in a 110
kV smart
sub
s
tation is used
as an exam
ple. The sy
stem
stru
cture whil
e adoptin
g "direct
-
samplin
g
-
direct-t
rippi
n
g
" is sho
w
n in
Figure 1.
3. Maintena
n
ce Stra
tegie
s
for Sm
art
Substa
tions
3.1. Primar
y
Mainten
a
nc
e
Method
s
The followi
ng
primary mai
n
tenan
ce meth
ods [10] may
be appli
ed in
sma
r
t sub
s
tat
i
ons:
1) Op
eration
a
l inspectio
n
. It include
s in
spe
c
tion
s
co
ndu
cted by b
o
th mainten
a
n
ce
crew
and onli
ne m
onitorin
g
syst
em.
2)
Co
rre
ctive
mainten
a
n
c
e (CM). It i
s
also kno
w
n
as re
pair af
ter failu
re. Si
nce
the
system
ha
s
alrea
d
y failed
whe
n
exe
c
u
t
ing the main
tenan
ce ta
sk,
this meth
od
may se
riou
sly
threaten the
safety of devices an
d maint
enan
ce p
e
rso
nnel.
3) Time
-ba
s
e
d
maintena
nce. It schedul
es mainte
nan
ce prog
ram b
a
se
d on time
which
belon
gs to
preventive
m
a
intena
nce (PM). Su
ch
maintena
nce
method
re
quire
s
optimi
z
ed
maintena
nce cycle, othe
rwi
s
e it will lead
to eit
her a lack of maintena
nce o
r
re
pair
surplu
s.
4)
Con
d
ition
-
based m
a
int
enan
ce
(CB
M
). Ba
sed
o
n
conditio
n
monitori
ng
a
nd fault
diagn
osi
s
of device, this
method a
r
ran
ges m
a
int
ena
nce ta
sks bef
ore d
e
vice fai
l
ure by an
alyzin
g
device
status and devel
opi
ng tren
d. Thi
s
mainte
nan
ce method
req
u
ire
s
that the
degradatio
n of
device fun
c
tio
n
is dete
c
tabl
e, and there i
s
a defina
b
le
potential failu
re co
ndition.
Comp
ari
s
ion
s
of the abov
e mainten
a
n
c
e method
s a
r
e sho
w
n i
n
T
able 1. Note
that in
this pap
er o
u
t
ages
cau
s
e
d
by time-ba
s
e
d
and
con
d
ition-b
a
sed mai
n
tenan
ce a
r
e
called
plan
n
ed
outage
s, whil
e outage
s ca
use
d
by corre
c
tive mainten
ance are
call
ed unpl
anne
d
ones.
Table 1. Co
m
pari
s
on
s of the Primary Ma
intenan
ce Me
thods
Item
Inspection
CM PM CBM
Will
cause component failure?
N
Y
Y
Y
Will
cause unplanned s
y
stem out
age?
N
Y
N
N
Will
cause planned s
y
stem outag
e?
N
N
Y
Y
3.2. Mainten
a
nce Str
a
te
g
i
es
Thro
ugh
com
b
ination
of the ab
ove pri
m
ary mainte
nan
ce meth
o
d
s, two
main
tenan
ce
strategi
es
cal
l
ed peri
odi
c maintena
nce and co
nditi
on
al maintena
n
c
e are co
nstit
u
ted, as sh
o
w
n
in Table
2. F
o
r exam
ple, the pe
riodi
c
maint
ena
nce
strate
gy use
s
in
spe
c
tion
and time
-ba
s
ed
maintena
nce method
s;
a
s
these
metho
d
s can
not
g
u
a
rante
e
expl
oring
all fault
s
of d
e
vice
s,
the
perio
dic m
a
in
tenan
ce strategy need
s to inclu
de corre
c
tive mainten
ance method
as well.
Table 2. Main
tenan
ce Strat
egie
s
for Smart Substatio
n
s
Method
Strateg
y
Inspection CM
PM CBM
Periodic maintenance
√
√
√
Conditional maintenance
√
√
√
4. Reliabilit
y
Anal
y
s
is Method
In ord
e
r to
cope
with the
compl
e
x sy
stem st
ructu
r
e
of sma
r
t sub
s
tation
s an
d
con
s
id
er
variou
s main
tenan
ce st
rat
egie
s
, this p
aper
com
b
in
es the RB
D and Monte
-
Carl
o sim
u
lat
i
on
method
s to constitute a practical relia
bil
i
ty simulation method.
4.1. Reliability
B
l
ock Diagram
The
RBD me
thod
can
de
scrib
e
the
logi
cal
co
nne
ctio
ns
amon
g all
the
comp
on
ents to
perfo
rm spe
c
ific system f
unctio
n
s. It is su
itable fo
r syste
m
s
where
com
pon
ents a
r
e fail
ure
indep
ende
nt and no
n-rep
a
i
rable [11].
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 5, May 2014: 3745 – 37
53
3748
Suppo
se
p
P
P
P
,
,
,
2
1
are the minimal path set
s
of the system,
i
X
is the state variabl
e
of the
i
-th co
mpone
nt in the system. As the whol
e
system is
co
nne
cted by the minimal p
a
th
s
e
ts
in parallel, the s
t
ruc
t
ure f
unctio
n
of the system i
s
[11]:
1
()
1
(
1
(
)
)
j
p
P
j
X
X
(1)
Assu
me th
e f
a
ilure
time
of
all th
e
com
p
onent
s exhi
bi
ts expo
nentia
l dist
ributio
n,
then fo
r
any comp
one
nt, the reliabil
i
ty function is:
t
i
i
i
e
t
p
t
R
)
(
)
(
(2)
Repl
aci
ng th
e corre
s
po
nd
ing
state variable
i
X
in eq
uation (1
) wi
th
the reliabi
lity
function
i
p
of each
component, the sy
stem reliability function
()
sys
R
t
is obtained.
4.2. Monte-Carlo Simulation
Suppo
se
()
T
Ft
is the di
stribu
tion functio
n
of ra
ndom
variabl
e
T
. If
()
T
Ft
is
a
monotoni
cally
incre
a
si
ng functio
n
, then for all
(0
,
1
)
y
,
1
()
T
Fy
is uniquely dete
r
mined. Let
()
T
YF
T
, then the distribution fun
c
ti
on of rand
om
variable
Y
is
[
11]:
1
1
()
P
r
(
)
P
r
[
(
)
]
Pr
[
(
)
]
[(
)
]
,
0
1
YT
T
TT
Fy
Y
y
F
T
y
TF
y
FF
y
y
y
(3)
Clea
rly, if random varia
b
le
Y
exhibits the uniform di
stribution on
(0
,1
)
,
1
()
T
TF
Y
will have the
distribution function
()
T
Ft
.
Acco
rdi
ng to
the ab
ove p
r
i
n
cipl
e, the M
ont
e-Ca
rlo si
mulation met
hod ca
n
be
use
d
to
cal
c
ulate the reliability of the protection syst
em
s by
using
repeat
ed statisti
cal
experim
ent and
the system
structure functi
on defined
in (1). The process
is illustrat
ed in more details in [12]. For
compl
e
x syst
ems like sma
r
t sub
s
tation
s, Monte-Ca
rlo
simulation is
a more practi
cal and effici
e
n
t
approa
ch in contra
st to the traditional a
n
a
lysis meth
od
s su
ch a
s
Ma
rkov
chain.
5. Case Stud
y
5.1. Reliability
Parameters
In view of th
e la
ck
of lon
g
-term
stati
s
tics
of reliabil
i
ty param
eters of
com
pon
ents
i
n
sma
r
t sub
s
tat
i
on, this pap
e
r
adopt
s the followin
g
hypo
these
s
:
1) The failu
re
and rep
a
ir ra
tes of all com
pone
nts exhi
bit exponenti
a
l distrib
u
tion
.
2) The
reliabi
lity of protection IED in sm
art
su
bstation
shoul
d not b
e
lowe
r than
that of
conve
n
tional
sub
s
tation. T
he ne
w p
r
ote
c
tion IED
in
sma
r
t su
bstat
i
on re
pla
c
e
s
the tran
sform
e
r
input (an
a
log
ue) mod
u
le
s by SV modules, and re
place the inp
u
t/output (bin
ary) mod
u
le
s b
y
GOOSE modules, therefore the reliability param
et
ers
can and should
be
close to that of
conve
n
tional
prote
c
tion. According to [13
],
this paper
set the failure rate to 0.01/y.
3) The reliability of merging units,
circ
uit breaker IED, Ethe
rnet switches
and
synchro
n
ization clo
c
k sh
ou
ld not be
less than that of protectio
n
IED.
4) Co
mmuni
cation media
contai
ns o
p
tical fiber
a
nd
optical tra
n
sceiver. The protection
and me
rgi
n
g
unit nee
d l
e
ss opti
c
al t
r
an
sceivers t
o
tran
smit S
V
and G
O
O
SE for network
architectu
re
u
s
ing Eth
e
rn
et swit
ch tha
n
t
hose u
s
ing p
o
int-to-point
fi
ber co
nne
ctio
n.
Con
s
id
erin
g
the fact that the more o
p
tical tra
n
sce
i
vers in
the
same IED, t
he high
er te
mperature
of the
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Com
p
rehensi
ve Eval
uation of Reliability
of Prot
ection
System
in Sm
art Substation (Ji
pu Gao)
3749
GOOSE/SV module, an
d the high
er bit error rate, the failure rate o
f
communi
cat
i
on media i
s
se
t
to 0.0033/y [14-15] when u
s
ing Ethe
rnet
swit
ch, and 0
.
005/y for point-to-p
o
int co
nne
ction.
5) A
s
the
reli
ability of no
n
c
onve
n
tional
transdu
ce
r i
s
still lo
w n
o
w,
the failure
rat
e
is set
to twice that
of p
r
ote
c
tio
n
IED. O
n
t
he
cont
rary,
as th
e
conv
entional
tran
sdu
c
e
r
i
s
m
a
ture
enou
gh, its failure rate is set to half of protection IED.
6) F
o
r
co
nditi
on-b
a
sed
mai
n
tenan
ce, th
e
inspe
c
tion p
e
riod
is set to
one
month
[16]. It is
assume
d that if the residu
al lif
e of a compone
nt is less than 50
%, the potential failure can
be
detected by inspection and perfe
ct repair will be performed.
Detaile
d para
m
eter setting
s for the reli
a
b
ility analysis are sh
own in
Table 3.
Table 3. Parameters for Reliability Analysis
Component Pa
ra
meter
Sy
s
t
e
m
P
a
r
a
m
e
te
r
Component
λ
/y
-1
μ
/d
-1
Protection IED
0.01
0.5
PM C
y
cle
2
y
e
a
r
s
Merging Unit
0.01
CBM Inspection C
y
cle
1 month
Circuit Breaker I
E
D
0.01
P-F Residual Life
50%
S
y
nchronization Clock
0.01
Repair Deg
r
ee
100%
Ethernet S
w
itch
0.01
Simulation End Time
10
y
e
a
r
s
Communication Media
0.0033
Number of Simul
a
tions
1000
Instrument T
r
ans
ducer
0.005~0.02
For conveni
e
n
ce, case 1-4
are define
d
for
differe
nt archite
c
ture
s a
s
sh
own belo
w
:
Ca
se 1: Co
nventional
sub
s
tation,
Ca
se 2:
Smart substatio
n
with direct
-sam
pling-dire
ct-t
ri
pping a
r
chite
c
ture,
Cas
e
3: Smart s
ubs
tation with direc
t
-sam
pling-network-trippi
ng archi
t
ecture,
Cas
e
4: Smart s
ubs
tation with network
-s
ampling
-
net
w
o
rk
-trip
p
ing a
r
chite
c
tu
re.
5.2. Reliability
Calculation
for Non-repairable S
y
s
t
ems
Assuming that the protection system i
n
the
sm
art substation i
s
non-repairable, reliability
indexe
s
ca
n
be cal
c
ulat
ed dire
ctly by usi
ng th
e RBD met
hod. For dif
f
erent netwo
rk
architectu
re
s,
the system reliability and MTTF are
sh
own in Fig
u
re
2 and Figu
re
3 resp
ectivel
y
.
Figure 2
an
d
Figure 3
sho
w
that, the
re
liability
of the
prote
c
tion
sy
stem in
conv
entional
sub
s
tation
(case 1
)
is
sign
ificantly highe
r than th
o
s
e i
n
sma
r
t su
bst
a
tions
(case 2-4
)
. For tho
s
e
in
sm
art sub
s
tation
s,
the dire
ct-sampli
ng-di
re
ct
-tri
p
p
ing a
r
chitecture (ca
s
e
2
)
has th
e hig
h
e
st
reliability and the network
-sampling-network-trippi
ng archit
ect
u
re (case4)
has the lowest
reliability.
Figure 2. Reli
ability for
Non-repairable S
y
stems
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 5, May 2014: 3745 – 37
53
3750
Figure 3. MTTF for No
n-re
paira
ble Syst
ems
The ab
ove result
s seem
to indicate th
at, t
he reliabi
lity of protect
i
on sy
stem i
n
sma
r
t
sub
s
tation
ca
n ha
rdly rea
c
h the l
e
vel o
f
the convent
ional o
ne. B
u
t in fa
ct, the RB
D a
naly
s
is
method h
a
s t
he followi
ng
defect
s
: (1) t
he self
-tes
t capability of optical fibe
r ca
nnot be exp
r
essed
in the mo
del,
(2)
advan
ce
d
mainten
a
n
c
e
strate
gie
s
su
ch a
s
CBM
cannot b
e
sim
u
lated. In o
r
d
e
r
to fully refle
c
t the adva
n
ta
ges of smart
sub
s
tation
s,
and g
e
t a m
o
re
obje
c
tive
re
sult, reli
ab
ility
analysi
s
for repairable
system need
s to be don
e.
5.3. Av
ailabilit
y
Calculati
on for Repai
r
able Sy
stems
Monte-Ca
rlo
simulatio
n
is
adopte
d
to calcul
at
e the a
v
aiability for repairable
systems. In
this pa
rt of a
nalysi
s
, two
maintena
nce
strategi
es li
st
ed in Ta
ble 2
are
con
s
id
ered for the t
h
ree
netwo
rk architectures (case 2-4
)
.
As
for
the
convention
a
l one (ca
s
e
1), only pe
riodic
maintena
nce strategy i
s
co
nsid
ere
d
for referen
c
e u
s
e.
a) MTTFF
MTTFF indi
cates mea
n
time to first failure of re
pairable sy
stems. The system
MTTFF
s
for variou
s sy
stem
s are
sh
own in Fig
u
re
4.
Figure 4. Mean Time to Fi
rst Failu
re
cas
e
1
c
as
e2
cas
e
3
c
as
e4
MTTF/y
25.
952
7.
084
6.
677
6.
453
0
5
10
15
20
25
30
MTTF/y
cas
e
1
c
as
e2
cas
e
3
c
as
e4
Per
i
o
d
i
c
Mai
n
tenance
49.
953
12.
52
12.
621
12.
309
Co
ndi
ti
o
n
‐
bas
e
d
Mai
n
tenance
396.
597
331.
13
305.
017
0
50
100
150
200
250
300
350
400
450
MTTFF/y
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Com
p
rehensi
ve Eval
uation of Reliability
of Prot
ection
System
in Sm
art Substation (Ji
pu Gao)
3751
b) Total downtime
The total
downtime
D
own
T
sho
w
n in
Figu
re
5 me
asure
s
the lo
ss bro
ught by
syst
em
failure.
D
own
T
is made up of pl
anne
d do
wnti
me and u
npl
anne
d do
wnti
me. As the u
nplan
ned
outage
cau
s
e
d
by corre
c
tive maintena
nce has a mo
re
serio
u
s im
pa
ct, the unplan
ned do
wntim
e
D
ow
n
CM
T
is use
d
in this pap
er to ind
i
cate the lo
ss
of utility brought by unplan
ned outa
g
e
s
.
(a)
Con
s
id
eri
ng all events
,
D
ow
n
T
(b)
Con
s
id
eri
ng unpl
anne
d
outage
s only
,
D
ow
n
CM
T
Figure 5. Total Do
wntime
c) Availability
Whe
n
co
nsi
d
ering all the f
a
ilure eve
n
ts,
availability is defined a
s
:
()
/
Dow
n
A
TT
T
(4)
Whe
r
e
T
is
the total s
i
mulation time, i.e.,
10 years
in this
paper.
Whe
n
co
nsi
d
ering o
n
ly the unplan
ned
o
u
tage event
s, it is defined as:
()
/
Do
w
n
CM
A
TT
T
(5)
The result
s of availability are shown in Fi
gure 6.
cas
e
1
c
as
e2
cas
e
3
c
as
e4
Per
i
o
d
i
c
Mai
n
tenance
60.
886
137.
042
148.
413
157.
025
Co
ndi
ti
o
n
‐
bas
e
d
Mai
n
tenance
9.
613
9.
718
9.
718
0
20
40
60
80
100
120
140
160
180
Total
Downt
i
me
/h
c
a
se1
c
ase2
c
a
se3
c
ase4
Per
i
o
d
i
c
Mai
n
tenance
40.
896
117.
072
128.
448
137.
031
Co
ndi
ti
o
n
‐
bas
e
d
Mai
n
tenance
1.
728
1.
248
1.
248
0
20
40
60
80
100
120
140
160
Unpl
a
nned
Downtime
/h
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 5, May 2014: 3745 – 37
53
3752
(a)
Con
s
id
eri
ng all events
(b)
Con
s
id
eri
ng unpl
anne
d
outage
s only
Figure 6. System Availability
d
)
. D
i
sc
uss
i
on
s
Via analyzin
g
Figs. 4~6, it
can b
e
co
ncl
uded that:
1) Mai
n
tena
nce
strategy
has great i
m
pact
on
system avail
a
b
ility. While a
dopting
peri
o
dic
maintenance, reliability of protec
tion
sy
stem in
sm
art substation i
s
obviously l
o
wer than that in
the conventi
onal
su
bstati
on. On
the
contrary,
whi
l
e
ad
opting
con
d
itional
maintena
nce,
the
availability of protectio
n
system in smart su
bstati
on is gen
erally higher than that of the
conve
n
tional
protectio
n
system. This
sho
w
s
that, applying
con
d
itional main
tenan
ce is t
he
prima
r
y meth
od to enha
nce the reliabilit
y of smart su
bstation
s.
2) When ap
p
l
ying perio
dic maintenan
ce
, case
2 amo
ng different n
e
twork a
r
chitectures h
a
s t
h
e
highe
st avail
ability and ca
se4 i
s
the wo
rst, wh
ere
a
s
after applying
conditio
nal
maintena
nce, the
differen
c
e
s
a
r
e very little,
whi
c
h im
plie
s applyin
g
adv
anced m
a
inte
nan
ce
strate
g
y
plays
a mo
re
critical rol
e
than usi
ng diffe
rent network
arch
itectu
re
s
in enha
ncin
g
the system a
v
ailability.
3) Th
e
key re
aso
n
why co
nditional m
a
i
n
tenan
ce i
s
a
b
le to g
r
eatly
improve
syst
em availabilit
y is
that it can si
gnifica
ntly r
educe the pro
bability of unplann
ed out
a
ges a
nd redu
ce un
ne
ce
ssary
plann
ed outa
ges. Th
e fact that MTTFF is greatly
improve
d
also owe
s
mu
ch to conditio
nal
maintena
nce
whi
c
h
ca
n de
tect pote
n
tial
failure
s b
e
fore fun
c
tional f
a
ilure
s, the
r
e
b
y red
u
ci
ng t
h
e
prob
ability of unpla
nned o
u
t
ages.
cas
e
1
c
as
e2
cas
e
3
c
as
e4
Per
i
o
d
i
c
Mai
n
tenance
0.
99931
0.
99844
0.
99831
0.
99821
Co
ndi
ti
o
n
‐
bas
e
d
Mai
n
tenance
0.
99989
0.
99989
0.
99989
0.
997
0.
9975
0.
998
0.
9985
0.
999
0.
9995
1
1.
0005
Availab
ility
(A
l
l
events)
cas
e
1
c
as
e2
cas
e
3
c
as
e4
Per
i
o
d
i
c
Mai
n
tenance
0.
99953
0.
99866
0.
99853
0.
99844
Co
ndi
ti
o
n
‐
bas
e
d
Mai
n
tenance
0.
99998
0.
99999
0.
99999
0.
9975
0.
998
0.
9985
0.
999
0.
9995
1
1.
0005
Availab
ility
(Unpl
a
nned
Failu
res
on
ly)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Com
p
rehensi
ve Eval
uation of Reliability
of Prot
ection
System
in Sm
art Substation (Ji
pu Gao)
3753
6. Summar
y
The followi
ng
con
c
lu
sion
s
are ma
de ba
sed on the sim
u
lation re
sult
s:
(1) While using periodic maintenance stra
tegy, the ad
opted
net
wo
rk arch
itecture
signifi
cantly a
ffect the avail
ability of the prot
e
c
tion system
in sma
r
t
sub
s
tation. Among
different
netwo
rk a
r
chi
t
ecture
s, the
dire
ct
-sampli
ng-di
re
ct-tri
p
p
ing archite
c
t
u
re ha
s the hi
ghe
st availabi
lity
and the
network-sampli
n
g
-
netwo
rk-tri
p
p
i
ng archite
c
tu
re is the
wo
rst. The re
sult
s
also
sh
ow th
at
the availabiltiy of the protection
sy
stem in sma
r
t sub
s
t
a
tion is lower than that of the co
nventio
nal
prote
c
tion sy
stem no matt
er whi
c
h n
e
twork a
r
chite
c
tu
re is ad
opted.
(2) Conditional
maintenance
strategy can greatly improve the availabilit
y of the
prote
c
tion
system in
sma
r
t su
bstatio
n
. Applyi
ng ad
vance
d
main
tenan
ce
strat
egie
s
is
mo
re
importa
nt tha
n
ado
pting o
p
timized
net
work
archite
c
tures with
re
gard
to en
ha
ncin
g the
system
availability, and imp
r
oving
mainten
a
n
c
e st
rat
egy i
s
the
pri
m
ary approa
ch
to ma
king t
h
e
availability of the p
r
ote
c
tio
n
sy
stem in
sma
r
t
sub
s
tation ex
ceed
th
at of conventi
onal
prote
c
tio
n
sy
st
em.
(3) Combi
nat
ion of
Monte
Ca
rlo
si
mula
tion an
d th
e
RBD metho
d
ca
n
ea
sily consi
der
variou
s m
a
int
enan
ce
meth
ods,
and
ove
r
co
me th
e
st
ate spa
c
e
ex
plosi
on
pro
b
l
e
m fa
ced
wh
en
using other
methods as
Markov ch
ain, thus maki
ng it an practi
ca
l way to analyze the reliability
of smart sub
s
taions.
Referen
ces
[1]
Z
hang Pe
icha
o
,
Gao Xian
g.
A
nalysis
of
rel
i
a
b
ility
an
d c
o
mpon
ent
i
m
porta
nce for
a
ll-d
i
git
a
l
protectiv
e
system
s.
Proc
eed
ings
of the CSEE. 2008; 2
8
(1): 77-8
2
.
[2]
P Zhang, L Portillo, M Ke
zunov
ic.
Reli
a
b
ility an
d co
mp
on
ent impo
rt
ance an
alysi
s of all-dig
i
tal
protection system
s.
Po
w
e
r S
ystems Confere
n
ce an
d Expos
ition, IEEE PES. 2006: 13
80
–
138
7.
[3]
Hou W
e
ih
on
g, Zhang Peic
h
ao, Hu Yan. Reli
ab
ilit
y
an
d
availa
bil
i
t
y
st
ud
y
of the dig
i
tal substati
on
sy
s
t
e
m
. Pow
e
r
System Protec
tion an
d Co
ntrol
. 201
0; 14: 3
4
-38.
[4]
PM Anders
on,
SK Agar
w
a
l. An improv
ed mo
del for pr
otecti
ve s
y
stem r
e
li
a
b
ilit
y
.
IEEE T
r
a
n
sactio
ns o
n
Reli
ab
ilit
y. 199
2; 41(3): 42
2–4
26.
[5]
R Bil
linto
n, M
F
o
tuhi-F
iruz
a
bad, T
S
Sidh
u. De
termi
nati
on of
the opti
m
um
routi
ne t
e
st an
d se
lf-
checki
ng i
n
ter
v
als in
protect
i
ve rel
a
yi
ng u
s
ing
a rel
i
ab
ili
t
y
mod
e
l.
IEE
E
T
r
ansactio
n
s
on Pow
e
r
System
. 20
02; 17(3): 66
3–
669
.
[6]
Han Xia
o
tao,
Y
i
n Xi
an
gg
en,
Z
han
g
Z
he. A
p
p
licatio
n
of fault
tree a
nal
ys
is m
e
thod
in
rel
i
a
b
il
it
y
an
al
ysi
s
of substation comm
unic
a
tio
n
s
y
stem.
Power
System
Technology
. 200
4; 01: 56-59.
[7]
N
y
oma
n
Rizkh
a Emill
ia, Su
ya
nto, W
a
rih Ma
hara
n
i.
Isol
ate
d
Word Rec
o
g
n
itio
n Usin
g Er
god
ic Hid
de
n
Markov M
ode
l
s
an
d Ge
netic
Alg
o
rithm.
T
E
LKOMNIKA In
don
esia
n J
our
nal
of E
l
ectrica
l
En
gin
eer
ing
.
201
2; 10(1): 12
9-13
6.
[8]
F
an Ch
en, Ni
Yimin, Do
u
Ren
hui, etc.
Anal
ys
is of ne
t
w
ork sc
heme
for process l
a
yer in sm
art
substation.
Automation of Elec
tric Power Sy
stem
s
. 20
11; 1
8
: 67-71.
[9]
IEC Std. 61
850.
Commu
nicati
on
net
w
o
rks a
nd
s
ystems in s
u
bstation-P
a
rt
9-2: Sp
ecifi
c
communic
a
tio
n
service map
p
i
ng (SCSM)- Sa
mple
d ana
lo
gu
e valu
es over ISO 8802-3.
[10]
Bo Ye, L
e
i
Xu
an, Bo
Xu, etc
.
T
he mai
n
ten
ance st
rat
e
g
y
for
optimiz
ing
distri
bution transformer life
c
y
cle cost.
T
E
LKOMNIKA Indones
ian J
ourn
a
l of Electrica
l
Engi
neer
in
g
. 2013; 11(
10): 60
01-6
007.
[11]
Raus
and M, H
o
yla
nd A.
System r
e
li
abi
lity theory:
mo
dels,
st
atistical met
hods, a
nd a
ppl
icatio
ns
. Ne
w
York: John W
i
l
e
y
S
ons. 20
04.
[12]
Jipu Gao, Xu
He, Chang
b
ao Xu. Impacts of lif
e distributi
ons on re
liab
ilit
y an
al
ys
i
s
of smart
substations.
T
E
LKOMNIKA Indo
nesi
an Jo
u
r
nal of Electric
al Eng
i
ne
eri
n
g
.
2013; 1
2
(4): 3
059-
306
7.
[13]
W
ang
Ruic
he
n
,
Xue A
n
ch
en
g
,
Bi T
i
anshu,
et
c. T
i
me-v
ar
y
i
n
g
fai
l
ure
rate
e
s
timation
of re
l
a
y pr
otectio
n
devic
es and th
eir regi
on
al diff
erenc
es ana
l
y
s
i
s.
Automation
of Electric Pow
e
r Systems.
20
12; 05: 11-
15+
23.
[14]
ABB Research Ltd. Re
liabilit
y
c
a
lculation for substation
autom
ation s
y
stems. S
w
itzerland: EP
248
09
41 B1. 2
013.
[15]
Cui
Limi
n, W
u
Yunn
a. T
he lif
e c
y
cl
e re
lia
bil
i
t
y
ev
al
uatio
n
o
f
optica
l
ca
bl
e.
T
E
LKOMNIKA Indo
nes
ia
n
Journ
a
l of Elec
trical Eng
i
ne
eri
n
g
. 201
3; 11(4)
: 2024-2
0
2
8
.
[16]
Mainte
nanc
e a
nd man
a
g
e
me
nt standar
ds of
po
w
e
r
equ
ipm
ent con
d
itio
n. Beiji
ng: C
h
in
a Po
w
e
r Press
,
201
2.
Evaluation Warning : The document was created with Spire.PDF for Python.