Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

642 IEEE TRANSACTIONS ON PLASMA SCIENCE, VOL. 25, NO.

4, AUGUST 1997

Dielectric Recovery of Copper Chromium Vacuum


Interrupter Contacts After Short-Circuit Interruption
Eduard Huber, Klaus Fröhlich, Senior Member, IEEE, and R. Grill

Abstract— The recovery of a vacuum interrupter gap after sive laboratory installation for short-circuit switching tests. In
short-circuit interruption was measured by application of an addition, these tests do not contain a quantitative result as they
overshooting transient recovery voltage (TRV) several tens of provide only “failure” or “nonfailure” information.
microseconds after current zero. Copper chromium contact ma-
terials were employed varying in composition (25 and 50% In order to improve this situation, a test procedure was
chromium content), gas content, and production method. The gap developed with the goal of delivering results which also
failure was either pure dielectric or it was dominated by a signifi- provide the manufacturer of contact materials with quantitative
cant postarc current. Therefore, postarc current phenonema were information as a development tool.
experimentally investigated focused on the relationship among the The primary basis of this method is the measurement of the
postarc current, the power frequency current amplitude, and the
gap length. It was found that two postarc current maxima exist: gap recovery both a few microseconds and a few hundreds
the first strongly dependent on the power frequency current, and of microseconds after current zero during an interruption of a
the second on the field strength. A correlation among postarc sinusoidal short-circuit current wave at a power frequency with
current facilitated failures, the ultimately dielectric recovery, and peak values up to 75 kA. In addition, the measured postarc
the erosion rate of the material was found. Strong indication is current after the short-circuit interruption was investigated in
given that all of these effects are dominated by the metal vapor
pressure rise given by the constricted rotating arc. A significant detail, suggesting that the postarc current can be considered
influence of the material properties can be drawn from these as a guide for the interrupting capacity [9].
experiments, allowing a good estimation of the capability for The results of the experiments are presented and discussed
short-circuit current interruption, thus providing a useful tool in this paper. Furthermore, they are correlated with metal-
for material development. lurgical aspects for the various materials which were tested.
Index Terms— Dielectric recovery, high current, interruption Conclusions on the material characteristics such as live en-
performance, postarc period, test method, vacuum arc, vacuum durance and switching capability are drawn.
contact material.

II. EXPERIMENT
I. INTRODUCTION

I T is a well-known fact that the interrupting capability


of vacuum interrupters depends strongly on the contact
material. A very extensive summary of the influence of various
A. Setup
As a rule, one loop of a power frequency current of 60
Hz with a peak value of 48 kA was applied to the contact
materials including the effect of additives is given in [1]. system. Contact separation was set such as to obtain an arcing
For medium-voltage interrupters, copper chromium mixtures time of 5 ms, the contact distance being 4 mm at current
turned out to be superior compared to other known materials interruption. The contacts were opened by a pneumatic drive
for the majority of the switching cases, such as interruption reaching a speed of 1 m/s after 1 ms delay time. The transient
of small currents, load currents, and short-circuit currents. Of recovery voltage (TRV) was chosen either with a rate of rise
course, the capability of a contact of the copper chromium (RRRV) of 0.7 kV/ s or with 30 kV/ s. The TRV was always
family depends strongly on the geometry and the material overshooting, i.e., at each interruption, a restrike was obtained
characteristics, such as the ratio of mixture and method of giving a quantitative value for gap recovery. In order to prove
production. Optimal solutions normally are found empirically the reproducibility for each type of material, two test series
by the various manufacturers and extensive switching tests. were carried out, employing two identical sets of the specific
In particular, to optimize material for a high capability of test material. The test circuit for this procedure was a standard
short-circuit interruption requires very cost-effective tests as synthetic current injection circuit (Weil–Dobke) as is normally
the latter usually are time consuming and require an expen- used in high-power laboratories. The same test circuit was
used for postarc current measurement using the 50-Hz power
Manuscript received April 1, 1996; revised January 6, 1997. The ex-
periments in this work were performed at the “Bundesforschungs- und frequency current amplitude and the gap length
Prüfzentrum,” Vienna, Austria. as a parameter.
E. Huber and K. Fröhlich were with the Technische Universität Wien,
1040 Vienna, Austria. They are now with the High Voltage Laboratory,
ETH-Zürich, CH-8092 Zürich, Switzerland (e-mail: huber@eeh.ee.ethz.ch; B. Test Procedure
froehlich@eeh.ee.ethz.ch).
R. Grill is with the Metallwerk Plansee, Reutte, Austria. Prior to each test series, each test gap was preconditioned by
Publisher Item Identifier S 0093-3813(97)05332-0. six diffuse switching arcs with 10 kA (peak value). After this,
0093–3813/97$10.00  1997 IEEE
HUBER et al.: RECOVERY OF VACUUM INTERRUPTER CONTACTS 643

TABLE I
FEATURES EMPLOYED MATERIAL
OF THE

depending on the withstand capability of the contact system,


up to 20 breaking operations at 48 kA (peak value) were (a)
performed, alternatively changing the RRRV from 30 and 0.7
kV/ s. The value of 48 kA was chosen so as to operate at
the interrupting limit of the investigated contact gaps; this
means that in any case, the switching arc was constricted. The
RRRV of 30 kV/ s was intended to give insight into the gap
conditions within microseconds after current zero, whereas the
RRRV of 0.7 kV/ s should give information on how the gap
would behave during a short current interruption according to
IEC Publication 56.

C. Test Specimen
As a test specimen, standard vacuum interrupter tubes with (b)
a spiral contact system providing a radial magnetic filed for Fig. 1. Restrike values depending on the test number for (a) 25% and (b)
arc rotation were employed. The contact geometry of all 50% Cr content. RRRV = 0.7 kV/s.
investigated specimens was the same. According to the goal of
the investigations, the contact material was copper chromium,
varying in ratio of mixture, residual gas content, and different
production procedure (see Table I).

III. OBSERVATIONS AND TEST RESULTS

A. Dielectric Behavior After 0.7-kV/ s RRRV Stressing


Fig. 1(a) and (b) indicates the dielectric breakdown volt-
age of the gap after applying a TRV of 0.7 kV/ s for the
various materials. The lines in the figures connect the lowest Fig. 2. Mean restrike values for an RRRV of 0.7 kV/s depending on the
breakdown values indicating a sort of withstand level, giving various materials.
an idea of the breaking capability of the gap. Considering
materials with 25% chromium content [Fig. 1(a)], there is a B. Effects Around Current Zero
significant correlation of the content of impurities and the
withstand capability. The higher the degree of impurities, the If a TRV with a 33-kV/ s rate of rise (prospective value)
lower are the breakdown values. Also, a higher degree of was applied after current zero, the breakdown of two groups
impurities causes a faster degree of degradation. Materials was clearly distinct. The first group occurred at a comparable
with 50% chromium content [Fig. 1(b)] show, as a rule, a high value having a position close to the course of the
faster deconditioning, and normally have a significantly wider prospective TRV. The second group occurred at lower values,
scatter of breakdown strength, but have the potential for a which are significantly below the prospective TRV. Fig. 3
high withstand level as well. illustrates this situation schematically.
In Fig. 2, the mean values and the scatter of the breakdown The reason for the latter lower group lies in a significant
values for the various materials are summarized. This illus- postarc current , which depresses the TRV to these lower
trates clearly the effect of impurity for the 25% chromium values. Fig. 4 shows in a measurement how the rise of TRV is
contacts; it also shows the wider scatter for those with 50% influenced by the postarc current. Obviously, the higher group
chromium content. It is obvious as well that material with of breakdown values corresponds to a small postarc current
a high chromium purity (6) produced by infiltration has causing a negligible deformation of the TRV by charge carriers
significantly lower values. in the gap.
644 IEEE TRANSACTIONS ON PLASMA SCIENCE, VOL. 25, NO. 4, AUGUST 1997

(a)
Fig. 3. Influence of the postarc current on the TRV shape.

Fig. 4. Typical shapes of postarc currents and TRV courses. Prospective (b)
circuit parameter: di=dt 
9 A/s, du=dt 10 kV/s, time delay 300 Fig. 5. Collapse of gaps along the TRV course. (a) 25% Cr content. (b)
ns. 50% Cr content.

Group 1 seems to be a clearly dielectric failure of the gap,


whereas the mechanism for the values of group 2 is clearly
correlated to the postarc phenomenon. Therefore, this group
of breakdowns is called “thermal favored failures” in the
following. Fig. 5(a) and (b) shows how the breakdown values
are distributed for the various materials. Dielectric and thermal
influenced failures are clearly distinguishable. For 25% copper
chromium content, the tendency for thermal failures is stronger
the higher the degree of impurity. Furthermore, it can be
observed that for these materials, the occurrence of thermal
favored failures decays with increasing test numbers.
For materials with 50% chromium content, the latter condi- Fig. 6. Portion of thermal favored failure depending on material.
tioning effect could not be observed. However, a significantly
higher tendency for the occurrence of these failure mechanisms two gap lengths are evaluated. For 10-mm gap length up to 54
is evident for the infiltrated materials. Fig. 6 summarizes the kA, there is only small influence on the postarc current peak,
ratio between the dielectric and thermal favored failures for but it is strongly enhanced if the current is increased to 68 kA.
the various materials. For a smaller contact gap of 7 mm, this dependency was not
Obviously, there are contact materials which cause more observed. Also, the postarc current amplitude was smaller and
thermal favored failures, and it is assumed that more postarc the current flow duration was shorter.
plasma must remain in these gaps than in other ones. In order In order to obtain more insight into the structure of the
to learn about the influence of the amplitude of the power postarc current course, high-resolution measurements were
frequency current on the postarc current , the following performed with a power frequency of 68-kA peak value and
experiment was carried out. The power frequency current was a gap length variation of 5, 7, and 10 mm. Fig. 8 shows the
increased in steps, but the and were kept constant. mean values of three tests with the gap length as a parameter.
In Fig. 7, the peak value of the postarc current for a 50-Hz As a rule, the postarc current has two distributed maxima. The
power frequency current of 34, 54, and 68 kA (peak value) and first is significantly higher, followed by a second lower one
HUBER et al.: RECOVERY OF VACUUM INTERRUPTER CONTACTS 645

thus, the arc releases a higher amount of gas and metal


vapor, leading to a higher spatial pressure. Another reason for
the increased pressure is a lower arc mobility for the virgin
contact.
For 50% contacts, the arc mobility is generally higher [4]
due to the lower thermal conductivity of these materials. How-
ever, the lower thermal conductivity may cause, in summary,
a longer cooling period. There possibly is a balance between
easy arc movement and slow cool down of the foot paths which
is decided randomly from switching operation to switching
Fig. 7. Postarc current amplitude versus power frequency current amplitude. operation. This way, a random scatter between thermal and
=
di=dt 8:8 A/s, du=dt = 11:2 kV/s, U^ = 58 kV. dielectric failures, as is observable for 50%, can be explained.
For the phenomena concerning the postarc current, the
following interpretation applies. If the gap length is 7 mm,
the first peak of the postarc current is negligibly influenced
by the power frequency current amplitude (Fig. 7) as the
and before and after current zero, respectively,
were constant. Obviously, no memory of the power frequency
amplitude exists, and the first peak of is related only to
the removal of charge carriers. For a longer gap length,
is generally higher because of the larger amount of charge
carriers in the longer arc, but also, an increase with the higher
power frequency current is the case (see Fig. 7). The reason
for the latter effect is not quite clear, but it is assumed that
a secondary production takes place by space-charge effects
Fig. 8. Postarc current courses versus TRV for different gap lengths
d: d = 5; d = 7; and d = 10 mm, I^ = 68 kA/50 Hz, U^ = 98 kV. preferred by the larger space volume. It is further assumed
that these effects are the reason for the very low reignition
values (compare Fig. 5) which occur when the interrupter is
which is in phase with the TRV oscillation. The shorter the gap
on its very limit. Normally declines almost to zero. But
length is, the smaller is the ratio between the first and second
after a first cycle (Fig. 8) a second much smaller rise appears;
maximum. As the experiment shows, the failure of the gap to
the amplitude is higher the smaller the contact gap is. As the
interrupt is always combined with a relatively high peak value
current is more or less in phase with the voltage, it can be
of the postarc current, as Fig. 8 indicates.
concluded that electron production takes place mainly by field
emission or thermo field emission effects. Normally, this is
IV. INTERPRETATION
the reason for the reignitions occurring at a higher TRV value
Considering the test results, the following hypothesis is (Fig. 5). Further research is necessary to clarify in detail these
introduced. The arc releases metal vapor, the amount mainly effects, in particular, the role of gas density.
dependent on the contact temperature. Contact material pa-
rameters such as
• thermal conductivity, V. CONCLUSIONS
• gas content, and • Vapor is assumed to determine the dielectric recovery
• density of material (as the density is practically equiv- of vacuum gaps primarily after constricted arc activity.
alent for all investigated materials, it is not considered A strongly nonuniform distribution and decay of metal
furthermore) vapor density combined with an inhomogeneous, locally
are responsible for the vapor density and the distribution in enhanced electric field provide a large scatter of recovery
the gap after the current zero. The produced metal vapor level, depending on the arcing behavior.
causes a pressure rise up to atmospheric values [2]. The vapor • As the experiments show, the postarc current normally
is nonuniformly well distributed and highest above the hot has two maxima, the ratio of these values depending on
spot area. Breakdown occurs at locations where the electric the gap length and the interrupted current.
field, vapor density, or both are highest. This way, the arcing • Degassed contact material (2) with 25% chromium con-
behavior, e.g., the arc mobility, decides the breakdown origin tent having a small scatter of restrike values and a small
ultimately by distribution of the heating load into the contact, degree of degradation is assumed to keep the thermal load
leading to more severe gap conditions after current zero or low and uniform even after several switching operations.
not, strongly dependent on the degradation state of the contact This proves again that degassing of the material is essen-
system. tial, as is evident for 25% material. In the case of high
For the material with 25% chromium, it seems to be purity, it is a more reliable material. Although material
primarily a matter of gas content. Obviously, these contacts with 50% Cr content seems to have the potential for a
erode faster than those with a lower gas content [see Fig. 1(a)]; higher recovery level, this is comparable to random low
646 IEEE TRANSACTIONS ON PLASMA SCIENCE, VOL. 25, NO. 4, AUGUST 1997

restrike values. The origin of these low values and their Eduard Huber was born in Horn, lower Austria, in
elimination should be the subject for further research. 1968. He studied electrical engineering at the Tech-
nical University of Vienna, leading to the degree of
• Measuring the gap recovery microseconds after current Dipl.-Ing. in 1993.
zero as well allows deep insight into the material quality, He was with the Institute of Switching Devices
thus providing a powerful tool for material development and High Voltage Technology, Technical University
of Vienna. He is now with the ETH-Zürich. His
regarding the short-circuit interruption capability of the field of research is in vacuum interrupters contact
contacts. materials testing.

ACKNOWLEDGMENT
The authors gratefully acknowledge the support with exper-
imental components from the Rensselaer Institute and ABB
Switzerland. They wish to express their thanks to Prof. A.
Greenwood and Prof. M. Glinkowsky, as well as to Dr. L.
Klaus Fröhlich (M’85–SM’88) was born in
Niemeyer for the stimulating discussions. Salzburg, Austria, in 1945. He studied electronic
engineering at the Technical University of Vienna,
REFERENCES leading to the degree of Dipl.-Ing in 1972, and
received the Ph.D. degree in electrical power
[1] P. G. Slade, “Advances in material developement for high power engineering in 1976.
vacuum interrupter contacts,” IEEE Trans. Comp., Packag., Manufact. From 1973 to 1979, he was a Scientific Staff
Technol. A, vol. 17, Mar. 1994. Member at the Institute of Switchgears, Technical
[2] B. Gellert, E. Schade, and E. Dullni, “Measurement of particles and University of Vienna. In May 1990, he became a
vapor density after high current vacuum arcs by laser technique,” IEEE full Professor at the Technical University of Vienna,
Trans. Plasma Sci., vol. PS-15, Oct. 1987. where he was head of the Institute of Switching
[3] E. Dullni and E. Schade, “Recovery behavior of vacuum circuit breakers Devices and High Voltage Technology until 1997. In May 1997, he became
at the current interruption limit,” in Proc. 14th Symp. Discharges Elect. a Professor at the ETH-Zürich.
Insulation in Vacuum, 1990. Dr. Fröhlich is a member of CIGRE Study Commitee 13 and the convenor
[4] E. Dullni, “Motion of high current vacuum arcs on spiral type contacts,” of the newly founded CIGRE Working Group 13.07 on controlled switching.
IEEE Trans. Plasma Sci., vol. 17, Dec. 1989.
[5] , “Investigations of high-current interruption of vacuum circuit
breakers,” IEEE Trans. Elect. Insulation, vol. 28, Aug. 1993.
[6] F. Unger Weber, “Wiederverfestigung des Hochstrom Vakuumbogens
bei höheren Spannungen,” thesis, TU Braunschweig, Germany, 1988.
[7] D. Heyn, “Untersuchungen zum einfluß des kontaktwerkstoffes auf das
löschverhalten im vakuum,” thesis, TU Braunschweig, Germany, 1990.
[8] M. Glinkowski, “Numerical solution of high current vacuum arcs,” Proc. R. Grill was born in 1961. He completed education in metallurgy at the
SPIE, vol. 2259, p. 153, 1994. Berg- und Hüttenschule, Leoben, Austria.
[9] S. Yanabu, Y. Satoh, T. Tamagawa, M. Honma, and E. Kaneko, “Post He has been Head of the Developement Group for Composite Materials
arc current of vacuum interrupter,” IEEE Trans. Power Delivery, vol. at Plansee AG, Reutte, Austria, since 1989. His field of interests are high-
PWRD-1, Oct. 1986. temperature materials and composite materials for electrical applications.

You might also like