Simulation of the 500 kV SF6 circuit breaker cutoff

process during the unsuccessful three-phase

autoreclosing
I. Naumkin, M. Balabin, N. Lavrushenko, R. Naumkin
Abstract--The paper presents a study on the electromagnetic
transients in 500 kV transmission line equipped with shunt
reactors. Conditions of unsuccessful three-phase autoreclosing are
discussed in the case when the DC component in current flowing
through a circuit breaker is significantly greater than the
amplitude of the steady-state current which prevents the current
from crossing the zero line for a prolonged period of time.
A model of the electric arc for simulation in a long time spans
was developed and employed in this paper. Simulations of an
accident in a 500 kV grid are presented.
Keywords: electromagnetic transients, three-phase
autoreclosing, circuit breaker failure, sulfur hexafluoride circuit
breaker, arc model
I. INTRODUCTION
t the present time in Russia overhead transmission lines
(TL) of the ultra-high voltage of 1150kV are operated at
500kV. To compensate the high reactive power consumption
of the 1150kV-type TLs they are equipped with an increased
number of shunt reactors (SR). For example, the transmission
line 1150kV connecting substation Altay and substation
Itatskaya is equipped with four SRs, two reactors on each
end of the line. If four reactors are turned on, then the line
capacitance becomes overcompensated; if three reactors are
turned on then the line works in a near-resonance mode.
Taking this into account, protective relaying is configured in
order to avoid commutations with three shunt reactors enabled.
A
Nevertheless, not all peculiarities of this type of
transmission lines were taken into account completely. As a
consequence, on the February 26th 2007 a major accident
occurred at the Altay substation (SS): phase A of three-
phase sulfur hexafluoride circuit breaker (ABB HPL-550B2)
was destroyed. The analysis of the accident has revealed that
prior to the breaker breakdown a one-phase line-to-ground
fault occurred followed by the three-phase line tripping and
autoreclosing. The fault still was not cleared successfully at the
moment of the autoreclosure and relay protection issued
another line tripping command. The breaker at the phase A
was not able to extinguish the electric arc because the current
did not cross the zero line for more than 130 ms due to a large
I. Naumkin is with Bolid Ltd., Novosibirsk, 630015 Russia (e-mail of
corresponding author: [email protected]).
M. Balabin is with SibEPRI, Novosibirsk, 630126 Russia (e-mail:
[email protected]).
N. Lavrushenko is with SibEPRI, Novosibirsk, 630126 Russia (e-mail:
[email protected]).
R. Naumkin is with SibEPRI, Novosibirsk, 630126 Russia (e-mail:
[email protected]).
Paper submitted to the International Conference on Power Systems
Transients (IPST2011) in Delft, the Netherlands June 14-17, 2011
DC transient presented in the current. The DC transient
appeared at the non-faulted phase A as a result of the
autoreclosing commutation of the line.
The accident happened during the three-phase tripping of
the line at a non-faulted phase. If a one-phase tripping was
used, the accident could have been avoided because the circuit
breaker performs successfully in faulted conditions. However,
such an accident can happen again if appropriate security
measures are not taken, even during routine switching
operations.
II. CONDITIONS NECESSARY FOR THE DC COMPONENT TO OCCUR
DURING LINE AUTORECLOSING
A degree of compensation of capacitive reactance of TL
using SR is defined by the formula:
K=
X
TL
X
SRT
, (1)
where X
SRT
=X
SR
/ N
r
is a total inductive reactance of N
r

shunt reactors connected to a TL; X
TL
is a total capacitive
reactance of TL, which is determined by the positive sequence
capacitive reactance of TL with sufficient accuracy:
X
TL
=1/(b
1
0
l
TL
) ; b
1
0
is a TL positive sequence capacitive
reactivity; l
TL
is a TL length. If compensation degree is close
to 1 then TL is in the resonance mode.
An electric scheme of the transmission line connected to a
power source on the first end and its equivalent circuit using
lumped parameter elements are correspondingly presented on
the figures 1a and 1b. The transmission line is represented by a
Pi network.
~ ~
~
a)
~
b) c)
I
I
L
I
C
Cl/2 Cl/2
Rr
Lr
Rl Ls Es Ll
Rr
Lr
L C
R E
Fig. 1. Scheme of the transmission line under study and its equivalent
circuits
To obtain an analytical solution, consider the simplified
scheme (Fig. 1c). On the scheme 1c inductance of line L
l
is
neglected, impedances in the TL branch and SR branch are
taken into account by adding them to the resistance R
C
. Such
simplification of the scheme 1b allows us to study qualitative
properties of the circuit based on the simple analytical
solution.
Solution for the steady-state current through the circuit
breaker on a simplified scheme using the complex
representation can be written as

I
ss
=

E

Z
, (2)
where the impedance of the circuit

Z is

Z=R j
oL
o
2
LC1
=Rj
X
L
X
C
X
L
X
C
, (3)
assuming that X
L
=oL , X
C
=1/(oC) .
As it follows from (2) and (3), steady state current

I
ss
=0
when X
L
=X
C
, e.g. in the resonance mode. Condition for the
resonance is
oLC=1 . (4)
Since natural frequency of the circuit is f
nat
=1/ ( 2n.LC)
and driving frequency is f
dr
=o/ ( 2n) then (4) means that the
natural frequency of the circuit is equal to the driving
frequency of the circuit: f
nat
= f
dr
.
When X
L
=X
SR
/ N
r
, (3) can be transformed to

Z=R j
X
SR
X
C
X
SR
N
r
X
C
,
which means that if no reactors are connected ( N
r
=0 ) then
Im(Z )=X
C
0 and the current has a capacitive nature. If
we start adding reactors to line, then eventually N
p
X
c
>X
SR

will become true and the current will become inductive by
nature (e.g. Im(Z )>0 ).
In the considered case sinusoidal driving voltage
e=E
m
sin(oT+) is applied to the resonant circuit under
study. According to (2) and (3) we can write as follows:
i
ss
=
(
X
L
X
C
1
)
E
M
Z
sin (ot ++) , (5)
where =arctan
X
L
R( X
L
/ X
C
1)
,
Z=
.
R
2
(
X
L
X
C
1
)
2
+X
L
2
.
Total current i through the circuit breaker is a sum of the
steady-state current i
ss
and the transient current i
tr
:
i=i
ss
+i
tr
. Transient current is found as a solution to the
following system of equations:
t
L
t
C
d
2
i
L , tr
dt
2
+t
L
d i
L ,tr
dt
+i
L ,tr
=0; (6)
u
tr
=L
d i
L ,tr
dt
; (7)
i
C , tr
=C
du
tr
dt
; (8)
i
tr
=i
L , tr
+i
C ,tr
, (9)
where t
L
=L/ R and t
C
=RC are characteristic time
constants of the circuit; i
L ,tr
and i
C , tr
are transient currents
flowing through inductance and capacitance, accordingly.
Let's consider characteristic equation of the differential
equation (6):
t
L
t
C
p
2
+t
L
p+1=0 .
Its roots are:
p
1
=
1
2t
C
(
1
.
14
t
C
t
L
)
;
p
2
=
1
2t
C
(
1+
.
14
t
C
t
L
)
.
(10)
According to the (6)-(9) solution for the transient
component of the current through the circuit breaker assuming
zero initial conditions i
L
t =0
=0 , i
C
t =0
=0 is expressed as:
i
tr
=
E
m
Z
sin (+)
1
p
1
2
p
2
2
|( 1+t
L
t
C
p
1
2
) ( p
2
2
+
+o
2
)e
p
1
t
(1+t
L
t
C
p
2
2
) ( p
1
2
+o
2
)e
p
2
t

.
(11)
Values of p
1
and p
2
are real if t
C
/t
L
1/ 4 . For a
typical set of parameters of 500 1150 kV transmission line
equipped with reactors the stronger assumption is valid:
t
C
t
L
. (12)
Under the assumption (12) we can apply the series
expansion
.1x1x / 2 to the expression (10):
p
1

1
t
L
;
p
2

1
t
C
. (13)
If we simplify the formula (11), the result will be as
follows:
i
tr
=
E
m
Z
sin (+)
|
X
L
X
C
e

t
t
C

(
1+
X
L
X
C
t
C
t
L
)
e

t
t
L

. (14)
Let's study two cases of the circuit breaker reclosing: the
first case with =0 , the second case with =n/ 2 .
In the first case =0 , the total current trough the circuit
breaker is equal to
i =
(
X
L
X
C
1
)
E
m
Z
sin (ot +)

E
m
Z
sin
|
X
L
X
C
e

t
t
C

(
1+
X
L
X
C
t
C
t
L
)
e

t
t
L

.
(15)
Taking the assumption (12) into account, the circuit breaker
current for t >3t
C
will be determined mostly by the steady-
state current
i
ss
=
(
X
L
X
C
1
)
E
m
Z
sin(ot+) (16)
and the transient current corresponding to the time constant
t
L
:
i
tr
=
E
m
Z
sin
(
1+
X
L
X
C
t
C
t
L
)
e

t
tL

E
m
Z

X
L
Z
e

t
tL
.
(17)
The value of X
L
/ Z may exceed X
L
/ X
C
1 significantly
because X
L
/ Z1 and X
L
/ X
C
1 approaches the zero line
when X
L
-X
C
. Therefore, the DC component in current
flowing a circuit breaker may exceed the steady-state
component and the current would not cross the zero line for a
long time. As follows from (16) and (17), time needed for the
transient component to decay to the amplitude of steady-state
component is approximated by the following formula:
t
1
t
L
ln
X
L
/ Z
X
L
/ X
C
1
(18)
and the value of t
1
can be quite significant. Thereby, we have
shown that the DC component is present in the current flowing
through the circuit breaker if reclosing is done with =0 .
The magnitude of a DC component may exceed the amplitude
of the steady-state sinusoidal current and delay the zero-
crossing event.
In the second case =n/ 2 the current through the circuit
breaker may be expressed as:
i =
(
X
L
X
C
1
)
E
m
Z
cos (ot +)

E
m
Z
cos
|
X
L
X
C
e

t
t
C

(
1+
X
L
X
C
t
C
t
L
)
e

t
t
L

.
(19)
Similar to the first case, for t >3t
C
the circuit breaker
current will be determined mostly by steady-state current
i
ss
=
(
X
L
X
C
1
)
E
m
Z
cos(ot+) (20)
and transient current corresponding to the time constant t
L
:
i
tr
=
E
m
Z
cos
(
1+
X
L
X
C
t
C
t
L
)
e

t
tL

(
X
L
X
C
1
)
E
m
Z

R
Z
e

t
tL
. (21)
Since RZ , it is apparent from (20) and (21) that DC
component is much less then steady-state current amplitude in
any case. Problem with delayed zero-crossing does not arise in
this case.
Plots shown on the Fig. 2 illustrate analysis made in this
section. Model parameters typical for 500 kV transmission line
were used. Transmission line is 400 km long with two shunt
reactors connected. Em=428.7 kV, R=25 ; XL=765 ,
XC=663 , L=97.5 ms, C=0.12 ms. The formula (18) yields
an estimate t10.182 s which is consistent with the results
shown on the Fig. 2A.
A)
B)
Fig. 2. Current through the circuit breaker A) if =0 B) if =/2.
III. HYPOTHESES EXPLAINING THE CAUSES OF SF6 CIRCUIT BREAKER
FAILURES.
The considered situation happens in non-faulted phases of a
breaker during the three-phase autoreclosing in the cycle
O-t-CO, when an unsuccessful reclosing of an overhead
transmission line is taking place. In this case, next to reclosing
the idle non-faulted phases after the dead time, a line tripping
signal is issued immediately. If there is a large DC component
flowing through the breaker, cutting off the breaker becomes a
considerable problem.
Until recently air circuit breakers have been used to cut off
the 500-750 kV lines. They handle the considered line
reclosing well. However, usage of the new SF6 circuit
breakers led to new unexpected problems. Analysis of the
accidents has shown that the cause of the failures of arc
extinguishing was the presence of the large DC component in
the cutoff current which delayed the moment of zero crossing.
We have shown conditions for the occurrence of such DC
component on a simple scheme.
We would propose the following hypothesis in order to
explain the reasons for the success of air circuit breakers, as
well as the reasons for failures of SF6 circuit breakers.
Firstly, during the reclosing of air circuit breakers, the
closure of an intercontact gap of the breaker is most likely to
occur at times close to reaching the voltage maximum, that is
when =n/ 2. It may be explained from a physical point
of view when the contacts are moving to meet each other the
contact closure is being carried out due to the breakdown of
the air intercontact gap, which occurs close to the voltage
maximum. As we already know, in these circumstances the DC
component flowing through the breaker is negligible, and there
are no problems during the subsequent cutoff.
Analysis of the oscillograms obtained during the operation
of SF6 circuit breakers has showed that in contrast to the air
circuit breakers the intercontact gap may also close when the
voltage is near zero, that is when =0 . The insulating
properties of sulfur hexafluoride compared to the air may be
the reason for that. When contacts are moving sulfur
hexafluoride does not break until the contacts are approached
to very short distances, which may be got over in time less
than a quarter-period of power frequency (5 ms), that is the
time during which the voltage does not reach its peak value.
Contact closure occurs with an equal probability for all values
of voltage, including values close to zero. Under these
conditions a large DC component in the cutoff current appears
to be leading to an accident.
Another reason for the difference in the behavior of an air
and gas-insulated circuit breaker may be the following.
Modern SF6 circuit breakers have a closed volume of arc
chute, in which the intensity of the arc extinguishing is largely
dependent on the energy of the arc defined by the cutoff
current. During the concerned reclosing of currents in non-
faulted phases their value is negligible, especially near the
resonance modes. Therefore, SF6 circuit breakers cannot cut
off these currents until they cross the zero line. Such behavior
of the breaker is considered as advantageous when handling
small inductive currents, because the overvoltages are minimal
in this case. However, in this particular case, these qualities
are represented as disadvantageous, as they may lead to
accidents. Air circuit breaker, having a supply of air under
constant pressure, extinguishes the arc off at a rate almost
independent of the current. A more rapid decay of the DC
component may also contribute to more rapid arc
extinguishing in the air circuit breakers, because the significant
resistance of the extinguishing arc, with its intensification
during extinguishing, is added to the active resistance of the
circuit.
Taken together, this leads to the fact that there is no
problem of cutoff currents for air circuit breakers in the non-
faulted phases during three-phase autoreclosing cycle. To test
the proposed hypotheses, the simulation of arc extinguishing
processes in the above scenarios of accidents was made. For
this purpose, a model of an extinguishing arc adequately
reflecting the process of arc burning over the entire range of
currents from peak values to zero-crossing was developed.
IV. THE ELECTRIC ARC MODEL
Based on the integral description of the arc as a cylindrical
nonlinear resistor u=Ri with with varying length l and
cross-sectional area S, it is possible under the assumption of
constant pressure to write the following relations:
R=( h)
l
S
; H=hlS ; P=p( h)lS , (22)
where R is the arc resistance; is the resisitivity; h is the
volumetric enthalpy; p is the heat loss per unit of the arc
column volume.
The energy conservation law with integral description of
the arc is represented as:
dH
dt
=iuP. (23)
A.M. Cassie's assumptions [1] are valid in the area of
current amplitude values:
S =var ; h=h
C
=const ; p=p
C
=const .
As a result
R=
C
l
S
, whence S =

C
l
R
; (24)
H ( R, l )=h
C

C
l
2
R
; (25)
P( R, l )=
p
C

C
l
2
R
. (26)
O. Mayr's assumptions [1] are valid near the zero-crossing
of current:
S=S
M
=const ;
h=var; =
M
e
h/ h
M
;
p( h)=p
M
=const .
(27)
As a result we get
R=
M
e
h / h
M
l
S
M
, whence h=h
M
ln
(

M
l
RS
M
)
; (28)
H ( R, l )=h
M
ln
(

M
l
RS
M
)
lS
M
=
= h
M
S
M
l
(
ln

M
S
M
+ln l ln R
)
;
(29)
P( l )=p
M
S
M
l . (30)
As it may be seen from given computations, enthalpy of
electric arc in asymptotic cases of high and low currents
depends only on the arc resistance and its length:
H=H ( R ,l ) In accordance with the assumptions of
A.M. Cassie and O. Mayr we get from (23):
dH( R ,l )
dt
=iuP;
H
R
dR
dt
=Ri
2
P
H
l
dl
dt
.
(31)
Introducing the notation:
H
M
=h
M
S
M
l ; R
CM
=

C
l
S
M
; R
M
=

M
l
S
M
;
P
M
=p
M
S
M
l ;
R
1
=
h
C
h
M
R
CM
; R
2
=
p
C
p
M
R
CM
; R
3
=

M

C
R
CM
.
(32)
According to (24)-(30) we can form the following Table I:
TABLE I
MODEL TYPE
H
R
H
l
P
A.M. CASSIE'S
MODEL
H
M

(
R
1
R
)
1
R
H
M

1
l

(
R
1
R
+
R
1
R
)
P
M

(
R
2
R
)
O. MAYR'S
MODEL
H
M
( 1 )
1
R
H
M

1
l

(
ln
R
3
R
+1
)
P
M
( 1 )
Both models may be combined to the one generalized
model by introduction of the functions:

m
(i )=

m0
+(1
m0
)
(
2
i
2
I
C
2
)
i
2
I
C
2
, if iI
C
;
1, if iI
C
,
(33)
where m=1,2,3; IC is transition current from low currents to
high currents (transition current from O. Mayr's model to
A.M. Cassie's model). The expression (33) is obtained if
functions

m
(i )
are in the following form:

m
(i )=

n=0
N

mn
i
2n
, if iI
C
;
1, if iI
C
,
providing continuity of function values and their derivatives at
IC point.
Generalized model of the arc for the entire range of currents
from peak values to zero-crossing is obtained as follows:
H (i , R)
d ln R
dt
=Ri
2
P(i , R)H
l
(i , R)
d ln l
dt
,
(34)
where
H( i , R)=H
M (
R
1
R
)

1
( i )
; P(i , R)=P
M
(
R
2
R
)

2
(i )
;
(35)
H
l
(i , R)=H
M

3
(i )
R
1
R
+
|
1
3
(i )

ln
R
3
R
+
(
R
1
R
)

1
( i )

.
(36)
When
m0
=0 (m=1,2,3) the equation (34) with current
value i=0 becomes O. Mayr's equation, and with i IC
becomes A.M. Cassie's equation. In general case the
assumption
m0
0 may be made, then with

m
(i )=
m0
=const , Schwarz model [3] is obtained.
V. THE ARC EXTINGUISHING MODEL FOR GAS CIRCUIT BREAKERS
In modeling of arc extinguishing process for gas circuit
breakers, the convention l=const is made. When using
equation (34), model of arc in the breaker is described by the
equations:
u
1
u
2
, if t t
I
;
u
1
u
2
k
ag
R
I
i=0 , if t=t
I
;

M
|(
R
1
R
)

1
( i )

d ln R
dt
=
Ri
2
P
M

|(
R
2
R
)

2
(i )

;
u
1
u
2
k
ag
Ri=0 ,
if t >t
I
i=0 , if RR
m
,
(37)
where
k
ag
is the number of arc gaps in the breaker;
R
I
is the initial resistance of arc burning;

M
=h
M
/ p
M
=H
M
/ P
M
is Mayr's time constant;
R
m
is the resistance of arc extinguishing.
To determine the parameters of equation (37) experimental
results are used. We introduce the Cassie's time constant:

C
=h
C
/ p
C
. From (32)-(35) we get a number of relations:
R
1
=

C

M
R
2
; (38)
the arc voltage is constant at high currents
(u
eff
=u
0
const )
,
then:
P
M
=
u
0
2
R
2
, (39)
when
R=R
max
(in case of failure of arc extinguishing during
a thermal breakdown):
u
R
max
2
P
M
=R
max
(
R
2
R
max
)

20
; (40)
if i=0:

M
(
R
1
R
i=0
)

10
d ln R
dt

i =0
=
(
R
2
R
i =0
)

20
. (41)
If the time constant of the arc near zero current

M

(Mayr's time constant) and time constant of the arc at high
currents

C
(Cassie's time constant) are obtained from the
experiment, then having the experimental oscillograms ue(t),
ie(t) for the arc model we may consistently determine all the
other parameters of the model (37). The minimum of residual
function should be obtained:
F(
20
)=

t
0
t
k
|
R( t )R
e
( t )

2
( t )dt , (42)
where ( t ) is the weighting function.
Sequence of calculations is as follows:
1)

20
value is set;
2)
P
M
, R
1
, R
2
,
10
are obtained from (38)-(41) and
experimental oscillograms;
3) solution of the equation (37) is obtained;
4) residual function is obtained according to (42);
5) parameters are determined when a minimum value of
residual function is obtained.
VI. SIMULATION OF ELECTRIC ARC IN AIR AND SULFUR HEXAFLUORIDE
CIRCUIT BREAKERS
On Fig. 3 results of simulation of air circuit breaker cutoff
process are presented for the case when the current does not
cross zero. As can be seen on the figure, arc resistance makes
the current decrease. When the current approaches the zero
level, voltage of electric arc increases and, eventually, current
chopping occurs.
Fig. 3. Arc extinguishing process during air circuit breaker cutoff
iidbr current through a circuit breaker without arc simulation (ideal
breaker);
iarc current through an air circuit breaker with electric arc taken into
account;
ich current chopping;
uarc voltage of electric arc;
utrv voltage on the circuit breaker contacts after extinguishing the
electric arc
On Fig. 4 results of simulation of a SF6 circuit breaker
cutoff process are shown. The same current as in the previous
case flows through the breaker. As indicated on the picture, the
arc in the breaker is not extinguished for a prolonged period of
time. During that period of time arc extinguishing conditions
are changing significantly due to considerable heating of
breaker contacts and arc chute walls. Conditions of fanning of
the arc by sulfur hexafluoride become significantly worse.
Changed conditions lead to a situation when the arc is not
extinguished even when the current crosses the zero line. To
model the breakdown process completely, the time-variability
of parameters must be taken into account in the arc equations.
Fig. 4. Arc extinguishing process during SF6 circuit breaker cutoff
VII. CONCLUSIONS
1. In case of an unsuccessful three-phase autoreclosing
during a one-phase transmission line fault it is possible that a
circuit breaker has to cut off the current that does not cross
zero for a long time. This situation occurs on 500 kV and
higher voltage lines with shunt reactors in a near-resonance
mode on a first harmonic. Reclosing of non-faulted phases of a
transmission line when voltage of the power source is close to
zero leads to a DC component of considerable magnitude
occurrence in the current through a circuit breaker. At the
same time, sinusoidal steady-state component is quite small in
the magnitude. As a result, the current does not cross zero for a
long period of time.
2. It was shown that air circuit breaker can cut the current
which does not cross zero. Current chopping occurs in this
case. Sulfur hexafluoride breaker can not cut off such current,
leading to the circuit breaker damage. Additional measures
should be taken to prevent such accidents when SF6 circuit
breakers are used.
VIII. REFERENCES
[1] Cassie A.M. A New Theory of Arc Rupture and Circuit Severity.
CIGRE, 1939, Paper 102, p.1-14.
[2] Mayr O. Beitrag zur Theorie des Statishen und Dinamischen
Lichtbogen. Archiv fr Elektrotechnik. 1943. Bd.37, 12, s.588-608.
[3] Schwarz J. Dynamisches Verhalten eines gasbeblasenen,
turbulenzbestimmten Schaltlichtbogens. Elektrotechn. Z., 1971, 92,
6, s.389-391.