Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
9, SEPTEMBER 1969
Abstract-The energization of a large capacitor bank with vacuum mechanical contact bounce, or both. This short-lived interruption
interrupters coupled with the simultaneous failure of a load trans- may persist up to several hundred microseconds before the cur-
former on a radial feeder prompted this study. The study reveals a rent is reestablished.
source of overvoltage accompanied by sudden and severe voltage Such interruptions and subsequent reignition, all on a high-
reversals at the transformer terminals. This phenomenon is char-
acterized by the momentary interruption and subsequent reignitions frequency base, can give rise to severe overvoltages and high rates
of the two predominant transient high-natural-frequency oscillatory of change of voltage on associated equipment. These observa-
capacitor closing currents. The momentary interruptions take place tions are influenced by the distributed circuit parameters, lines
at numerous higher natural-frequency current zeros, which occur and cables, as well as the randomness in the three-phase closing
during the short interval between the prestrike of the arc and the operations.
metal-to-metal closure of the contacts and at the first lower natural-
frequency current zero, which occurs either during the short interval RELATION TO PREVIOUS STUDIES
between the prestrike of the arc and the closing of the metallic con- The phenomena revealed, although new to the science of cir-
tacts, or during the contact separation- due to mechanical contact cuit interruption, have been postulated earlier as the cause of
bounce, or both. Analysis is in two parts, field tests and PACE overvoltage on transmission lines and defined under arcing
computer analysis. The computer studies duplicate field test results
and pernit the evaluation of various combinations of overhead and ground phenomena [1], [2]. In these earlier papers, the inter-
underground circuit parameters and corrective measures. mittent arcing across line insulators has been postulated on the
basis of the currents being interrupted at the natural current
zero of the high-frequency phenomena, and restriking on the
natural frequency recovery voltage [2, Theory I, Table I(a)].
INTRODUCTION This is the case reported here for the first time as associated with
IN AN ADVANCED program of system planning, the Depart- a circuit interrupter. In [2, Table I, case II ] is the well-known re-
ment of Water and Power (DW&P) of the City of Los Angeles striking phenomenon of early power breakers in which both the
has embarked on a program of adding shunt capacitor banks to leading power factor interruption and the reignition took place
the 34.5-kV distribution system to improve system voltage and on the 60-Hz power frequency basis. The present revelations
supply the kilovars required during the system peak load were made possible by the introduction of an interrupter so
periods. In keeping with the advanced technology of capacitor powerful that it is capable of interrupting the circuit with very
switching equipment, vacuum switches were introduced for small contact openings. When these openings occur during the
capacitor switching. During routine preservice testing, a trans- "closing" operation and are associated with the numerous cur-
former on an industrial site, one circuit mile from the capacitor rent zeros of the high-frequency closing transients, the resulting
substation, was severely damaged (Fig. 4). voltage transients at the time of reignition of the closing con-
This is a report of the extensive field testing and computer tacts gives rise to a family of phenomena which, on most counts,
analysis of the cause of this transformer failure; it comprises the are highly undesirable.
technical concepts of capacitive switching where the rates of
change of voltage and current are extremely high. None of the THE PHENOMENON
equipment used for these tests was designed, manufactured, or When a large ungrounded, uncharged, capacitor bank is
sold by the authors. In these studies, high-speed cathode-ray energized in a substation fed by lines or cables, the closing of the
oscillographic (CRO) records of switching operations, coupled first breaker pole is uneventful since no circuit has been estab-
with analytical treatment, and expanded through a PACE com- lished. Closing the second pole at the moment of the maximum
puter study, have revealed an important aspect of such capacitor value of line-to-line voltage between the two phases causes both
switching that is new to switching technology. The phenomenon phase voltages to collapse to zero, the uncharged potential of the
is believed to be unique to the vacuum interrupter under closing capacitors. As a result, traveling waves of voltage and current
operations in which an interruption of the circuit can occur at radiate simultaneously from the capacitor on lines and cables
one of the several current zeros of the two predominant high- associated with the two phases affected. These conditions are de-
frequency closing currents either following the prestrike and picted in Fig. 1 where all four linles have a surge impedance of Z1
prior to contact make, or following contact separation due to ohms. The voltage waves will all be alike and have a magnitude
of half the crest value of the line-to-line voltage. The waveshape
will be exponential with a sheer front as shown, and expressed as
Paper 69 TP 83-PWR, recommended and approved by the Switch-
gear Committee of the IEEE Power Group for presentation at the
IEEE Winter Power Meeting, New York, N. Y., January 26-31, e .= Em(L-L) _2t/Zic
1969. Manuscript submitted September 16, 1968; made available for (2
-printing May 29, 1969.
E. W. Boehne is with I-T-E Imperial Corporation, Philadelphia, where C is the capacitance in farads per phase to neutral of the
Pa. 19130.
S. S. Low is vith the Department of Water and Power, Los Ange- bank. When C is in microfarads, time will be in microseconds.
les, Calif. 90054. The time to half value T/2 will be 0.346Z1C. If Z1 = 30 ohms and
BOEHNE AND LOW: SHUNT CAPACITOR ENERGIZATION WITH VACUUM INTERRRUPTERS 1425
Fig. 1. Origin of higher natural frequency (see Appendix II): Fig. 2. 91/4-inch transformer bushing rod gap
conditions microseconds after closing of second pole of three- similar to type that flashed over during cap-
phase switch upon large ungrounded, uncharged capacitor bank. acitor bank energization.
C = 12 pF, the value of T/2 is 124 /ss. If Z, = 400 ohms, the half-
time will be increased to 1660 ps.
These waves reflect and refract at junctions of lines and cables 0
-0
and at transformer terminals, to create what is here termed the I-
o
higher natural frequency of the observed phenomena (Appendix IDsIn 150
o
II). In some cases of composite circuits consisting of cables and <G 140
lines in series, the phenomena can be amplified to give rise to o 130
extremely severe transformer stresses. The possibility of the
higher natural frequency being in resonance with the internal
natural frequency of the terminal transformer cannot be over- <:100
looked. The above high-frequency oscillation due to the multiple
reflections on the radial feeder to the load transformer is super-
imposed upon a lower natural frequency of the capacitor bank
oseillating with the source inductance and described in some de-
0
0
7O-----_717
80
hL
I I' . .. L is | .
_
____
_ _ _
___
I_
_ xsw:X:i:ii{_
-
.
(a) (b)
Fig. 8. Typical transient overvoltages produced during capacitor energization witth vacullm switCih caahile
of interruptiorn during closing. (a) Voltage measured at capacitor buis. (b) Voltages measurled
simlnltaneoutsly at open end of test line. Second high-frequency voltage oscillationi was iiitialed bY in-
terruption and reignitioni of capacitor current at first current zero of low-freqimenvY oscillatioii.
(a) (b)
Fig. 9. Transient overvoltages as in Fig. S. (a) V'oltage measured at capacitor bus. (b) Voltages measured
simultaneously at open end of test line. Rleignition following interruption timed by chaiice so that voltages
neutralize to cancel original high-frequeincy voltage of initiating transienit.
(a) (b)
Fig. 10. Transient overvoltages as in Fig. 8. (a) Voltage measured at capacitor bus. (b) Voltages measured
simultanieously at open end of test line. Reignition following interruption timed by chance so that only
momentary distortion of high-frequency wave occurs.
BOEHNE AND LOW: SHUNT CAPACITOR ENERGIZATION WITH VACUUM INTERRUPTERS 1429
Following the field study, a computer program was carried out CORRECTIVE MEASURES
to duplicate the field results and to explore the effects of numer- Fig. 14 demonstrates that a 15-ohm resistor, the surge im-
ous circuit parameters and several protective measures. This pedance of two 30-ohm surge impedance cables in parallel, will
work was carried out on the PACE computer, which accepts the essentially eliminate the undesired phenomena when first in-
differential equations of the entire circuit, solves for any desired serted in the capacitor circuit. The short circuiting of this re-
quantity, and plots up to eight time traces. It permits manual or sistor produces results just as mild. This practice, familiar in
automatic switching of any portion of the circuit as desired, and circuit switching, would increase considerably the complexity of
will yield the results of this switching on the pacographic traces. the vacuum switch device used to energize large capacitor banks.
This method of analysis is exceptionally well adapted to momen- To reduce the rates. of change of voltage at the terminal trans-
tary changes and explorations of the switching sequences and former, a 1.75-AF protective capacitor was considered to be in-
intervals, inasmuch as the graph speed can be reduced far below
the real speed of the phenomena. Figs. 12-17 exhibit samples of
stalled, line to ground, at the terminal of each of the three phases
of the terminal transformer. Fig. 15 displays the effects of this
the 150 solutions carried out. capacitor-reduced voltage and rate of change of voltage at
Since the circuit under study is ungrounded, the first and most terminal. A comparison of Fig. 15(a), where no protective
important phenomenon develops when the second pole of the capacitor is used, with Fig. 15(b) and (c), where the protective
energizing switch closes. Upon the second pole closing, a single- capacitor is present, will reveal its value in reducing internal
phase circuit is created through two phase capacitors and two stresses in the terminal transformer. This expedient might be
source inductances (Fig. 1). Therefore, a single-phase computer further explored for cases now in existence that exhibitthe
representation was employed. The value of the driving voltage phenomena reported here.
represented essentially 50 kV, the crest value of the phase-to-
phase voltage of the 34.5-kV circuit. The speed of the phe- EFFECT OF DISTRIBUTED CIRCUITS FROM CAPACITOR Bus
nomena was such that 60-Hz representation was not warranted. All the studies above, both field and analog, have considered
A comparison of field oscillograms and computer results will con- compound service and radial feeder consisting of cables, adjacent
firm the wisdom of this choice. to the capacitor bus, that enter open lines to both the source and
The distributed circuits were simulated with 7r and T sections to the load transformer. Of the three case studies, this combina-
designed in keeping with the known constants of the cables and tion proved the most severe for the terminal transformer. When
lines used in the field and studied oscillographically. Because of only cables were employed the terminal voltages were reduced
the lumped circuit representation, common to all analog treat- and the rate of change of voltage markedly reduced. This im-
ment of distributed circuits, the results will fall short of display- provement was at the sacrifice of heavier duty on the energizing
ing the true severity, particularly with respect to the initial
wave impact upon the transformer as discussed in Appendix II.
switch. This, case approaches the severity of the back-to-back
Fig. 16 gives one example of this case.
switching requirements.
In addition, the rate of change of voltage at the terminal trans- When the cables were replaced with open lines, the duty on
former will be softened due to presence of a lumped capacitor, both the transformer and the switch were reduced. One such
which is part of the last ir section of the appended line. Figs.
12-17 display the 12 records made on each test, labeled with re-
sample is shown in Fig. 17. This example suffers from the
curacies associated with the lumped representation of tWhe inac-
open
spect to the designations of Fig. 6. line for the single-phase circuit involving two-phase capacitors.
1430 IEEE TRANSACTIONS ON POWER APPARATUS AND SYSTEMS, SEPTEMBER 1969
-1250 -1250
-2 50S -2500
0 -0
VJATTAGE IN CLOSING RES.
W-2 E IIJidt WATTAGE IN CLOSING RES.
R = 1.0r
WAT T WATT
SECONDS 1250 SECONDS 1250
--
KV5i.i
V/ 250S
R 0E5001.0
C 0 j'D NOTE; MAGNITUDE IS KV/MICROSECOND NOTE: MAGNITUDE IS
-. v RATE OF CHANGE RATE oF CHANGE
(a) (b)
1250 F REIGNITION
-
1250 REIGNITIINTIIIN
-2500 -2500
(C) (d)
Fig. 12. Samples typical of 150 PACE computer runs simulating case of DW&P field test. (a) Traces of
normal closing; no interruption. (b) Same records allowing interruption at first current zero 'of low-
frequency transient current at P; no reignition. (c) Same as (b) but allowing reignition 125 us after interrup-
tion at P. (d) Same as (b) and (c) but allowing reignition 500 ,us after interruption at P.
BOEHNE AND LOW: SHUNT CAPACITOR ENERGIZATION WITH VACUUM INTERRUPTERS 1431
FREQUENCY OSC
100I S INTERRUPTION I5
;
250, JS INTERRUPTION
FEUNp 15OpS- -
LOW Q/i lt
FRE~QUENCY OCC
e TRAPPED VOLTAGE ON
C 3~ IIOfls*j ~~~~~~~I -
CPACITOR
RECOVERY TRANSIENT
'1ll 1111
FOLLOWING INTER.AT P
_2 200 JS INTERRUPTION T 450 JS INTERRUPTION
_ 0 TO; ~ ~~
-, ~ P.
jeP- =
e3 ~ ~
1i PACE
COMPUTER
TRAPPED
3ON CAPACITOR7 VOLTAGEIM1,il111
REPRESENT-
ATION OF
IIPHENOMENA
~
0 ~ O T CURRENT,
IZERO
INT~ERVA L
(a) (b)
Fig. 14. Subdued overvoltage occurring when 15-ohm resistor is introduced in series with closing switch.
(a) Interruption at P; no reclosure. (b) Interruption at P; reignition 150 ,s later at Q.
E12
IN KG PHASE -TO -PHASE
5as, CLOSING SWITCH CURRENT 110250
____
,.CLOSING 7; ISWITHCURRENT --
JJ 501250 II... CLOSING SWITCH CURRENT
-1250 pJ JI -1250
-2500
-2500
lOoS5-.
-2500
100iS-
100 uS
RATE OF CHANGE OF TERM. VOLT.
-
100
-so
e2
- 5o
TE-- , e 2 r0
-50
-
2500 p CLOSING SWITCH CURRENT 2500 CLOSING SWITCH CURRENT 2500 ^CLOSING ..SWITCH CURRENT
Y U2 50 - G , 0\
- 1250 1250 -1250 P
-2500 -2500 -2500
TEa R t2RdtO'
jiI WATTAGE IN CLOSING RES. R (IUdt WATTAGE IN CLOSING RES. MI R(IOdt WATTAGE IN CLOSING RES.
WATT WATT
JOR- I.o'A IOdt
125 WATT _
SECONDS 5 SECONDS 1250
SECONDS 1250
_ 0- 0 --
RATE OF CHANGE OF TERM.VOLT. RATE OF CHANGE OF TERM.VOLT. RATE OF CHANGE OF TERM. VOLT.
.de3 0 3 1D
di _E _dtt}-
'iS->
KV/MICROSECOND NOTE: MAGNITUDE IS
RATE OF CHANGE
KV/MICROSECOND NOTE:MAGNITUDE IS KV/MICROSECOND NOTE: MAGNITUDE IS
RATE OF CHANGE RATE OF CHANGE
Fig. 16. Evidence of reduction of overvoltage phenomena when supply and radial feeder are cable circuits
only (no overhead line sections). (a) Closure only; no interruptions. (b) Interruption at P; no reignition.
(c) Interruption at P; reignition at Q, 150 Ms later, showing lower transformer voltage but more intense
switch current on both closmg and reignition.
- 0-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
RfIdt WATTAGE IN CLOSING RES. PI dt WATTAGE IN CLOSING RES.
Jo0
VII
R - 1.0
WATT WATT
SECONDS 1050 SECONDS 1250
(a) ()
Fig. 17. Cables of Fig. 16 replaced with overhead lines-terminal voltages essentially same; voltage rates
of change increase; duty on closing switch reduced. (a) Single closure; no interruption. (b) Interruption at
P; closure at Q, 150 As later.
BOEHNE AND LOW: SHUNT CAPACITOR ENERGIZATION WITH VACUUM INTERRUPTERS 1433
The short, highly damped phenomenon in trace II (el), and subject requires more study with more highly refined computer
trace V (Io) are questionable, and are perhaps the result of the representation or instrumentation if the vacuum interrupter can-
local circuit of lumped parameters in oscillation with the capaci- not be cured of its eagerness to interrupt at all current zeros in
tor bank being energized. More field studies of this case are war- its arcing path.
ranted to evaluate damping characteristics of the higher natural
frequency over open lines. MECHANICAL FORCES ON CLOSING CONTACTS
The extremely high calculated rate of rise of current of 1960
EFFECT OF SOURCE BUS IMPEDANCE A/us (Appendix II) gives rise to a high rate of rise of wattage,
An examination of the computer graphs with particular ref- since this current must be drawn through the cathode and anode
erence to the voltage of the source bus, labeled eo and shown at drop of the vacuum switch. Since this arc drop increases with
the top of Figs. 12, 16, and 17, reveals a large voltage dip as- current [121, it follows that, before the prestriking contacts can
sociated with the energization of the 5400-kvar capacitor bank close mechanically, there is a considerable release of energy
some distance away. This comes about because the source bus is which could create forces in melting metal or ejecting electrons
considered as having small capacitance and is not reduced in that would encourage the contacts to retard their motion. In
impedance by having many lines radiating from the bus. The this manner, more time would be created to permit interruption
computer representation exhibited would be more accurate, at a current zero. The defined motion of the contacts under
since the source bus was essentially a transformer terminal or a conditions simulating actual field conditions should be required
bus of essentially high impedance compared with 12.1 uF per procedure. Current measurements should be made with concen-
phase. This fact will have an influence upon the nature of waves, tric noninductive shunts.
reflected at this bus, which return to the capacitor bank. Future WHISKER HYPOTHESIS OF INCREASING
work should examine the effect of the impedance of the source PRESTRIKE DISTANCE
bus in relation to current zeros in the capacitor switch. In the
above light, the results presented here are believed to be slightly One unusual phenomenon sometimes occurring with certain
pessimistic; however, any source bus ahead of lumped inductive metals in a vacuum and exposed to electrostatic stress is the
impedances is not likely to be large enough to avoid being con- growth of long, extremely small diameter crystalline filaments,
siderably influenced by 5400 kvar on one closing. which rise normally to the contact surface. These have been
termed whiskers. These fine filaments are hardly visible to the
INTERRUPTIONS AT CURRENT ZEROS OF naked eye. Should a breaker closing upon a capacitor circuit en-
HIGHER NATURAL FREQUENCY counter a whisker, it could easily increase the prestrike distance
During the computer study numerous current zeros were ob- and permit an interruption of the type described here without
served early in the growth of the closing current. These were the necessity of a reversal of the contact motion. This phe-
caused by the high value of high-frequency current demanded by nomenon may account for the variation in performance observed
the low surge impedance of the adjacent cables. These current in the initial closing of vacuum switches that have been exposed
zeros were very evident during the field tests. Since the high-fre- to open-circuit voltages for long periods, as compared to tests
qi iency current uses the linearly rising low-natural-frequency cur- made in relatively rapid succession.
rent as an axis, it is easy to show that early current zeros can exist CONCLUSIONS
on the basis of no attenuation of either current, only up to a time
designated by the time 2L/Z, where L is the inductance per phase Supplementing the conclusions presented after the description
of the source and Z is the surge impedance of each of the four of the field tests, the following are pertinent to this application.
radiating conductors from the capacitor bus (Fig. 1). Should an 1) Interruptions of current, no matter how small, during the
interruption occur on the last current zero up to a time 2L/Z, energization of large capacitor banks may be a hazard to ap-
then a current Em(L-L)/Z would have been established in the pended equipment, and should be avoided through proper design
source inductance, but reduced by the degree of attenuation or and application criteria.
damping of the high-frequency current. Should- this attenuation 2) Of the three circuits studied, the compound cable-line
be 50 percent up to this current zero, then a current of Em(IL)/ feeder proved the most severe under the influence of the high-
2Z would have been interrupted in the single phase, line-to-line frequency momentary interruptions during closing.
circuit of Fig. 1. This inductive current would then be required to 3) Two corrective measures to reduce stress on the terminal
seek parallel paths through the terminal transformer, build up transformer were effectively studied: a) the insertion of low-
voltages in the positive sequence capacitances remaining, or ohmic-value closing resistors; or b) the use of small capacitors
seek eddy-current paths in the iron circuits associated with the shunting the terminal transformer. Both proved effective.
source, or all three, and in this manner establish a new natural 4) The higher voltage applications will experience longer pre-
frequency capable of giving rise to some high overvoltages. ignition intervals and hence will have a greater probability of
It was somewhat surprising to observe that when the PACE having current zeros during the interval between preignition and
computer was programmed to interrupt at one of the early cur- ultimate contact closing.
rent zeros described above, some of the highest overvoltages were 5) Some lightning arresters protecting terminal transformers
encountered. 'rhis phenomenon is not unlike the high-frequency subjected to the high-energy surges reported are vulnerable to
pulsating of a dc circuit to create a current zero. It is also closely damage or destruction.
allied to the work of Damstra [9], in which capacitance in APPEN3DIX I
parallel with the nonlinear arc-voltage characteristics can pro-
duce ,agrowig-instability in a decaying current, which leads to a LOWER NATURAL FREQUENCY
premature current zero (heretofore called chopping). The phe- The lower of the two natural frequencies associated with the
nomenon described above occurs on the rising current and is energization of large capacitor banks is initiated with the in-
dictated by the circuit topology and switch characteristics. This stantaneous collapse of the line or bus voltage to zero as the
1434 IEEEE TRANSACTIONS ON POWER APPARATUS AND SYSTEMS, SEPTEMBER 1969
1. 5L (C/1.5)
closing, are so numerous that it is necessary to reduce our con-
This current is 15 percent higher than the second pole closiing, sideration here to the simplest case. Consider the schematic dia-
which is the magnitude of the single-phase case of (1) and has the grams of Figs. 1 and 6 in which the three-phase ungrounded bank
same natural frequency. is located in a radial cable-line feeder as shown. We shall, at this
The restoration voltage that accompanies the above curreent, time, neglect the closing of the third phase and be concerned
in each case, has the general form only with the chronological events that take place in the micro-
seconds following the closing of the second pole, the first pole
Em (1 - cos <) = Em(I - cos (6) having already closed.
At this moment of energization the voltage between the two
This voltage rises from zero to 2Em in a time 7rV/LC or V Ae/ phases collapses to zero, permitting voltage waves, having a mag-
120 seconds, and occurs at the moment of the first current zerco of nitude of Em(L-L)/2, to radiate from the capacitor on all four
the associated transient currents shown above in (1), (4) and (5).
( connected cables. These voltage waves have the form
It follows that should these currents be interrupted at the fi
current zeros, then the capacitors would trap a charge approa Ee=
Em(LL)
m -2t/CZi (7)
2
ing twice the crest of the associated voltage.
This is what has occurred in the cases reported in the pa where C is the capacitance in microfarads per phase, Z, the surge
when the contacts of the vacuum switch permitted an interr impedance of each of the four cables, and t is time in micro-
tion to occur at the first current zero of the lower frequency seconds. Each voltage wave is accompanied by a current wave
cillation. These conditions are described in Fig. 18. of the form
Just as the lower natural-frequency currents and volta
oscillate about the 60-Hz current and voltage as an axis, so Em(LL) - 2t/CZi (8)
high-natural-frequency phenomena (Appendix II) use the lo) 2Z1
BOEHNE AND LOW: SHUNT CAPACITOR ENkERGIZATION WITH VACUUM INTERRUPTERS 1435
As a result, the switch current is twice the wave current of (8). each other within the transformer, the potential to ground is re-
It is also important to note that the actual current from two of duced at the expense of placing a squeeze upon the internal coil
the cables flows into the capacitors while the same total current and turn insulation within the transformer. The PACE computer
flows out into the other two cables as directed by the polarity of results, reported in the text, measured the magnitude of the rate
the voltage. of change of voltage at the tranisformer terminal under the condi-
In the 34.5-kV system of DW&P, where the surge impedance tions of the circuit simulation.
of each cable is 30 ohms, the switch current initially is 1660 am- The above description is normal for the case of a cable-line
peres, while each wave current initially is 830 amperes. The cor- radial feeder terminated with a transformer. When a vacuum
responding wave voltage is initially 30 X 830 or 24 900 volts, switch permits an interruption at one of the current zeros as-
which is half the crest value of 34.5 kV. sociated with the lower natural frequency described in Appendix
When we consider the action of the current and voltage of a I, demonstrated in the field study, and duplicated in the analog
particular cable when the wave encounters a line of surge im- study, it becomes apparent from Fig. 18 that a reignition (fol-
pedance Z2, the majority of the current reflects negatively with lowing an interruption) can occur when the phase capacitors
respect to the incident wave and has a magnitude and shape have been elevated to essentially twice the crest of the line-to-
line voltage and when the cable-line circuit is essentially at zero
voltage. As a result, the voltage waves radiating to the trans-
i Lz1iz i (9) former terminal are twice as large as described above, resulting in
When the four cables of Fig. 1 have the same length, the waves 3.64 X 50 000 or 182 kV to ground at the transformer bushing.
return to the capacitor bank simultaneously. As a result, the Here, there is little or no reduction due to the original voltage on
wave currents permit a complete reversal of the switch current, the line, since the recovery voltage of the system has been lowered
which in turn creates the first current zero in the closing switch. to essentially zero following the interruption. It is at this point of
Reflected current waves of opposite polarity to the original out- maximum voltage separation from the trapped capacitor voltage
going waves are again dispatched into the outgoing cable cir- that reignition is assured. Under these conditions, 364 kV could
cuits where they reflect again from the cable-line junction [see appear across one A winding and at a rate of rise that would es-
(2)] with opposite polarity and return to the capacitor to com- sentially double the former case.
plete the cycle and the second current zero in the closing switch. From the above analysis, it is easy to account for the flashing
These multiple current wave reflections on the cables create of the bushing gap since 182 kV is 6.2 times the normal line-to-
what is here termed the high-frequency-closing transient current. ground crest.
The frequency of this initial cable oscillation in the case of This "high-speed" study of the events which can occur in the
DW&P, where the cable lengths are about 800 feet, is about 188 first two milliseconds following the closing of the second pole, in-
kHz. The origin of this current rises to its first crest of 1660 cluding an interruption and a reignition at the most unfavorable
amperes at about 1960 amperes per microsecond for the case moment, is clearly on the pessimistic side of probable occurrences.
depicted in Fig. 1. On the other hand, further complications associated with the
In the meantime, the initial voltage wave that passed the third pole closing can yield still higher voltages which are not
cable-line junction entered the line with an amplified value of presented here.
The high rate of rise of switch current through the 30-volt arc
L2Z21 drop of the vacuum switch develops an energy release that per-
-2
= , +z el. (10)
[Z1 + Z2]_0 sists until the contacts are closed. Any tendency for this energy to
slow the closing contact will increase this energy to the detriment
When this voltage wave arrives at the transformer terminal at of the contact life.
the end of the line, the wave will essentially double and reflect
back to the cable-line junction. Since the surge impedance of the REFERENCES
cable is so small with respect to the line, the majority of the re-
flected voltage wave that arrives at the junction will again re- [1] J. F. Peters and J. Slepian, "Voltages induced by arcing
grounds," AIEE Trans., vol. 42, pp. 478-493, April 1923.
flect back to the line. These multiple reflections create what is [2] R. D.transients
Evans, A. C. Monteith, and R. L. Witzke, "Power-sys-
caused by switching and faults," AIEE Trans.,
here termed the higher natural frequency. The period of this tem
vol. 58, pp. 386-397, August 1939.
frequency is 4 times the wave travel time of the line. [3] A. Klopfenstein, E. York, and J. W. Kalb, "Effect of preceding
When the cable surge impedance Z, is 30 ohms, and the line bias voltage on switching surge operation of spill gaps and
surge impedance Z2 is 300 ohms, then e2 = 1.82el. This wave at lightning arresters," AIEE Trans. (Power Apparatus and
Systems), vol. 81, pp. 320-331, August 1962.
the transformer terminal is 3.64e1. [4j E. W. Boehne, "EHV surge suppression of interrupting light
But, the cable and line were initially at the potential el and currents with air switches-I: capacitive currents," IEEE
Trans. Power Apparatus and Systems, vol. PAS-84, pp. 906-
opposite to the polarity of the arriving wave. Hence, the line-to- 923, October 1965.
ground potential at the transformer bushing is only 2.64el. The [5] T. W. Schroeder, E. W. Boehne, and J. W. Butler, "Tests and
value of e1 in the DW&P case was 25 000 volts. Hence, in the analysis of circuit-breaker performance when switching large
normal closing of the capacitor, the first peak voltage to ground capacitor banks," AIEE Trans., vol. 61, pp. 821-831, November
1942.
is 66 kV at transformer bushing. [6] E. W. Boehne, "Traveling wave protection problems-I: a
It is important to note, however, that the bushing of the other graphical analysis with special application to the switching of
lines and cables," AIEE Trans. (Power Apparatus and Sys-
phase involved in the initial closing will, simultaneously, ex- tems), vol. 73, pp. 920-928, August 1954.
perience the same voltage, but at opposite polarity. Hence, the [7] ,"Traveling wave protection problems-II," AIEE Trans.
A winding of the transformer will have 132 kY across the wind- (Power Apparatus and Systems), vol. 74, pp. 880-887, October
1955.
ing. The steepness of the above waves differs from shear vertical [8] J. E. Knudsen, "Measurements oftransformer capacity-current
fronts only by the wave sloping action due to line losses and the switching overvoltages andvol.a resulting failure,"
Ingenioren (Internatl. ed.), 2, 4,
no. November 1958.
capacitances of insulators, the bushing, and the impact capaci- [9] G. C. Damstra, "Current chopping overvoltages relation
and in
tance of the transformer winding. As these waves move toward to system parameters," CIGRE, Paper 120, 1964.
IEEE TRANSACTIONS ON POWER APPARATUS AND SYSTEMS, SEPTEMBER 1969
1436
[10] R. Rudenberg, Transient Performance of Electric Power I also recall conducting some tests with an oil circuit breaker in
Systems. New York: McGraw-Hill, 1950. which the pre-arcing at "make" of a capacitive circuit produced ex-
[11] - , Electrical Shock Waves in Power Systems. Cambridge, cessive overvoltages on the supply transformer. These conditions
Mass.: Harvard University Press, 1968. were actually worse than the breaking of the capacitor load.
[12] T. H. Lee, D. R. Kurtz, and J. W. Porter, "Vacuum arcs and This is of particular importance today with the trend toward un-
vacuum circuit interrupters," CIGRE, Paper 121, 1966.
[131 H. C. Ross, "Switching in high-vacuum environment," IEEE derground distribution coupled with the use of the vacuum inter-
Trans. Component Parts, vol. CP-10, pp. 155-169, December rupters.
1963.
[14] L. V. Bewley, Traveling Waves on Transmission Systems. New
York: Wiley, 1933.
[15] E. W. Boehne, "Voltage oscillations in armature windings
under lightning-I1nDn
impulses-I," AIEE Trans., vol. 49, pp. 1587-
1u1uo,
toir
uctober 11-430.
9
T Coxial Shunt
it is stated that when only cables were employed there is a heavier
duty on the energizing switch. Can the authors explain to me the (a)
reason for this? In effect it seems to me that a piece of overhead line
connected at the end of the cable should not change the value of the
peak current through the switch, as I can see also from Figs. 12-17.
3) Have the authors attempted to ascertain whether a small in-
ductance coil permanently connected between the capacitor bank
and the switch can help in reducing both the duty of the switch and
II9
the overvoltages in the system at a reasonable cost? This coil is
beneficial also for the case of back-to-back capacitors switching.
I thank the authors in advance for their reply, and I congratulate (b)
them for their very fine work.
Fig. 21. Circuits for studying reignitions in vacuum interrupters.
(a) Single-frequency currents. (b) Double-frequency currents.
til its amplitude has "rung down" to a very low level, at which time,
as indicated, the trapped voltage is negligible.
Similar tests were conducted with several interrupters in series. I
An oscillogram from such a test is shown in Fig. 23. The upper trace L fOL 1O0L IOOQ L
lrlAUctfAr.c
Manuscript received February 14, 1969.
Fig. 24. High-frequency currents cleared by vacuum interrupter
at given voltage; no interruptions in barred area.
BOEHNE AND LOW: SHUNT CAPACITOR ENERGIZATION WITH VACUUM INTERRUPTERS 1439
is a replica of part of the lower trace with an expanded time scale. Hugh C. Ross (Ross Engineering Corporation, Saratoga, Calif.):
This test was conducted on a two-frequency circuit. It shows that the Although this paper treats a rather isolated case of trouble with a
switches would not interrupt 12.5 kHz with the current shown, but number of questionable assumptions, it does touch on several basic
did clear 300 Hz at the first current zero. studies and facts of interest.
Several thousand pieces of test data were collected in this way. With experience from the large number of similar installations
Different contact materials were investigated and for some of the that have been operating essentially trouble free since as far back as
tests part of the circuit was changed to an artificial line to simulate 1955, and literally millions of large and small capacitor bank close-
the effects of a cable. Presenting such data in a meaningful way be- open operations, it does not appear that the excitement we sense here
comes quite a problem. The method shown in Fig. 24 was chosen. is as serious as the paper describes it.
This chart has logarithmic scales of inductance and capacitance for The paper does contain an excellent recommendation for the solu-
ordinate and abscissa, so that lines inclined at 450 are lines of con- tion of this possible problem, which could occur under the right
stant surge impedance (the quotient of L and C is constant), whereas conditions with a particular type of vacuum contacts. The use of in-
lines inclined at 1350 are lines of constant frequency (the product of rush-limiting resistors on the interrupter, or in some cases small
L and C is constant). From the experimental data an area can be shunt capacitors at the load in question, has proven satisfactory when
blocked out on such a chart: the shaded area in Fig. 24, for which the the economics and probabilities can justify them.
switch will not interrupt either because the current is too high (as Most experienced vacuum switch manufacturers have used re-
determined by the surge impedance) or because the frequency is too sistive insertion devices as a routine part of some of their installa-
high. tions where there is a question of contact life or of inrush oscillations
From the experience gained by these tests I would say that the and resonances creating overvoltages on closing in with or without
problem described by the authors arose as a consequence of some or prestrike reignitions.
perhaps all of the following contributing factors: From tests on other types of breakers, including oil breakers where
resistors were not used, we have seen this same prestrike oscillation
1) closing speed too slow, phenomenon, therefore the paper's claim that vacuum interrupters
2) the bouncing of the contacts, are unique in having prestrike reignition characteristics is not valid.
3) inadequate synchronizing of the closing of the three poles and The inrush current through closing contacts is certainly severe and
the two interrupters of a single pole, the resistors are also helpful to contact life. But the suggestion that
4) the kind of material used for the switch contacts, arc vapor pressure itself tends to force the contacts apart does not
5) the amplifying effect of the particular circuit layout. appear valid in our experience up to 20 000 amperes nns, 60 Hz, and
over 500 000 amperes of high-energy storage capacitor discharge.
From the oscillograms inl the paper it is evident that the frequency Vapor expansion from the small amount created has not been found
of the current being interrupted is about 470 Hz, and that there were sufficient to raise internal pressures enough to overcome normal clos-
a number of opportunities to clear. This indicates slow closing of the ing forces significantly.
switch. On the other hand, it is our experience that closing at 2.5 Magnetic forces from high currents appear to create the major
ft/s, which is not inconsistent with essentially bounce-free operation problems along this line and can generally be compensated.
with the correct contact material and mechanisms design, an 800- With high closing currents, contact losses certainly are higher due
1200 As prestrike is typical. This scarcely allows time for interruption to increases in arc energy and resultant temperatures, localized cur-
of frequencies on this order. Moreover, if such an event would occur rent spots, and erosion from the resultant EI and coulomb values.
the contacts are by then so close together as to be unable to support However, during preignition on closing, generally beginning in the
a significant voltage thereafter. order of 0.010-0.030 inch (100-1000 ,us) before touching, the arc
Were the contacts in the devices described made from a primarily voltage usually remains on the lower to moderate plateaus as as-
refractory material? The propensity of such materials to recover sociated with a very close contact spacing [13].
rapidly a very high dielectric strength, after "whiskers" or other sur- It should be noted that the particular vacuum interrupter de-
face exerescences have been removed, is well known. This is one of the scribed (not of our manufacture) has contacts of a sintered pure re-
reasons that such materials are rarely used nowadays for vacuum fractory metal. Many types of vacuum contact materials do not
circuit breaker contacts. have nearly as much tendency as these pure refractory metals to in-
I feel certain that 5) above was an important contributor to the terrupt at the close spacings where the prestrike is initiated or bounce
unfortunate events described. Perhaps the incident would never have occurs. There has been some tendency to conclude that excessively
occurred but for the circuit configuration. The authors point out high recovery rate and the prestrike or preignition interruption is
that any disturbance created at the switch is amplified by the surge inherent and may not be modified in the vacuum contact, however
impedance transitions, cable-line-transformer. Such a voltage will be the controlled recovery characteristics of the new alloys show this
of short duration because it is quickly modified by further reflec- conclusion would be invalid. The proper contact alloy with controlled
tions from the cable-line junction and the capacitor bank itself. impurity (gaseous or high vapor pressure solid) can modify or mini-
Also, the steepness of the surge at the transformer is probably not as mize current chopping, and slow recovery characteristics for both pre-
great as the authors fear. If the surge impedance of the line is 400 ignition and interruption at close spacings even better than adjusting
ohms and the capacitance at the transformer location is 5000 pF, closing or opening rates. Most vacuum contact manufactuLrers now
the time constant for the rising voltage is 2 As. The fact that damage use these new minimum current chopping alloys for power interrup-
occurred suggests voltage magnitude or repetition, rather than rate tions.
per second. As the authors suggest, the expedient of a small surge Some of these new alloys also have less tendency for rough sur-
capacitor would radically change conditions at the transformer. faces, pits, and points to develop during high-current closing. The
The use of resistors to damp the transients is always a solution, so-called whiskers occasionally associated with improperly processed
but again, as a consequence of our experience, we question the need sintered-type refractory metals such as molybdenum and tungsten
of such an elaboration. Traditionally, reactors have been used with oil were often found to be occluded gas molecule buildups from the
switches, especially for bank-to-bank capacitor switching. This is be- strong "gettering" action of these particular metals, as well as crystal
cause in an oil switch the excessive inrush current flowing through a growth. Properly sintered and degassed units rarely exhibited these
prestrike arc represents a potentially explosive situation. This condi- characteristics, and newer alloys appear to be consideraly superior in
tion does not apply with vacuum. Indeed, reactors should be used these respects.
with caution, for they may reduce the magnitude and frequency of In reference to higher voltage circuit breakers conclusion 4) is in-
the high-frequency current components that are not presently inter- valid since at higher voltages more contacts are used in series with no
rupted to a level at which interruption can occur. greater voltage stress per contact, and therefore the preignition dis-
On the matter of bank-to-bank switching, it would seem that the tance is not increased. With proper closing speeds, the higher ap-
phenomenon described would be still less a problem in that the plications do not therefore have longer preignition intervals and do
blocks of capacitance being switched are usually of comparable size. not have greater probability of current zeros. In fact there may be
Thus when they are connected together, they exchange charge by less preignition time since the voltage stress per contact is usually
both momentarily reversing polarity. The voltage excursions are
thus diminished and the chance of voltage escalation is greatly
reduced. Manuscript received January 29, 1969.
1440 IEEE TRANSACTIONS ON POWER APPARATUS AND SYSTEMS, SEPTEMBER 1969
less than in single contacts. There is more probability of bounce since Finally, I want to congratulate the authors with their masterly
there are more contacts, however with the proper alloys, and proper work and would like to urge them to further investigate the interac-
mechanism to minimize bounce, it should be insignificant. tion of circuit and discharge.
A major value of this paper is that it represents another recom-
mendation for "softer" closing characteristics for certain types of REFERENCES
high loads of capacitance and inductance capable of creating high [16] R. C. van Sickle, "Influence of resistance on switching tran-
in-rush currents or ringing possibilities. Also the low energy but sharp sients," AIEE Trans., vol. 58, pp. 397-404, August 1939.
spike transient suppressing benefits gained by placing small shunt [17] P. Baltensperger, "Switching high-voltage capacitors with air
capacitors across the transient source and/or across the unit to be blast circuit breakers," Brown Boveri Rev., vol. 8, p. 1, 1956.
protected is reinforced by the suggestion in this paper. This method of [181 E. P. Butz, "A revolution in are furnace switchgear," Steel
transient suppression has proved invaluable for many types of appli- Foundry Facts, p. 12, December 1963.
cations connected with high-speed interruption. [19] W. M. C. van den Heuvel, "Interruption of small inductive
The authors are to be commended for a considerable amount of currents in ac circuits," thesis, University of Eindhoven,
The Netherlands, 1966.
thought and applied theoretical analyses of problems possible under [201 W. M. C. van den Heuvel and D. T. J. ter Horst, "Interrupting
certain conditions, combined with actual field tests, and determining small inductive currents with oil filled breakers, part I," Allis-
the possible reasons for this particular transformer failure. In the Chalmers Rev., vol. 30, no. 3, p. 8, 1965.
past, actual field test and life experience has shown many of the [211 -, "Interrupting small inductive currents with oil filled
theoretical problems originally attributed to the vacuum interrupter breakers, part II," Allis-Chalmers Rev., vol. 30, no. 4, p. 26,
to be nullified by other effects, or to be either insignificant or easily 1965.
remedied for the general application.
AIJanuscript received February 17, 1969. Manuscript received February 13, 1969.
BOEHNE AND LOW: SHUNT CAPACITOR ENERGIZATION WITH VACUUM INTERRUPTERS 1441
be limited to the arrester discharge voltage plus the voltage drop of the use of a 40-AH reactor would allow the capacitor bus voltage to
arrester lead and thus the voltage rate of change would be less than decay at a rate defined by a time constant of 2.67 ,us. Should lines of
the case of a rod-gap sparkover. 400-ohm surge impedance replace the cable, the time constant of the
Under these stated conditions, the problem is now addressed to the decay is 0.20 ,As! In both cases, the above rate of collapse would per-
energy level that the arrester would be required to discharge and the sist until wave reflections arrived from either direction. The reactor
seal-off capability of the arrester. It appears that the energy dis- shown by Dr. Lerstrup is 200 ,uH. When working against 400-ohm
charge requirements would be dependent on both the high- and low- surge impedance lines, the minimum time constant of decay would be
frequency voltage waves generated by the capacitor bank. The 1 ,us. This would prevent the high-frequency phenomena from ap-
authors made brief mention of lightning arrester application and pearing on the open end of a 1000-foot line.
seem to prefer circuit parametric changes in the form of closing re- Conditions immediately improve when we include the presence of
sistors or protective capacitors. Do the authors have information the inherent or installed capacitance to ground of or at the capacitor
relative to arrester performance or optimum location under these bus. A small capacitance C2 on the bus prior to energizing a much
conditions, and what was the final or interim solution to the actual larger capacitor Ci through a series reactor L will permit the bus volt-
field installation problem? age to be lowered as controlled by the period T/2 of the new
natural frequency of the tank circuit
T/2 7r-\/LCo
where
Eugene W. Boehne and S. S. Low: The authors are indebted to the Co_ VC,CIC2A
numerous discussers for their contributions and their interesting + C2/
viewpoints. It now appears that the paper has tended to draw to-
gether the body of experience describing interruptions that have Here when L = 40 ,H and Ci = 0.5 IAF and C2 is much larger, the
occurred during the closing operation of numerous types of circuit value of T/2 essentially becomes 14 iAs. This condition would protect
breakers, revealing that the vacuum breaker is not unique in this re- a much longer radial line terminated in a transformer by avoiding the
spect. It is recognized that when such interruptions occur in some buildup of the high-frequency component of the radial line under
circuits they have little significance or importance. However, where shock excitation. Conditions further improve when the total tank-
capacitive circuits are coupled with some overhead lines or cables, circuit inductance is included in which the reactor L becomes a part.
there is the possibility for trouble unless certain precautions are When both lines and capacitors are considered together, the pro-
taken. This fact has brought to light the possible growing importance tection is im-proved. The point to be made here is that the choice of
of this area of surge generation due to the wide popularity of the the reactor size would depend upon the length of line to be pro-
vacuum switch and the growing need to reduce the index of switch- tected, whether it was open line or cable; it would also be dependent
ing surges that reach solid-state circuitry and computer installations, upon the value of the capacitance at the bus at the time the larger
to say nothing of their impact upon terminal equipment. The present capacitance is to be energized. It is felt that this method of correct-
study is motivated toward application criteria that will minimize ing the circuits that are vulnerable to the phenomena described in the
these occurrences and thus take advantage of the excellent interrupt- paper will prove to be the most economical.
ing characteristics of the vacuum interrupter and its low maintenance In contrast, the other three protective measures are: 1) resistor
features. switching, 2) lightning arrester protection of terminal equipment,
At the same time, the study reported here has served to aid in the 3) capacitor protection of terminal equipment.
explanation of numerous apparatus failures that have occurred on Two-step switching resistors will provide an excellent solution, but
energizing large capacitor banks with vacuum switches that have not is difficult to apply to vulnerable circuits in the field. Moreover, it
employed one or more of the protective features described in the creates a complexity of equipment and expense which should be
paper and reviewed here as a result of the comments and questions of weighed with alternative methods.
the discussions. Several discussers have pointed out that arresters should provide
Since the eleven discussions tended to duplicate numerous ques- an excellent means of reducing the voltages at the transformer termi-
tions and, at the same time, make similar contributions, it is our in- nals. This is certainly true, and modern arresters are available which
tent here to divide our rebuttal into five specific technical areas and, provide this protection. Unfortunately, arresters located at the
as closely as possible, to credit each discusser with his contribution. capacitor bank are of little use in controlling the overvoltage at the
Dr. Lerstrup's analysis of the prestrike phenomena is based upon end of feeders radiating from the switching point. In addition,
an assumed closing contact speed of one meter per second, which is should such arresters be initiated in the presence of a large capacitor
three times faster than the speed measured for the particular device bank they are vulnerable to heavy energy surges, which must be
used in the reported case. Separate tests in which a travel record allowed for in selecting the arrester thermal capacity. It should be
monitored the closing stroke clearly demonstrated that interruptions made clear that the arrester has little or no control of the rate of
occurred before the contacts engaged. This characteristic was also change of voltage at the transformer terminal. This is certainly true
confirmed by Dr. Greenwood in his discussion. Of course the overall if the magnitude of the voltage is below the arrester sparkover. Since
performance, including contact errosion, would be complicated by a it is believed that initial transformer failures are the result of turn-
bouncing contact. The field oscillograms in the paper give evidence to-turn or coil-to-coil sparkover, either the rate of change of the in-
of some bouncing later in the voltage traces. These are so indicated coming wave should be modified at the source by lowering the
on the oscillograms. voltage more gently (as described above) or the steep incoming
We are indebted to Dr. Lerstrup for the photographs showing the wave should be modified at the transformer by means of a shunt
series reactors used in each -capacitor leg and suggesting, with others, capacitor.
that this artifice would go a long way to minimize the generation of Shunt capacitors at the transformer terminal, either alone or in
the higher superimposed natural-frequency voltage by permitting a parallel with arresters, provide excellent protection. Standard rotat-
slower collapse of the circuit voltage to the capacitor at the moment ing machine protection will be recognized in this combination [15].
of capacitor energization. Such series reactors are in common use in This form of protection was reported in the paper. Traces III in
suppressing the high tank-circuit currents and voltages during back- Fig. 15 exhibit the remarkable protection provided by a 1.75-,uF
to-back switching. capacitor. In most cases, a 0.5-,uF capacitor would do a splendid job.
Although extremely effective when a similar lumped capacitor is Use of unswitched capacitors on the bus at the point of energiza-
present the bus (as in back-to-back switching), the use of series
on tion is discussed above. Here the unswitched capacitance, working
reactors L in the separate legs of the initial bank to be energized with the series reactor allows a much slower collapse of the sending-
bears close study and analysis. When the capacitor bus is sandwiched end voltage. The rate of collapse of sending-end voltage defines the
between incoming and outgoing lines or cables of surge impedance shape of the front of the traveling wave that radiates to the terminal
Z the collapse of voltage is controlled by the time constant 2L/Z. transformers. For example, when Em collapses to zero at the ex-
Hence when cables of 30-ohm surge impedance radiate from the bus, ponential rate f-ZI12L the frontof the wave that leaves the capacitor
bank has the well-known shape of e(wave) = Em(l - &-Zt/2L).
As a result, it seems more,expedient to correct the problem at the
Manuscript received June 20, 1969. source by providing a small unswitched capacitor on the bus where
1442 IEEE TRANSACTIONS ON POWER APPARATUS AND SYSTEMS, SEPTEMBER 1969
can be observed to occur at the sending end rather than at the dis- interruptions and restriking. At the same time, however, such a
tant open end as usually occurs on shorter lines. switch would have exhibited restriking on the 60-Hz base when
We are indebted to Mr. Ahlgren and Mr. Hansson for pointing out opening a large bank.
that many modem arresters do have the thermal capacity to deal It is somewhat ironical that the remarkable restrike-free char-
adequately with the overvoltage problem described. This is particu- acteristics of the vacuum switch on interrupting large capacitive
larly true of arresters that depend principally upon a high internal circuits are good enough to permit an interruption at a current zero
arc drop to arrest the overvoltage. in the closing path, which in turn gives rise to the piggy-back type of
Dr. Pflanz has pointed out, like others, that oil and airblast restrikes on the high-frequency base. The suppression of the super-
breakers have been observed on occasions to interrupt at current imposed high-frequency voltage transients that accompany this
zero in the closing stroke and, therefore, could produce the observed phenomenon can be accomplished, as discussed earlier in the closure,
phenomena. Indeed the case reported by Knudsen [8] documents by the introduction of standard circuit artifices, which avoids the
very similar behavior of an oil-poor breaker. We disagree, however, sudden collapse of circuit voltage at the capacitor at the time of
that load break switches and air-blast breakers will also produce the energization. In this manner, the resulting overvoltages and rates of
observed phenomena. An air-make interrupting switch was tested change of voltage are, in general, halved. It is fortunate indeed that
with the identical circuit parameters. This type of switch, on closing, only a small percentage of application circuits require this special
excited the circuit natural-frequency currents without subsequent attention.
Late Discussion
the many tests we have performed, it has been confirmed that there
exists a systematic correlation between the flashover voltage of a
contaminated surface and the surface layer conductivity. >: L
Each time we prepare a saline solution of sodium chloride and a II