You are on page 1of 10

IEEE Transactions on Power Dclivety, Vol. 14, No.

4, October 1999

1317

OVERVOLTAGES DUE TO SWITCHING OFF AN UNLOADED


TRANSFORMER WITH A VACUUM CIRCUIT BREAKER
Marjan Popov, incinber IEEE
University St. Cyril and Melhodious, Skopje
Deparlinent o f Electrical Engineering. P.O. Box 574,
Rcuublic o f Macedonia
Absrriicf-When a vacuuni breaker i s used to disconnect an
uiiloaded transforiner, high overvoltages m a y result on tlic
secondary sicle. I ' l i c plieiinmena of c u r r e n t chopping and
multiple reigoitions iiftcr coiitact openiiig. sliurtly before current
zero, play a significant role in generating these overvoltages.
Furthermore, t h e transient recovery voltage (I'KV) is also
influenced by virtual current chopping wliicli inay lead tu f a s t
interruptions and reignitions. 'l'lic paper reports on a vacuuin
brealier model implemented in the ATP-EMTP prngrain.
Simulations liiive been performed to cletermine overvoltages
generated by the vacuum breaker, before and after suppression.
Network eleiiients were simulatecl by simplified models of liniitecl
complexity.

Key words:
transformer

vacuum circuit breaker, TIIV, ATP-EMTP,

I. 1NTROI)UC'I'ION

Expcriencc shows that tlic t~insforiiiersarc exposed to


high overvoltages when switched by vacuuin circuit breakers,
and when repetitive reignitions takes place. I1 was sliowii iii
previous work 151 tliat due lo tlic mulliple reignition in the
vaciiiiiii circuit breakcr. voltage cscalatioii can occur duriny
load shedding wlicii l l i e Vaciiiiin breaker is iiscd as ii
generator circuit breaker. L.oadcd arc furnace traiisforiiicrs
iiiay also be subjected to high voltage transients due to
repetitive reignition in the circuit breaker [I I]. Unloaded
trunsforiners were found to be subject to severe overvoltages
when switched by vaciiiiin breakers. The possibility o f tlic
vacuum circuit breaker to chop the current due to the iinstablc
arc burniiig around current zero. and the virtual chopping
~iliciioineiiaIlia1 iiiiglit occiir wlicii switching olf a three
phase load, lead to very liigh 'I'KV wliicli iiiay cause damage

Enrique Acha,

member IEEE
l l i e University orGlasgow

Departinen1 of Electrical Engineering and Electronics


Scotland, UI<

to tlic transformer windings. The reason is lhat the stored

magnetic encrgy iii the transformer arter the current has been
chopped i s converted into clectroslalic energy by charging tlic
tramforiner capacitance. By equating [lie stored inagnctic
energy with the stored clectrostatic encrgy. and neglecting
tlic damping, tlic voltage which i s built up at the transformer
tcriiiiiials is:

where I+ i s llie magnetising induclance o f thc transrormer, C


transformer capacitance and I,,,i s the chopping curreiii.
According to (I) it caii be sccii that llie generatcd overvoltagc
depends oii the chopping ciirreiit and t i e surge iinpcdance of
the transformer wliicli might be very high due lo the low
lransforiiier capacilance. lr tlie arc i s reignited in one pole,
liigh frequency current flows in the otlicr poles o f the circuil
breaker (fig. I),which is also followcd by vollage escalation
(lig. 2).

I .

1.1.

PE-072-PWRD-0-05-1998 A paper recommended and approved by


the IEEE Switchgear Committee of the IEEE Power Engineering
Society for publication in the IEEE Transactions on Power Deliveiy
Manuscriptsubmitted October 8, 1997; made available for printing May
18, 1998.

-~

~~

I Iiillucace nl'tlic ieignitcil ~ ~ i r r e iii


i i lonc liolc io lhc otlicr iw pdcs

This may cause enforced current zeros in tlic other two poles,
wliicli happens if tlie aperiodic reignition current induced by
l l i e pole tlial is already cleared is greater in magnitude Lliaii
the iiisLiiiitaiie~usload current and opposite in polarity. 'I'liis
iiiccliaiiisiii lkiiown as virtiial current clioppiiig caii also cause
rcpetilive rcignilion which leads to vcry high TRV. T h e
resulting TIIV in nnc of Lliese IWO poles will be of tlic Sanie
polarity as tlic first pole, but lhc oppnsite oii the other.

111 [ I ]

an expression is used to calculate tlie chopping current


'S

sauce sde

.
UIS

Lk

load sde

MOD=

Fig. 3 Model oftlie VU

which is valid for load currents between 45 A and 170 A.


Here the mean chopping level is calculated according to [4,
141:
Fig 2 Vdtage escalation

Id, =

caused by icigiiilcd CLI!ICI>I

This results i n possible double magnitude on one line-to-line


voltage [ I O ] . The phenomena mentioned above are caused by
the properties of thc circuit. On the other hand the vacutiin
gap with its statistical character plays significant role in the
generation of the high voltage lransients. In ordcr to simulate
these conditions it was required to develop a vactiuni breaker
model with the appropriate statistical behaviour.
The paper focuses on the circuit breaker model and the
parameters are set to simulate the most severe TRV at the
transformer terminals. The EMTP gives an opportunity for
modelling both the circuit breaker and the associated
network
11. CIRCUIT BREAKER MODEL

I n order to simulate the phenomena that appears when


switching off a vacuum circuit breaker it is necessary to
consider the statislical properties of the vactium breaker:
chopping current,
withstand voltage ofthe vactiuiii gap,
high frequency quenching capability.
The vacuum breaker can be inodelled with a variable arc
resistance [I, 6, 81. Another possibility [4, 5, 141 is to
represent the breaker with an ideal switch, type-13 element
(fig. 3) which opens or closes whenever the voltage across tlie
breaker is lower or higher than the withstand voltage of the
given cii-cuit breaker.
ATP's MODELS language was used to control the opening
and closing of the switch, including current chopping at
power and high frequency, withstand voltage or the gap and
the high frequency quenching capability. Due to the random
variation ofthe 60 t l z chopping level it is very hard to predict
the chopping ctirrenl. Former research was rocused on
determining Lhc chopping level or the vactiiiin circuit breaker.

(2 n . .f',
'

I,m .p)"

(2)

/ - power frequency (60 Hz),

7 - amplitude of the load current,


a = 6.2 10-16s,p=
14.3,q=-0.07512
The chopping currents calculated by expression (2)
correspond to those of vacuiiin breakers that use CuiCr
contacts. This value is varied by a uniform distribution with
15% standard deviation. Depending on the moment of contact
opening the chopping current has different value. The closer
the contacts open to current zero, the higher the chopping
current. This implies to higher overvoltage production. The
speed of contact opening and quenching capability are also
very important. The first coiitrols the speed of withstand
voltage build up, whilst the second arc extinguishing at high
frequencies. The withstand voltage is considered to be
linearly dependent on the gap distance and is given by [I]:

U,,

= Ekr . v . t

(3)

where:
Exr- withstand electric field between the contacts,
v - speed of contact separation,
t - timestep ofthe simulation.

After the current has been chopped a 'race' between the


TRV and withstand voltage appears. Whenever tlie TRV
exceeds the withstand voltage reignition O C C I I ~ S . Reopening
can occui if the slope of the current at current zero has a
value lower than the critical current slope, and the magnitude
orthe high frequency current is lower than the high frequency
chopping level. The value of critical slope is found to be
randomly distributed. A random generator is used to select a
critical slope between 30 - 75 Aips.
111. SIMUILATION OF A SINGLE PHASE ClRCUll~

In order to v e r i 6 the validity of the proposed model a


modified single-phase circuit is used, which was first

1319
proposed by Carnerki and prof, Lindinayer (fig. 4). 'l'lie
source U is connected by a sinall resistor (not shown on the
picture) that represents the internal source resistance. The
resistor R , , inductance I./,and capacitance C,, represent tlic
paramctcrs of the bus bars atid cable which are connected to
the load. l'lie rest o f the elements in fig. 4 are load
parameters. T h e model used yields h e saiiie results as the
Lindinayer tiiodel.
b

Lh CO

,
I
R>
CB

Li

- ,

b)L

big 4 Test ciicciit Ibr the VI3 model

u=5134 v
1.?=8.4 intl
RK-lO' n
cb=2nnpl:

R,-8.2 n L.,-SO LIH' ~ ? = n . n


m
R,,,=2.58 n I,,,,=6.X m l l L,,-O.2 I~I
C,,=15 iiF Rl,=50
L,,=SO n l l

Thc circuit hreakcr includes an additional parallel branch that


rcpresents tlic opeii gap The parallel branch is used 10
represent the post zero current which i s usually a few
amperes, due to remanent ions iii the gap and tlic
displacement current dqidt. I t s peak lnagnitiidc dcpcnds on
the slope o f the current before current zero [ I ? ] . I n fig. 5 the
results of tlic simulation in the single phase test circuil are
given. Fig. 5 a) presents the current through the switch at
successful arc extinguishing. B y varying the parameters in
the circuit, as well as tlic inoinent o f contact separation it is
easy to get an unsuccessful arc extinguishing with a few
reignilions. 111 fig. 5 b) and c) the load side voltage and V B
voltage are given respectivcly.

~~

.~

~~~.
~~

c)
Pig. 5 Simolatioiis in a aiiiglc liliarc CircLiit
a) ciirreiit tlirougli llie brcakcr
b)
load sidc voltsgc
c)
vollagc across the brcahcr

Whenevcr [lie hf-current i s quenched successfully the TRV


rises. When the V B voltage withstands tlic T R V no hf-cui-rent
flows through the V13, and it approaches power frcqiicncy.

IV. M O D E L I M P L E M E N T A I ' I O N FOR TRANSFORMER


ANALYSIS
The developed V B tnodcl is suitable for analysing
switching overvoltages 011 an unloaded transformer. An o i l
filled transformer and cable arc connected by a V B as it is
shown in fig. 6. The transformer was modelled by EMTP's
satorable tvansforiiier model. The model provides the
transient response at transformcr terminals. For calculating
transient response along the wiiiding, a more detailed model
is required, but this was bcyond the scope of this p p c r . 'The
cable was modelled by EM'l'P's distributed parameter option.
A l l system data are given in Table I.
For the case studicd excessive overvoltages inay occur at
the secondary terminals of the transformer. This can damage

1320

winding insulation and cause winding flashover to core.


Possible suppression of such overvoltages includes the
installation of capacitors on the low voltage side [l21, or
using preinsertioii resistors on the primary side [I I ] . Another
one is to install an R-C branch phase to ground at the priinary
side as close as possible to the transformer terininals.
TAII1.E I

S Y S I E M DATA

Pare,neter

Value

lrensiollllcr tiatil
lraliiformcr rated uowcr

13MVA
13.810.2 kV, Y,,d 5
60 HZ

596
7 111.'
7 "F
4 111.'

I%

It is important to mention that the surge capacitance values


varicd and they depend on the connection of the HV
transformer windings and chopping voltage factor.
Expression (4) is derived by equating the stored magnetic and
clectrostatic energy and assuming that primary transformer
capacitance is sinall enough so that can be neglected. The
chopping voltage factor k limits no-load overvoltages to lttimes phase to ground rated primary voltagc. I n appendix a
table is given, that shows the calculated values of the R-C
branch Tor different chopping voltage factors. The time
constants of the R-C branch are also included. The selected
resistor assists in damping the transient oscillations. Another
fiinction of this resistor is to prevent current zeros being
forced in the otlier switch poles when ihe first pole to clear
suffers a reignition. It is suggested to lteep the surge
capacitor as small as possible. There are two reasons for this.
The smaller the capacitance, the smaller tlic reignition
current magnitude when the switch reignits.
The resistor should be rated for the contiinioiis capacitor
current, hence its sizc is function of the surge
capacitance selected.

V. RESULTS AND DlSCUSSlON

il 2

IidLV

L\'

Using the network shown in fig. 6, inumerous simulations


were performed with and without the R-C suppression
branch Fig. 7 ihrough fig. I O present the reignition current
througli the vacutim circuit breaker and voltages at the LV
and HV side in phase T before suppression.

I*
Fig. 6 System coiiligoratian for EM1'1' simuliitions

l h e values ofthe surge capacitance and resistor are chosen


by equation (4) and ( 5 ) :

where:
S,, - Icated transformer power i n MVA,
io -magnetising current in %,

f - power tkequcncy i n Hz,

~~~~

~~

Fig. 7 VI? wrrent iii liltiise T

U,, - line-to-line supply voltage in kV,


k -chopping voltagc factor.

r T

it..

R,t2 2
whcre:

L, - inductance ofthe cable,


C, - surge capacitance

The results of the sinidations atier supprcssion are given i n


fig. 11, 12 and 13. According to these results it may be
concluded that significant reduction in overvoltage and
redtictioil in the magnitude of reignited current can be
achieved. Simulations were also made with a capacitor
installed at the LV terminals of the transformer as a protective
dcvice. This yields lower secondary overvoltages but has no
substantial effect on the reignited ciirreiit.

1321

iio

~.~

~~

~~

~.

~~

~~-

sulilxcssioon

It is suggested tliat if virtual current chopping

OCCIIIS,

the use

i~t'lighliiingarresters sliould be considered. Due to the virtilal


chopping tlie peak valucs o f t l i e voltages in two pllases might
rcach oppositc polarity and produce very high line to line

voltages.

~~~

111 I ~ X C i n plinsc v o l t a ~ ca1


i l l i c I IV tr;inslhrmct side bclirrc

I:ig 13 I'liasc to ground voliegcs 81 thc I.V iraoil'orincr sidc aI\cr


sL,(~prcs\I"Il
111 this case liglitiiiiig arresters should be connected no1 only
line to ground but also between tIV tcrininais in order to

Iprotect the transformer. This also depends 011 thc coiiiiectioii


ofthe I-IV windings, and protection requirements for Wye and
DelLa connected HV windings arc different.

1322
VI. CONCI.USION

It was found that the transformers under no load can be


cxposed to vcry high rcpetitive overvoltages if they are
switched off hy a vacuutii circuit breaker. The magnitude and
frequency of such overvollagcs depend on the cliaracteristics
of the systein and tlic statistical behaviour of the switch. A
gcneric E M I P inodel of a vacuum breakcr is described Tor
the siniulalion of these overvoltages. The chopping ciirreiit
strongly depend on tlie type or contact inaterial, tlius can lead
lo different overvoltages. Such a rcsearcli was not inade here
and that is bcyond tlie scope of tliis paper. A properly cliosen
I<-C circuit connected at the primary terminals of tlie
transrormer provides a good overvoltage suppression. The
cost of tliis suppression device is sinal1 compared to the cost
of a potcntial transformer failure. Should tlie network
conliguratioii lead to virtual current chopping, phase to
ground atid phase to phase lightning arresters inay also be
rcquircd in addition to the R-C suppression circuit. Tliis
would provide complete surgc protection for an unloaded
translbriner. Increasing the capacitance connccted to tlie
secondary sidc of the transformer, by adding extra sliuiit
capacilors, or cablcs, would decrease the overvoltages inainly
on the secondary side of the transformer. The use of the ATPE M I P prograni was shown to be a very powerful tool for the
modelling of a vacuiiiii circiiit breaker and the study or
overvoltages caused by switching off an unloadcd
transrormer.

VII. FUTURE POSSIBLE WORI<


More detailed vacuum breaker tnodel will be developed by
taking into accouiit more accurate measured characteristics.
Additional work will be done in order to estimate tlie
statistical overvoltage distribution at different opening timcs.
VIII. ACKNOWLEDGEMENT

1998 rcspeclivcly.. In 1997 lic \\'as a reseilrcli


assistiiiit at the University of Livcrpaol. His
inajor liclds of rescarcli are tire modelling and
transient phenonicne iii power systcias using
EMII'. Mr, I'opov is a membcr o f IEEE.

The first author would like to thank MI'. Gabor Furst,


Vancouver. Canada for his valuable suggestions during the
preparation oftliis paper.

'l',\BLt? II. I'ARAMITFRS 01: 'IIII: R-C RRANCI-I

0 I8

30.43
3651
45.64
43 03

10.59

4.5

1323

DISCUSSION
Toshlhisa Funahashi, Senior Member, Meidensha Corporation,
36-2, Nihonbashi-hakozakicho, Chuc-ku, Tokyo, 103-85 15, Japan,
Yoshihiko Matsui, Meidensha Corporation, Shizuoka, Japan,
Akihiro Amctani, Fellow, Doshisha University, Kyoto, Japan :
1. The authors havc proposed a vacuum circuit breaker model for
a TRV simulation of the breaker. The proposed model is based on
statistical properties of the vacuum breaker.
- chopping current,
- withstand voltage of the vacuum gap,
- lligh frequency quenching capability.
The very powerful simulation capability of ATP-EMTP p r o g m
was demonstrated in the paper. However, the simulation result by
this kmd of models is essentially deterministic even if the
simulation includes the statistical properties of the vacuum
breaker. To ensure the designed surge suppression circuits
effectivencss, numerous simulations are needed. More powerful
method is desirable.
2. As stated in the paper, there is another approach to model the
vacuum breaker, which uscs the arc equation such as Mayrs
equation or Cassics equation. In this approach the arc resistance
is represented as a fimction of breaker currents ruld sourcesidehad-sidc voltages. Alternatively, in reference [AI] an
artificial neural network (ANN) is used to represent the arc
resistance nonlinear characteristic, ANN implementation might
be useful, if it is well timed for a specific breaker. However, it is
not always applicable to anothcr typc of breakers. As a generic
model, the model with a function of voltagcs and cment seems to
be better than an ANN type.
3. We would like to have the authors comments on why eqiiation
(2) was selected for the calculation of the mean chopping level of

the breaker that usc Cu-Cr contacts7 The units for thhc parameters
0 and q are to be stated fur the readers better undcrstanding of
the eqnations physical inclining. The dependency of thc chopping
current on the moment of contact opening the chopping current
was stated in the paper. Could the authors show thc way how to
includc tlus characteristic into the proposed method7 It is
considered when the type of a contact material of the hrcaker is
fixed, the mean value and thc standard deviation of the chopping
cwent could be estimated [ N I
4. Could the authors show the way how to includc the statistical
properties of the withstand voltage in the proposed model? In
equation ( 3 ) , the speed of contact separation is considered to he
constant. This speed might vary \nth t h e . At the t h e to start
moving it might be rclativcly low and then become higher. Also,
the withstand voltage might he considered to be nonlinearly
dependent on the gap distance rather than lincarly. For instancc,
the wiIhstand voltage might saturate or might he the exponential
function of the gap distance, where the exponent equals 1/2 or 314.
5 . The condition of the reopening is.

- the slope of the current at zero has a value lower than (not
then) the critical current slope,

- tbc magnitude of the high frequency current is lower than the

high frequency chopping level.

Is thc high frequency chopping level in the second condition, a


constant parameter or that calculated from the equation (2) with f=
high frequency?

6 . For the suppression of the overvoltages generated by the


vacuiiin breaker, an R-C surge suppresser is uscful. Surge
capacitance value and resistor value are calculated by using
equation (4). Could the authors show the way to derive the
equation (4) by equating the stored magnetic aid electrostatic
energy? In this paper the calculated capacitance is 0.1 to 0.4 U F,
which secms to he acceptable from the point of conventional
scheme. In reference [AS] a conservative approach could he the
addition of 0.25 U F surge capacitors installed on each phase on
the incoming side of the circuit breaker.

7. l h e installation of the surge arrester is ellectivc for the


suppression of ovcrvolkges. In refcrence [IQ], the ovcrvoltagc of
the unloaded transformer, when it is switchcd by tbe vacuum
breaker, is discussed. It is written that +-type transformers

with low-impulsc voltage rating are more vulnerable and the


recommended protection schemcs including the surge arrestcr
application are shown. In cases of motors frequently switched on,
a combination of the R-C suppresser and ZnO-varistors is oiten
used and thk combination has higher performance than the R-C or
ZnO applications although the cost is also higher [A4].

In Fig.6 and Table.1, the meaning of the symbol 5 in the


transformer rated voltage 13.8/0.2kV, YndS is not clear.

8.

References
[AI] J.Kosmac and P.Zunko, A Neural Network Vacuum Circuit
Breaker Model for Simulalion of Transient Overvoltages,
European W P L!wrs Croup Meetinp; (EEUG95).
IIannover, German, November 13-15, 1995
[A21 P.G.Slade, Vacuum lntempters: The New Technology for
Switching and Protecting Distribution Circuits, IEEE
lrunsuctions on Indirstty Applicarions, Vo1.33, No.6,
November/December 1997, pp.1501-1511
1.431 U.L.Swindler, P.Scwartz, P.S.Hmer and S.R.Lmhert,
Transient Recovery Voltage Considerations in the
Application of Medium-Voltage Circuit Drcakers, B E E
Trunsuctions on Indirstty Applications, Vo1.33, No.2,
MarcWApril 1997, pp.383-388
[A41 A.KeinniQcr, The Effect of Protection Devices against
Surges When Switching on Large AC Motors, Blcctric
Muchines and Power Svstems, 20.103-1 12, 1992,
Heinispherc Publishing Corporation

Ogiiz A. Soysal (Frostburg State University, MD, USA):


I would like to congratulate the authors for thcir work.
Vacuum circuit hrcakers are preferred in many industrial
medium-voltage systems; however, thcir major limitation
appears to he the probability of current chopping and arc
reignition, particularly during the interruption of inductive
currents. I agree that RC circuits uscd in addition to surge
arresters improve the transformer protection in such systems
by reducinglhe risk of reignition in vacuum circuit breakers,
and partial internal resonance on transformer windings.
The paper suggests choosing lhe surge capacitancc and
resistor values by cquation (4), which is based on stored
magnetic and electrostatic energy at power frequcncy. The
main purpose of these RC circuits is to filter the high
frequency components that arise during switching transients.

1324

Experience with various industrial networks has shown that


selection of the capacitor and resistance values affects the
transient behavior of the whole system. As the authors
indicated, for smaller capacitance values the magnitude of the
reignition current will be smaller. Or, in other words, a larger
value of capacitance would result in a larger reignition
current, Consequently, there must be an optimum
combination of R and C parameters that would provide an
effective protection. In fact, while (4) and (5) provide an
initial guess for R and C values, the paper suggests keeping
the surge capacitor as small as possible.
1 would appreciate comments o f t h e authors on the following
questions:
Could the authors explain how they determine the final
design parameters?
1 understand that the presented simulation results are
obtained for RC values selected by using (4) and (5).
Were other simulations made for different combinations
(for example using smaller capacitors) to search for an
optimum RC circuit?
Transformers are usually designed to withstand transient
overvoltages up to the basic insulation level (BIL)
defined in standards for each system voltage. Would it
be more appropriate to consider this voltage level in
determination of the RC values?
The transformer model used in this study is apparently
based on the rated values at power frequency. High
frequency performance of transformers has, however, a
significant effect on the system transients. I believe that
wide frequency range representation of transformers
such as presented in [A] and [B] would be more
appropriate in the study of surge phenomena. What is the
authors' opinion about implementation of a transformer
model based on terminal impedance characteristics in the
simulation of switching operations?
[A] Soysal, A. 0. "A Method for Wide Frequency Range
Modeling of Power Transformers and Rotating
Machines," IEEE Trans. on Power Delivery, Vol. 8, No.
4, Oct. 1993.
[B] Soysal, A. 0. and Semlyen, A. "Practical Transfer
Function Estimation and its Application to Wide
Frequency Range Representation of Transformers,"
IEEE Trans. on Power Delivery, Vol. 8, No. 3, July
1993.

comments wc would like to start with the comment that the


point of this work was to introduce dangerous transformer
switching transients that can appear when the circuit is
switched by a vacuum circuit breaker and a way of their
suppression.

T. Funabashi, Y . Matsui, A. Ametani:


Ouestions 1. 6, 8, I n (0.Soysal):
The exprcssion for C, is derived (as mention in the paper) by
equating the stored magnetic energy in the transformer with
thc stored electric energy in the prolcctive capacitor at linear
magnetising induclance. Some limitations are considered, but
the calculated values does not differ a lot from those used in
practice. From a physical point of view after the magnetising
currcnt has been chopped the flux-current characteristics is
above the steady state hystercsis curve and ends with residual
flux. Therelore not all from the released magnetic energy will
charge the protective capacitor. In order to derive an exact
expression it is inevitable for particular transformer measured
B-H loops to be considercd [C3,C4] in ordcr to calculate the
exact released energy that charges the capacitor. According
to referencc [I31 (pp.81-84) and [ l l ] the following equations
are used:

where Lpt is the magnetizing inductance (linear) at rated


magnetizing current io and uq is the rated transformer phase
voltage. The capacitance is expressed as:

L,io' where V=k U,,J is the maximum overvoltage.


C,$= 7

The transformer data were taken for a transformer type Y,d5.


If the type of connection of the primary windings is different,
then different expression for the protective capacitor can be
derivcd. Fig. i l shows simulation of the primary voltages with
the paramcters in Table 11. Referencc [ I I]proposes a valuc oE
suppression resistor greater then a half of the value calculated
hy (5). Our study showed that better suppression is madc if
thc value is chosen by (5).

15,

lo4
..

..

..

..

..

Marjan Popov, member, University St. Cyril and


Methodious, Skopje, Department of Electrical Engineering,
P.O. Box 574, Republic of Macedonia,
Enrique Acha, member, The University of Glasgow,
Department of Electrical Engineering and Electronics,
Scotland, UK:
The authors kindly appreciate the thoughtful remarks of the
discussers that can hopefully lead to further interest and
improvement of this kind of studies. In reply to discussers

'I

0,001 0002 0003 0 004 0.005 0006 0 007 0 000 0009 0.01
lima Is]

Fig, il Primary phase voltage after suppression

1325

2. So far it is well known that the Mayr-Cassic equations


dominate in the representation of SF6 CB arcs. An approach
of this kind of model for gas blast CB reactor switching is
given in [91. When switching oflsmall inductive currents with
a VB arc voltage drop is small (13-25 V) and the current is
chopped before natural currenl zero. Generated overvoltages
depend on the chopping currcnt and hf reignitions. The
amount of reignitions depends on the peak overvoltage due to
the chopping, critical current slope at hf currenl zero, type of
the load (for example length of the buses can influence the
amount o l reignition) and moment of contact opening. When
switching off high vacuum arcs, a modcl of post-acro current
is needed. This modcl is in different time scale (a few Ls
rather 100 ps) and is bascd on dynamic plasma equation ICs,

GI.
3. The chopping current calculated according to expression
(2) is in a good agreement with that calculated by expression
used in [I]:

200

-s

,
,

,
,

,
,

,
,

,
,

,
.

,
.

,
,

.-..-.-

.2w

2:'"

-21

'

O..r0.1

'

-21

0.1

02

0.3

0.4

0.2

0.3

0.4

06

0.7

0.6

0.9

0.5 0.6
time js)

07

0.6

0.9

0.6

x 10~?

Fig. i2. Siinulatcd CB reignited current, CU voltagc and load side


vollage

i, = a - bi - c l o g Z,,
where

a,b,c
are constants which depend on the contact
.

material, i is the load current at the moment of contact


opening and 2, is the surge impedance of thc load thal is
switched. The validity of this expression is limited for load
currents between 45-170 A, so we followed the approach in
reference [4,14] where chopping current is calculated by
expression (2). Complete research for this expression is done
in [CI], and it is calculated according to rcference 11,141.
According to the measurements [C7] of the chopping current
behaviour it cBn be seen that the values vary in a wide range.
The calculated values by our approach in motor switching
circuits vary between 3-7 A. In our example the switched
currents are io = 1.25 A. This is Car below the chopping
currcnt level of the modern VB so the current is chopped
immcdiatcly.
4. Implementation of a variable speed of contact motion is
possible if additional data are known: the force of opening
spring, the mass of mechanism and movable contacl, the
rigidity of CB mechanism etc. Recovery of the full dielectric
slrenglh after arc cxtinguishing in vacuum takes place for a
very short time (4-8 !.IS), so the reignition is possible when the
contacts open shortly before current zero. Since the gap in
that time is vcsy small, thc withstand voltage is significant for
the all process up to the full gap distance. We considered the
niost sever case when the withstand voltage is zero, although
it does have some value which is normally taken to be I kV.
Stalistical properties of the withstand voltage can be
considcred by including SO%
values of the measured
statisLica1 distributions [14].
5 . The hf-chopping level due to lack of information is
considered constant.
In gcncral, the discussors expressed their concern about the
CB model. In order to prove further the validity of our
approach for the test circuit use in [4,14] measured
characteristics proposed by Glinkowski at al. [5], with a
variable critical slope at hf current zero were implemented in
the model and the results are presented in Fig.i2 and Fig.13.
These results show good agrecment with the measuscd givcn
in reference IC21 and sirnulaled given reference 14, 141.

-100

3.25 326 3.27 3.28 3.29 3.3

331 3.32 3.33 3.Y 3.35

x io*

40'

I
L
L
L
_
L

Smeets

Current

Low Current Behaviour and


Chopping of Vncuum Arcs, Ph.D., 'IU

R.P.P.:

Eindhoven, 1987.
Smects R.P.P.: Essential Parameters of Vacuum
Interrupters and Circuit Related to Occurrcnce of
Virtual Current Chopping in Motor Circuils,

International Symposium on Power and Energy,


Sapporo, Japan 1993.
Tuohy E.J., Panek J.: Chopping of Transformer
Magnetising Currents, Part I: Single Phase
Transformers, IEEE Trans. On PAS, PAS-97, No.1
Jan/Fcb 1978.
W.M.C. van den Heuvel, at.al.: Interruption of a DsyType Transformer in No-load by a Vacuum Circuit
Breakers, EUT Report 83-E-141, Eindhovcn 1983,
ISBN 90-6144- 141-2.
Childs SE., Crccnwood A.N.: A modcl Cor DC
interruption in diffuse vacuum arcs, IEEE Trans,
1980, PS-8, pp, 289-294.
Childs S.E., Greenwood A.N.: Events Associated
with Zero Current Passage During thc Rapid
Commutation of Vacuum Arcs, IEEE Trans, 1983,
PS-11, pp. I8 1-188.

1326

[C7].

Gibbs J.D., at al.: Investigation of Prestriking and


Current Chopping in Medium Voltagc SF6 and
Rotating Arc Vacuum Switchgear, IEEE Trans. On
PWD, Vo1.4 No. I , Jan. 1989.

0. Soysal:
Thc discussion by prof. Soysal addresses a few questions
that strongly helps in the further improvement and research
in this field.
2. Simulations with different parametcrs havc been performed
but no research has been donc in order to find the optimum
RC circuit. The best suppression was providcd for thc
parametcrs of the RC branch which give time constants
around 10 ps,
3. Thc generated overvoltages depend on thc chopping level
of the circuit breaker which is not related to the transformer
(and system) BIL in gcneral. Since in this case we had
transformer under no-load, that has a magnetising current

which is below the chopping level, system BIL might be


considcred. However in the motor circuits [CZJ where thc
rated current is above the chopping level, different chopping
levels lead to different overvoltage levels, so we think that the
estimated overvoltage level is more accurate for calculation
of the paramcters of protective elements.
4. Reignitions in vacuum contain high frequency components
that can be in the range of a few kHz up to MHz level. For a
study of a high frequency behaviour high frequency
transformcr model would be more appropriate. If the contacts
open so Pdr before current zero that no rcignitions occur, then
the only load that is seen by the CB is the transformer core.
So the transformer model should consider the non-linearity of
the core [C3, CSl.
[CS]. Ihara S, Panek J., Tuohy E.J.: Chopping of
Transformer Magnctising Currents, Part 11: Single
Phase Transformers, IEEE Trans. On PAS, PAS-102,
No.5 May 1983.

You might also like