I. INTRODUCTION
uA3
uA2
RELIABLE
uA1= uA
D-l
iA3
iB3
iA2
iB2
iA1
iL
iR
ua
ia
iB1
uB3
uB2
uB1 = uB
Electric arc
uF
RF
(1)
The arc voltage model (1) is used in developing the new
algorithm for overhead line protection. By using the
symmetrical components approach, the left line terminal
voltage va can be represented as follows:
di
di 0
(2)
u A = RAP iL + K RiL0 + LPA L + K L L + u F
dt
dt
L0A LPA
RA0 RAP
and
are
K
=
L
RAP
LPA
2
sequence current flowing through the line. These coefficients
can be calculated in advance and for a given line they are
constant. The fault current ia is the sum of the currents flowing
to the fault location from both line terminals:
ia = i L + i R
(3)
Taking into account the arc model (1), the fault voltage at
the fault location, uF, can be expressed as follows:
u F = U a sgn(ia ) + RF ia
(4)
U a sgn(iL + iR ) + RF (iL + iR )
l +
u A = r P i L + K R i L0 l + l P L + K L
dt
dt
sgn(i L + i R ) U a + (i L + i R ) R F
di L0
(6)
(7)
u A = # = #
u A (t N ) u N
(9)
di
di 0
E N r P iL + K R iL0 + l P L + K L L
dt
dt
sgn(K 3 ) =
#
#
=
sgn( K 3, N ) sgn[iL + iR ]
K 4,1 iL + iR
K4 = # =
#
K 4, N iL + iR
( t = t1 )
(t = t N )
(t = t N )
( t = t1 )
(t = t N )
( t = t1 )
(10)
(11)
(12)
E = K1 + K 2
(8)
l
l
u A = [E sgn(K 3 ) K 4 ] U a + = J U a +
R F
R F
where is a matrix formed by vectors E, sgn(K3) and K4:
J = [E sgn(K 3 ) K 4 ]
(14)
[l
Ua
( )
RF ]Topt = J T J
( )
Where J T J
III. LEAST ERROR SQUARES PARAMETER ESTIMATION
(13)
1 T
1 T
J uA
(15)
3
IV. SIMULATED TRANSMISSION SYSTEM
A. Arcing Fault Model
Transmission line arcing faults are widely experienced in
power systems and are usually categorized as transient faults.
The modelling of arcing faults and their interaction with the
power system is quite important for the design phase of the
proposed adaptive autoreclosing [13-16].
Several models are used for describing the behaviour of
arcs. Most models are used for circuit breaker arc modelling
[17-20] and several of them have been applied to the
modelling of long arcs and arcing faults [13-16]. Recently the
existing dynamic model has been modified for the
representation of short arcs [21].
There are various concepts for arc modelling and the most
popular rules depend on thermal equilibrium. The thermal
model has the longest history of the dynamic arc models, since
Cassie [17] and Mayr [18] introduced the first description of
arc conductivity in the form of a first order differential
equation. These dynamic equations are improved and
modified to increase the models' validity and to reduce the
computational burden. The thermal long arc model of Kizilcay
[13] and model of Johns [14] have been recently introduced
for modelling the arcing fault as:
g=
(G g )dt
Transmission Line
Power Network
(16)
SW
R
TACS Field
90
91
91
i(t)
v(t)
9000
[V]
[V]
6000
4000
3000
1000
0
-2000
-3000
-5000
-8000
-20
-6000
-10
-SW) v :SW
10
20
*10 3
30
-F_R
a- Case 1; =20 s
Fig. 3. Two typical arc characteristics.
-9000
-20
-10
-SW) v :SW
10
20
-F_R
b- Case 2; =550 s
*10 3
30
4
B. Fault Nature Detection
l [km]
V. ESTIMATION RESULTS
9 1
9 0
8 9
8 8
8 7
0 .0 2
0 .0 4
0 .0 6
0 .0 8
0 .1
t [s ]
98
3
2 .8
96
2 .6
94
2 .4
90
2 .2
R []
l [km]
92
2
1 .8
88
1 .6
86
1 .4
84
1 .2
82
80
0 .0 2
0 .0 4
0 .0 6
0 .0 8
0 .1
t [s ]
0 .0 2
0 .0 4
0 .0 6
0 .0 8
0 .1
t [s ]
T w o po rt
O ne po rt
4 .5
3 .5
3
3 .5
1 .5
1
2 .5
U [kV]
2 .5
a
U [kV]
0 .5
1 .5
0 .0 2
0 .0 4
0 .0 6
0 .0 8
0 .1
t [s ]
1
0 .5
0
0 .0 2
0 .0 4
0 .0 6
t [s ]
0 .0 8
0 .1
2.1
0 .4
2.05
0 .2
R []
U [kV]
0 .3
0 .1
-0 .1
0 .0 2
0 .0 4
0 .0 6
0 .0 8
1.95
0 .1
t [s ]
1.9
0.02
0.04
t [s]
0.06
0.08
10
8
6
4
U [kV]
The real nature of the fault arc is not exactly known. Thus, the
arc voltage can have different characteristics. The algorithm
presented assumes a pure rectangular arc voltage wave formsee equation (1). The reason for such a selection is its
simplicity and because the fault model is linear. The
introduction of nonlinear arc models consequently makes the
fault model non-linear, so the non-linear estimation methods
should be used. This would be essentially a much more
complex problem for practical realisation.
By processing voltages and currents for the case of a 10 km
arcing fault over tower footing resistance RF = 2 , and by
changing the arc characteristics, the sensitivity to the arc
voltage wave form is investigated. In Figs. 10 and 11 the
estimated fault distance and tower footing resistance for an arc
with = 20 s are respectively presented.
For the above case, in Fig. 12 the estimated arc voltage
amplitude and the arc voltage wave form are presented
respectively.
For the same fault conditions, but with the changed arc
voltage wave form, in Fig. 13 the estimated arc voltage
amplitude for = 55 E-5 s is presented.
2
0
-2
-4
-6
-8
Estimated
Real
-10
0
0.02
0.04
t [s]
0.06
0.08
Fig. 12. Estimated arc voltage amplitude and original arc voltage wave form
for = 20 s.
10
8
6
4
l [km]
12
0
-2
-4
11.5
-6
11
-8
Estimated
Real
-10
0
0.02
0.04
0.06
0.08
t [s]
Fig. 13. Estimated arc voltage amplitude and original arc voltage wave form
for = 550 s.
10.5
10
0
U [kV]
12.5
0.02
0.04
t [s]
0.06
0.08
6
Table I Sensitivity to arc voltage wave form.
[9]
Case
10-5 s
RF
Uarc
kV
Lfault
km
Err RF
%
Err Uarc
%
2
1
55
2
2
2
5.2
5.2
10
10
3.9061
0.0872
-20.2923
-1.2109
Err
Lfault
%
2.6677
0.4482
[10]
[11]
[12]
[13]
VI. CONCLUSION
In the paper, a new time domain adaptive estimation
numerical algorithm for overhead line protection, based on
synchronized phasors technology and implementation of
parameter estimation theory, is presented. The algorithm was
used as an example for discussing modelling, simulation,
testing and algorithm aspects in developing modern fault
locators. It is shown that through realistic arcing fault
modelling, the accuracy of fault locators can be improved.
Simultaneously it was shown that by introducing arc in the
design of fault locators, the adaptive autoreclosing schemes
can be designed. On the other hand it was shown that testing
procedures must include the essential components related to
the physical nature of line faults. In this sense, the critical
point of arc modelling was investigated and it was proved that
realistic arc modelling can improve the quality of fault
locators.
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
REFERENCES
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[23]
[24]
VII. BIOGRAPHIES
Vladimir Terzija (M95, SM2000) is the EPSRC Chair Professor in Power
System Engineering at the School of Electrical and Electronic Engineering,
The University of Manchester, UK.
Nagy I. Elkalashy (S06) is a PhD student at the Helsinki University of
Technology, Helsinki, Finland.
Gary Preston is an MSc student at the School of Electrical and Electronic
Engineering, The University of Manchester, UK.
Vladimir Stanojevi is a PhD student in the Control and Power Group at
Imperial College, London, UK.
Goran trbac (M94) is a Professor in the Control and Power Group at
Imperial College, London, UK.