|
DYNAMIC IMPULSE CONDUCTION
IN ZnO ARRESTERS |
|
A Huddad
Cardiff Universisy, Wales, UK |
P Naylor
BICC Supertension & Subsea Cables Ltd, UK |
|
|
ABSTRACT
In. this paper, Ilew fast-impulse test data obtained all ZnO surge
arreters with a special coaxial tell cell are presented.
The rine to peak of the discharge current was observed to become
shoner as Ihe magnirude was increased. An
explanation for this phenomefloll is suggested based on changing
current paths within tne material. To accounI for this dynamic behavior,
we propose an equivaleflf circuil which uses parallel paths wirh
associared inductances and resistances to model the conduction process
in the inrergranular layer. |
|
l.introductton |
Since the introduction of zinc-oxide material in 1968, much
research has been directed towards the
characterisation of the electrical behaviour of the material under
various stress conditions. There is now an
extensive published literature on the response of the material to
impulse current stresses of different shape
and amplitude. An aim of some of these investigations is to achieve
an equivalent circuit representation which
would adequately simulate the observed test resuits[ 1-7].
A feature of many of the published equivalent circuits[ 1 ] is the
representation of the zinc-oxide material by two
series sections; i) to account for the resistance of the zinc oxide
grains and ii) to simulnte the properties of the
intergranular layers. The zinc-oxide grains are represented by a low
resistance Renin wh o s~ effect is of
importance at very high impu lse discharge currents only .
The intergranular layers are represented by a paralJel resistance
(RrJ - capacitance (Cjg) network, the resistance branch having a strong
non-linear voltage current characteristic. For impulse currents in
the kiloampere
range, the resistive curront dominates and the capacitance component
can be neglectcd.
The impulse response in the low conduction regime of the material.
however, s hows a significant capacitive component of the current.
For impulse currents in the low-conduction regime, the materia! may
be simulated by the capacitallce branch. Evidence has been published
that indicates that this capacitance is also non¡¤linear. In addition
to this basic representation, other components may be included such
as an inductive component to represent the equivalent inductance of
the metal oxide materiaVarrester body or to account for the mater
ials response to stcep currents.
Much recent research on zinc-oxide surge a!Testers has conCCDlrated
on the very fast transient response
c haracterisation [2-5). Due to the difficulty in obtaining reliab
le test data for fast-rate-of-risc impulse currents, the
IEEE working group 3.4.11 (Application of Protective Devices Subcommittee,
Surge Protective Devices
Committee)[4] has limited its efforts to the modelling of metal oxide
surge arresters to current impulses with rise
times of 0.5).1s or greater. Extensive reviews have been publiShed
[1.4,6] on the ZnO impulse response, voltage
measurement and equivalent circuits.
The present work is concerned with the transition from low to high
conduction under fast-impulse cond ition s.
Firstly, new impro .... ed impul se tCSt data are presented, These
test data were obtained using a fast transient
coaxial test module and improved voltage measurement methods[2,7J.
The new test data obtained show that there
is no evidence of an overshoot on the residual voltage trace. In addition,
they show, for the first time, that the
discharge current exhibits a longer time-to-peak at low amplitudes
than at high a.mplitudes. Finally, an
equivalent circuit is proposed to account for these new observations.
The cin::uit response is also computed and
compared with the test data. |
|
2, FAST-IMPULSE TEST DA TA |
2. 1. Test Pracedure
In order to cany out fast transient tests with minimumcircuit inductance
a new coaxial test module has been
designed and constructed(7 J. The module is a lowinductance, test
facility i!lcorporating integra] voltage
and current transducers. It allows tests to be carried out on I SkY
arresters with measurements to SOkV at lkV{ns
and 5kA at IOAlns. For the very fast fronts , the source capacitors
wcre arrnnged in a coaxial cOllfiguration to
minimise source inductar.ce.
The inductive effects on the measured values could be further minimised
by measuring thc value of the residual
voltage at the installt of peak current when the rate of change of
current is zero. In this way, a more realistic
representation of the resisti ve behaviour of the arrester can be
achieved independently of the test circuit by
e liminating the effects of dildt on the measuring system, thus allowing
comparison between different results. |
|
2.2. Effect of Current ShApe
Tests were carried out applying impulses of similar amplitude but
of different waveshapc in order to determine the effects on the voltage-current
characteristics of the arrester. Two capacitor banks with a similar
capacitance value but of different internal inductances were used
in order to produce the desired current impulse shapes. It was obser
ved that after the voltage front (-4Ons) the shapes of the resul ting
residual voltage waveforms are quite dissimilar, the difference becoming
more notablc at higher dischargc currents .
Because the residual voltage has not yet reached its penk value according
to the voltage-current curve, it is
continuing to increase at a slower ratc t.han the initial jump due
to the high non-linearity of the material. |
|
The arrester branch inductance produces a voltage fall which is
proportional to the ralC of change of currem which
become smaller towards the current.peak vlaue.
The residual voltage at peak current was found to vary lillie for
the twO sources for similar discharge current
amplitudes, suggesting a unique resislance-vo ltage curve for the
zinc-oxide material for this range of fast fronts. |
|
2.3. Effect of Current Amplitude
Tests at ir.creasing charging voltage of the capacitance bank were
conducted in order to check the voltage-current
characteristics of various distribution z.inc-oxide surge arresters.
Figures 1.a to 1.d show voltage and current records measured on a
15kV rated surgc arrester.
These tests revealed, for the first time, thalthe time_to_ peak of
the current decreases as the amplitude of the
current increases. Furthermore, it can be seen that the voltage traces
do not show the initial voltage overshoot
previously_associated with z.inc-oxide tests even for these voltage,
risetimes of not more than 400s. In the case of
Ihe Jow-inductance source, it was fowtd that the time-tocurrent peak
decreases from about 2.SIfS at a peak
current of 100A to approximately 1.5 us at 5kA.
Figure 2 shows the relationship between time-to-curre nt peak and
the amplitude of the cur-ent peak for three
ar resters of different manufac tUIe (arresters A, B and C).
It can be seen that the time to current peak reaches a constant value
(-1.5jJs) when the arresters are operating
in the high-conduction regime (above -lkA). In Ihis region, most of
the intergranular layers have broken down forming many current paths
through the m!uerial. 1t is in this region that the resistance of
the zinc- oxide grains becomes the main limiting (actor for conduction.
In contrast, the low-conduction regime shows time-Iocurrent peaks
ranging from aboui 1.6).15 for CUTTent
amplitudes of approximately 500A to lime-to-peaks in the order of
5~lS for currents less Ihan l00A. |
|
3. STATIC YOLTACE-CURRENT CHARACTERISTIC OF ZlNC-OXlDE
SURGE ARRESTERS |
The V-I characteristic may be divided into three regions;
the pre-breakdown region, the breakdown region and the upturn region.
The pre-breakdown region of the characteristic is determined from
direct or powerfrequency
voltages. The amplitude of the applied voltage is such that the resulting
current is usually less than 10mA. Cominued application of current
amplitudes greater than this value can result in excessive heat dissipation
which may lead to premature ageing and thermal runaway of the arrester.
Consequently, for characterisation where currents are in the hundreds
of amperes and into the kilo-.lffipere range, impulse currents are
applied. For current up to about 500A, switching impulses may be used.
For characterisation in the kiloampere range, lightning or fast impulses
are used because of their lower energy content.
|
|
4. M ULTIPLE-CURRENT-PATHS CONCEPT |
The experimental results (Figs. 1 and 2) show that zinc¡¤
oxide arrcsters have a dynamic vo][age-currem characteristic. The
salient features of this dynamic behaviour are; i) a dependence of
currcnl time-la-peak on current amplitude and ii) a residual voltage
reaching ils peak value before the discharge CUTTent reaches its peak.
The time-to-current peak is seen to decrease to a minimum as the current
amplitude increases. The simple
standard representation of zinc-oxide surge arresters cannot account
for the decrease seen in the time-IO-peak
of the discharge current. A constant inductive element would produce
a shorter time-ta-peak at a lower current
than at a higher current, which is the opposite of the measured effect.
A non-linear inductance could be used
to simulate this effect, but it is difficult to determine its parameters
from measurements. An alternative simpler
approach is to consider the development of discrete current paths
through the zinc-oxide material, with the number of paths increasing
as the impulse voltage amplitude increases. The low current (when
the level of impulse voltage is low) will flow through the zinc oxide
taking a path where inter granular layers are the easiest to break
down. This path may not be the most direct path through the material,
but as the level of the impulse voltage increases the intergranular
layers which could not preViously be broken down by the lower voltage
amplitude are now bridged. This results in the current paths increasing
in number and becoming more direct.
To simulate this physical relationship would ideally require a circuit
containing a large number of paths. The paths would have differing
characteristics to simulate this current growth as the level of the
applied impulse increased. However, for simulation purposes a model
containing two parallel paths is proposed. It was found that the model
gave good correlation with the laboratory test data. |
|
5. PROPOSED EQUIVALEJT CIRCUIT |
he proposed equivalent circuit is shown in Figure 3. It
comprises two series sections; one to represent the
resistance of the zinc-oxide grains (Rgrai,,) and the selfinductance
(Lbcd) due to the physical size of the arrester
body and a parallel network to represent he properties of the intergranular
layers. One branch of the network
carries the high amplitude discharge current, so that the branch has
a highly non-linear resistance Rg and a low
value inductance Lc:. The second branch has a linear resistance R~
and a higher value inductance Ld to
account for the delay in low-current fronts and the multiple --current
path concept. A capacitive element Cg
to represent the arrester shunt capacitance was also included in the
equivalent network.
|
|
6. CIRCUIT SIMULATION |
The laboratory test circuit incorporating the equivalent
circuit representation of the arrester described in the
previous section is represented in Figure 4. The circuit parameters
were determined from laboratory
measurements or where applicable obtained from manufactures data.
The stray components were estimated on the basis of the test circuit
physical arrangement.
Figure 5 shows voltage and current oscillograms from the circuit simulations
representing the residual voltage
of the arrester at the point of measurement (R:ccL,~J and the diSCharge
current through the arrester. Examination
of the records show thai the current shapes give good agreement with
lhose obtained in the laboratory (Figwe
I). Both Ihe labormory lest resulls and .he simulations show simi
lar lime-to-currenl peaks for a similar current
amplitude, Furthermore, for current amplitudes in excess of J leA
the res idual voltage al currenl peak shows good
agreement between laboratory tesl results and simulations . |
|
|
7 . CONCLUSIONS |
The laboralory tests showed thai the time-la-peak of
the arresler discharge current is dependent upon the
amplitude of the current. For low amplitude currents, the arrester
exhibits longer time-Io-peaks than at higher
amplhudes. |
|
Since a single non-linear resistance function cannot
reproduce the dependence of curren! time -to-peak on
current amplitude, an equivalent circuit for zinc oxide has been proposed
to simulate the observations made in
the laboratory, namely the effect of a decreasing time-tocurrent peak
as me-amplitude of the c urrent increases.
The equivalent circuit is based upon the assumption that multiple
current paths are formed through the zinc oxide
and that the number of paths increases and the path lengths decrease
as the current amplitude increases. The
results from the simulations show agreement with those obtained in
the laboratory. |
|
REFERENCES
1 . Haddad A., Elayyan as.B., German D.M. and Waters R.T.: nO surge
arrester elements with mixed direct and 50 Hz voltages, lEE Proceedings,
Part A, VoU 38, No.5, pp.265-272, 1991.
2. Haddad A., Nayl or P., Metwal ly 1., German D.M., Waters R. T.:
An Improved Non-rnductive Impulse Voltage Measurement Technique for
ZOO Surge Arresters, IEEE Trans. on Power Delivery, Vol.lO, No.2,
pp.778 - 784,1995.
3. Schmidt W., Meppclink 1., Richter B., Feser K., . Kehl L. and Qiu
D.: Behaviour of MO-surge arrester blocks to fast transients, IEEE
TraRs. onPower Delivery, Vol.4, No:l , pp.292-300, 1989.
4. IEEE working group 3.4.11.: Modelling of metal oxide surge arresters,
Transactions on Power Delivery, Vol. 7, No. I., pp. 302-309, 1992.
5. Kim I., Funabashi T., Sasaki R. , Hagiwara T. , Kobayashi M.: Study
of ZnO arrester model for steep front wave, IEEE trans. On Power
Delivery, VoL I I, No.2, pp.835-84 I, 1996.
6. Haddad A., Naylor P.: Finte Element computation of capacitance
networks in multiple electrode systems: application to ZnO surge arresters,
lEE Proc. Science, Measurement and Technology, Vol 14S.No.4 pp.l 29-135,
1998.
7. Haddad, A., Naylor, P., German, D.M., Waters, R.T.: A fast transient
test module for Zno surge arresters, Meas. Sci. Techno!. , Vol. 6,
pp560-570, 1995. |
|
A CKNOWLEDGEMENT
The authors thank Profe ssor RT Waters and DM German for their assistance
and useful technical discussions. |
|
ADDRESS
Electrical Division , School of Engineer ing,
Cardiff Uni versity, PO Box 687,
Cardiff eF2 3m, Wales, UK.
TeL +44 12228759(>1
Fa x; +44 1222874735
S-MAII..: HADDAD@CP.AC,UK |
|
|
|
|
|