You are on page 1of 8

Corona Phenomena from Water Drops on Smooth

Conductors under High Direct Voltage


M. Akazaki

Abstract: The effect of high direct voltage on the formation In this paper, the author presents laboratory studies to
and dislodging of water drops hanging from a smooth conductor further clarify the understanding of corona phenomena from
will be shown and used to explain the resulting corona character- d d
istics, such as pulse amplitude and repetition rate, audible noise, conductors uring ram.
radio influence (RI) and loss current, and corona current wave-
forms. Phenomena taking place during the conductor-drying Apparatus and Method
period are also explained. The relationship between the number Figure 1 gives a general picture of the apparatus. A test
and spacing of points on a smooth conductor, on one hand, and conductor of smooth aluminum, 3/4 inch in diameter, forms
RI and loss current, on the other, is demonstrated. Results ob-
tained from water-drop experiments are compared with those the inner electrode of the concentric cylindrical system. This
from experiments using metal points of various shapes. is energized by dc voltages up to 95 kV, positive or negative,
by the voltage multiplier at the left. The outer electrode
consists of a mesh cylinder and two copper-end cylinders
Because of the extremely high voltage levels now in use on which can be connected to the mesh cylinder by appropriate
power systems in North America (and elsewhere), it has be- resistances. Water may be dropped through the outer mesh
come necessary to determine more exactly the corona charac- electrode to strike the inner energized conductor. Various
teristics of conductors under high-voltage stress. Both RI numbers of drops or streams can be supplied bv the 21 stop-
and the power loss created by corona discharges on conductors cocks on the water manifold.
are important factors in the planning and designing of new Figure 2 issadarmo
Fgr
a diagram of this apparatus when ittiis used
hsaprtswe sdt
to
tragnsmission asess.ent
transmission lines.
A general assessment of the characteristics of these two
of the characteristics ofthesetwo
examine the shape of corona pulses and to measure radio noise
influence voltage (RIV) and loss current. The mesh cylinder
corona phenomena for particular conductors can be obtained is connected to the end cylinders of the same inside diameter
by using test lines such as those of the National Research through resistances r2 and to the ground through measuring
Council (short dc) [1 ] and those of The Hydro-Electric Power resistance rsi
Commission of Ontario at Coldwater, Canada (ac) [2]. The values of these resistances are chosen to minimize dis-
However, detailed laboratory studies are also required, if tortions on the output corona pulse waveform which, other-
the mechanisms involved are to be understood. wise, would be caused by reflectionse These resistance values
This is particularly true when an attempt is made to analyze were determined experimentally al
by applyingrS0ns
a square
corona characteristics during rainfall. Here, complex effects wave pulse between the center oint of the mesh cylinder
are caused by raindrops falling in close proximity to the and the ground.
conductor [3], by droplets splashing when they hit the surface,
and by drops hanging from the conductor [4].
Pauthenier, et al. [5] have studied the onset of the me-
chanical vibration of a stationary water drop at voltage Fig. 1. Apparatus
gradients below the normal breakdown strength of air, for dc corona
These authors have shown the change in the shapes of sus- m e a s u re m e n t
pended water drops with increasing voltage by a method of during simulated
shadow photography and have plotted the discharge currents rain
as a function of voltage gradients at the surfaces of conductors
with varying diameters.
English [6] also has investigated the characteristics of
corona discharges from a water drop, in a point-to-plane gap.
He suggests that there are various shapes of corona discharge
current waveforms, depending on polarity and voltage. In
addition to these fundamental laboratory investigations,
there is some experimental evidence that RI from a dc test
line of positive polarity is less during rainfall than it is in dry copper cylinders
weather and that it gradually increases to normal value after / tchmg esistonceS
the rainfall. F F.
Power loss, however, is greater during rainfall than in dry
weather [7]. By way of contrast, it appears widely accepted Coscode OC nmesh cylinderi
that, for ac lines, both power loss and RI are larger during Generoorto
rainfall than are the normal fair weather values [8], [9]. -*1 c F
IEEE Transmission and
A paper recommended by the IEEE Transmission and Distribution
Distribution I
otching -i
- --
2 ~ ~oscioscope
~ ~ ~ TeWtoni.58i
Committee and approved by the IEEE Technical Operations Com- IOV resistonce L .
mittee for presentation at the IEEE Intemational Convention, New 400cps ' Stoddort NM-208
York, N. Y., March 23-26, 1964. Manuscript submitted November b L NOise meter
4, 1963; made available for printing August 5, 1964.
M. AKAZAKI, formerly with the National Research Council, Ottawa, Fig. 2. Schematic diagram of apparatus for measuring corona
Ont., Canada, is now with Kyushu University, Fukuo_ka-Shi, Japan. phenomena from water drops

JANUARY 1965 Akazaki-Corona Phenomena from Water Drops on Smooth Conductors 1

Authorized licensed use limited to: Tallinn University of Technology. Downloaded on April 08,2021 at 12:29:16 UTC from IEEE Xplore. Restrictions apply.
0 5 15.5 2675 290us
0 0 ;. 0^00 .-000 w0nd.
00000000000000 0 94< o~~~S.1u5 s 5 ^2.0.'3
Photo Pho

Fig. 4 (above). Oscillograms showing effect of


conductor voltage on dc corona from water drops

0o Z5 15.. 27.5.. 52.5* w


112.5 150 u 30

o 2.5. 125 Is1.5.. 35 so 77. ma 140. 155

'+ Fig. 3 (left). Effect of voltage on water-drop forma-


tion on smooth aluminum conductor 3/4 inch in
diameter: individualindicate times
frames were taken at base
0 15 u 25 55 A 45 60 u indicated; arrows line of conductor

In this way, the surge impedance of the mesh cylinder to photographs of the appearance of water drops on the under-
the center point was determined as 150 ohms. To measure side of the conductor, as water is allowed to drip onto the
individual corona pulse shapes and RIV, a Tektronix 585 conductor. Exposure time was 1 ins.
oscilloscope and a Stoddart NM20-A noise meter were con- In this case, water drops were supplied through only one
nected through the 50-ohm matching impedance, as shown. stopcock, at the rate of one drop per 3 seconds, by adjusting
This corresponds to an 8-ns time constant for the circuit, the cock and keeping the water head constant. The con-,
which is made up of the 50-ohm resistance in parallel with the stant water head was obtained by feeding the manifold from
160-pF stray capacitance of the mesh cylinder to ground. a small water bucket which was always kept in an overflow
It will be shown, later, that corona discharge from the con- condition by the arrangement shown at the right in Fig. 1.
ductor, under water supply conditions, generates two distinct The arrows at the right of Fig. 3 indicate the line of the
kinds of audible noise which are best described as a "crackling" conductor base for each row of photographs. In some cases,
noise and a "hissing" noise. These two types of corona dis- the full outlines of the pendant drops have been lost in re-
charge consist of many individual pulses, each of which has production (photographs 1, 2, 3, and 5 of row 1, and photo-
the samie shape when displayed on the scope, if the meas- graphs 1 and 2 of row 2). In general, though, the photographs
uring circuit with the short (8-ns) time constant previously are easy to follow.
described has been used. In the top row of Fig. 3, there is no voltage on the conductor.
When it was desired to show the difference in these two In this case, one arriving water drop gives only one main drop
types of corona, the measuring circuit was changed to have below the conductor. The main drop falls about 1/4 second
a long time constant. This enabled the scope to display the after the arriving water drop touches the upper surface of the
waveforms resulting from many pulses over a considerable conductor. The second, third, and bottom rows show the
time. In this case, a resistance of 250 kilohms was used for type of hanging drops created when the conductor is energized
ri, instead of the 50-ohm matching resistance. Moreover, the at voltages of 30, 40, and 50 kV, respectively.
resistances r2 were removed, which gave the circuit a new time At these voltages, one hanging water drop develops a
constant equal to 40 /As. mechanical oscillation, and a long sharp point of water ap-
pears on the bottom surface of the conductor (photographs 5
Results and Discussion and 6, row 3). Corona discharge occurs easily from such a
CORONA DISCHARGE FROM WATER DROPS ON A
point. (No corona is visible in Fig. 3 because of the lighting.)
When the sharp water point becomes too long to keep its
SMOOTH CONDUCTOR form, it breaks into several droplets (photograph 7, row 3)
Under dry conditions, the inner electrode did not go into and corona discharge ceases for a moment. It resumes when
corona even at the maximum voltage available (4-=95 kV). the volume of water in the hanging drop increases because
Figure 3 shows high-speed (400 frames per second) movie additional water flowing down from the surface of the
2 Akazaki-Corona Phenomena from Water Drops on Smooth Conductors JANU3ARY 1965

Authorized licensed use limited to: Tallinn University of Technology. Downloaded on April 08,2021 at 12:29:16 UTC from IEEE Xplore. Restrictions apply.
Table I
Sweep
Photo Voltage, Time,
No. kV Seconds Observations and Comments
1 +30 0-0.5 Three crackling spikes from three con-
secutive sharp water points formed
from one water drop. Each "spike"
consists of many individual corona
pulses.
0.5-1.7 Voltage too low to cause corona from re-
mainder of drop.
1.7-2.7 Five spikes in 1 second, as a result of a
newly arrived water drop which forms (A) (B)
five sharp points.
2 +40 0.3-1.5 Approximately 16 spikes in 1 second.
Larger amplitude and higher rate of
formation of sharp water points be-
cause of higher voltage.
1.5-3.5 Voltage too low to cause corona from re-
mainder of drop.
3.7 Start of crackling corona spikes from a
new drop.
3 +50 0-0.5 Rapid crackling from consecutive sharp
water points. Amplitude higher than
for previous voltages.
0.5-2.1 Voltage sufficient to cause hissing corona
from remainder of water drop.
2.2-4.0 Approximately 20 spikes per second from
water points from a new drop.
(C) (D)

Fig. 5. Positive corona pulses at various times after single


conductor forms a new point (photograph 8, row 3). The water drop falls on smooth conductor aluminum
time between the formation of one water point and that of the
next is several milliseconds, but the higher the voltage, the A-Thirty seconds, 0.2 volt per division, 0.1 ,s per division
more this time decreases. From a study of the original film, B-Three minutes, 0.2 volt per division, 0.1 lAs per division
average times for each case are given as: C-Eight minutes, 0.2 volt per division, 0.1 js per division
D-Ten minutes, 0.2 volt per division, 0.1 jus per division
No voltage: 267 ms
30 kV:78 ms
40 kV: 53 ms
50 kV: 22 ms
These recurring sharp water points permit the formation of a shown by Fig. 4. Scope sensitivity is 1 volt per division,
corona which produces the spiked current waveforms to be scope sweep time is 0.5 second per division, and the conductor
described. diameter is 3/4 inch. The rate of water-drop supply is ap-
While each water point exists, many corona pulses are proximately one drop every 3 seconds.
emitted from it. There is then a momentary relatively For comparison purposes, the same time constant was used
quiet period, until the next point starts to form and grow and to obtain both the hissing and crackling waveforms. It
another burst of corona pulses appears. If the large time con- should be recalled that the shape of these current waveforms
stant (40 As) is used in the measuring circuit, each burst of depends, not only on the shape of the individual corona pulses,
corona pulses appears as one spike on the cathode ray oscillo- but also on the time constant of the measuring circuit and on
scope. the height and repetition rate of the corona pulses.
That is, each spike on the trace represents one group of For example, each crackling corona spike of Fig. 4 (which
pulses from one water point; several spikes correspond to a is made up of many individual corona pulses) would appear as
series of grouped pulses resulting from several water points an approximate trapezoid on a faster sweep. This trapezoidal
created, in succession, from one water drop (photograph 1 of spike is similar to one of the shapes of corona current wave-
Fig. 4). forms reported by English [6]. After careful analysis of in-
The rate of occurrence of spikes on the trace increases as the dividual corona pulses for the two types of corona noise
voltage is raised because raised voltage increases the rate and for corona from different water and metal points, it was
of water-point formation. (This has already been explained.) concluded that individual corona pulses have a standard
A corona generated by this process gives a characteristic shape [10].
crackling noise, and will be referred to by the author as As soon as a water drop hits the conductor, there are many
"crackling corona." small water particles on the conductor surface in the neighbor-
At low voltages, when the water on the surface of the conduc- hood of the main pendant water-drop point. These small
tor is nearly all gone and insufficient water remains to make a water particles cause hissing corona discharges similar to that
sharp point under the influence of the electrostatic field, the just described. At the same time, the main water-drop point
corona discharge will be finished. However, if the applied produces a crackling corona for a second or two; it then
voltage is very high, a short sharp point is produced by the changes to a hissing corona, when recurrent creation of long
remaining slight amount of water and a corona discharge con- points ceases. (In the 50-kV case of Fig. 4, hissing takes
tinues. place at the same time as the crackling, but the crackling
This discharge differs from that previously described (long and the choice of sweep time mask the effect.)
water-point discharge) by emitting a hissing, rather than a Figure 5 shows individual positive corona pulses which
crackling, noise. Current waveforms illustrating both types were recorded using the fast (8-ns) measuring circuit. These
are shown in the third photograph of Fig. 4. Table I gives the pulses originated in water-drop points on the surface of a
voltage, sweep time, and peculiarities of all three oscillograms conductor energized at 76 kV after the water drop supply
JANUARY 1965 Akazaki-Corona Phenomena from Water Drops on Smooth Conductors 3

Authorized licensed use limited to: Tallinn University of Technology. Downloaded on April 08,2021 at 12:29:16 UTC from IEEE Xplore. Restrictions apply.
450 1 1 l I I I m loss current variation with time under positive polarity during
_____ _____ _____ the conductor-drying period after the water-drop supply has
been turned off. From Figs. 5 and 6, the relationship for
positive polarity can be seen between increase in RIV and
300 ] increase in pulse amplitude with the time after "rainfall"
stops, even though loss current steadily declines as a result of
the lower pulse repetition rate.
CORONA DISCHARGE FROM METAL POINTS ON A SMOOTH
65 KV CONDUCTOR
-1 0 0
X To verify the conclusion that RIV goes up with time after
Test Run No. 1:
the water-drop supply is turned off until the conductor is dry,
0 - -I - 2: (3) corona characteristics from metal points having shapes
3: x-x-x similar to those of water drops were investigated. It was
4: assumed that no significant difference exists between the two
(A) types of points, except the change in the length of water
points during the conductor-drying period.
50 ______ Single Point
Figure 7 gives the dc corona characteristics for both positive
and negative polarity of a smooth conductor having a single
40 metal point; the shape of the point is altered for different tests.
Each curve shows the spread in characteristics caused by eight
different points of the same shape, applied singly.
30 < '< At Under either positive or negative polarity, the sharper metal
points give lower RIV, but have higher loss current, when
____ ___ _ _ _ _ I compared
l l to dull points. This is in agreement with the ob-
20 t <- servation of Liao, Keen, and Powell [11] that for the nega-
_,-x-l
tive half cycle of an applied alternating voltage, the sharper
10
I0
____ I - X-______
3
_____
\
the metal point, the smaller is the individual corona pulse
amplitude. They also reported that pulses originate more
rapidly from sharp points than from dull ones.
01 1 Figure 8 shows the effect of the shape of metal points on
0 2 4 6 8 10 12 14 corona pulse amplitude for positive polarity. Note that
Time (minutes) these results agree with the results obtained for corona from
(B) a water point during the conductor-drying period (Fig. 6).
These results have been qualitatively explained [11] on the
Fig.6. RIVandlosscurrentvariationwithtimeunderpositive basis of field distribution differences between the two cases.
polarity The sharp point has a high gradient at the tip so that corona
onset occurs readily. However, the gradient falls rapidly with
A-RIV in ,V distance from the point; thus the pulse amplitude, which is
B-Loss current in MA governed by the field in the ionization zone, is relatively low.
The next pulse is then propagated. The gradient at the tip
of the dull point, while lower than that of the sharp point,
does not fall away as rapidly. The net effect is a greater total
had been turned offndnlyhissngoroimpetus for the pulse and, therefore, a higher amplitude.
crandconlyking radfinihemhed
hadbeen turnedof
is, when the crackling corona corona (ha However,
had finished). The hissing isntgigt.cu tip, onseti
lower gradient at the rate.
because of thesraiyadterptto
corona continues for a much longer time than the crackling lower.
corona, often lasting for minutes.
This is because a hissing corona does not cause as rapid a Spacing and Number of Multiple Points
depletion of the water. The many pulses of hissing corona, To study the interaction between corona discharges from
from many small water points of various heights, have a neighboring points, loss current and RIV were measured with
variety of amplitudes, as can be seen in Fig. 5 (left-hand the distance between the metal points along a conductor
picture). varying from 3-12 inches.
However, after some time (for instance 2-5 minutes, de-
pending primlarily on the applied voltage), only the original, Corona loss current: Figure 9 shows the total dc corona loss
main but depleted, pendant water-drop point is left, causing current under positive and negative polarity, from four and
hissing corona discharges. Gradually, because of spray effect, three metal points, as a function of the spacing between points.
even this point decreases and the amplitude of the hissing Three types of points were tested here; their dimensions are
pulses becomes greater (see Fig. 5). At the same time, the given in Fig. 10.
pulse repetition rate decreases because the voltage gradient at There were indications at the higher voltage levels, and
the water tip decreases as the point becomes shorter. particularly for the sharper points, that some interaction takes
Rakoshdas [10] has observed essentially the same effect for place, even with a 9-inch spacing. This is also shown in Fig.
wires of increasing diameter. 10, which is a plot of total corona loss current against the num-
Figure 6 shows the RIV (quasi-peak value at 1 Mc/s) and ber of points spaced 9 inches apart. Here the curves,
4 Akazaki-Corona Phenomena from W~ater Drop8 on Smooth Conductors JANUARY 1965

Authorized licensed use limited to: Tallinn University of Technology. Downloaded on April 08,2021 at 12:29:16 UTC from IEEE Xplore. Restrictions apply.
100 100

< <

a 1
o
a, lo
~~~~~~~~~~~~0
0

_
__|I __ ____ _| IOmmNTE

108 ____ 32.4 500 5O8.S

_____--r
~-n
~~~~~~~~~~~~~ ~ ~ ~
A]Omm
~~~ mm
NOE
I~ ~ ~. Dimensions of the four types of
~~metal points are shown of the leff.
I AliT A
-T
5mm 5 mm
Conductor size: 3/4"
~~~~~~2.
ICC A ~~~ ~~~~~~~~~ ~~~~~~~~~~~00 \73 5mm 3. During eoch test only one point was

I_ ___ l_ l7K F KT]1

10.8 21.6 32.4 43.2 54.0 64.8 75.6 86.4

(A) (B)

Fig. 7. Comparative characteristics of dc corona from different metal points mounted on smooth conductor

A-Conductor potential in positive kV B-Conductor potential in negative kV

although straight lines, do not produce back to zero for the interaction seems to be 9 inches for the particular points tested.
higher voltages.points, :~~~~~~~~~~~~~~~~~~~~~~~~~~~ifeet
mounted It is felt that the choice of points wasont sufficiently
f oh ye varied to
In terms of field distortion, it seems logical that both in- cover practical "raindrop" cases.
creased voltage and the increased sharpness of points would In Fig. 10, the relationship between loss current and the
affect the spacing required to avoid interaction. To a reason- number of points n is linear, even at higher voltages and with
able degree of accuracy, however, the distance required for no sharp p)oints where some interaction was expected. This is

JANUARYComp1at6v characteristcsofodc
Pheomena from WatferenmeotalpoinsmonedoSmooth conductorr

shape of metal

ductor, on corona Bacgru _


pulse amplitude gtl

kV)NUARY
1965 (Wn

Authorized licensed use limited to: Tallinn University of Technology. Downloaded on April 08,2021 at 12:29:16 UTC from IEEE Xplore. Restrictions apply.
uA
240 x 77.6KV

220 x

,A 200 200 200 x 77.6KV


200 K77.6 KV 200A

180 8
180
Point Shopes 7807 KV 180
1800 / a- /669KV
X 4

160 x 69KV 160 60 160


77/6 KV x 77.6KV 76V . 69KV

460 4 60 _140o60.46V60KV/40 77VKV


°3K69KV o

120
0
120
O 100
77_ o X 47..6KV
o
7
8 /~~~~~~~~~~~~~~~~~~~~x~
120
776K 6076 K V
2077.0<^
KV604
77.6 KV
369 KK 3 6 KKV
'00 x6Fa60.49KV
10 corona
X l00 - 7.K 100
60.4b 60I .KV o
69 KV 240x 77.69KV

t40 40 - / 7 7i8V40 /4 6.K


820 L...4KV 80 -8-860.4KV 80 220 6KVK5
200- _ _ /\ 5mm 77 60.4KV 2 0 0a6 ,
/ KV
KV 69KVa

6084 KVmm 18/ j


- 60.46KV 60 60
60 60 t / 776KV 0 80
o lOO 1 / / //// / i / 1 51 8 KV a) < _ x ,!s * 6KV
60.4

40 60 6494KV 60 40 17-u60.4KV
60 .-...a4KV 60 KV 60.0KV a 43.2KVKV
2040 200 .8K
20
0-4 51..V L /X/<
S5 8 KV 20

0 3 6 9 0 3 6 9 12 0
0 3 6 9 0 3 6 9 12
Number of Paints; 4 Number af Points: 3 Numnber of Points: 4 Number of Paints:
3

(A) (B)
Fig. 9. Dc corona loss current as function of spacing between points
RD 120 0X 776KV C ^/ 240 69 OKV t a 69,KV
5
A-Positive polarity B-Negative polarity

240 240 77.6KV

220 -a- AiI m220 a A

o P a fmm
4

200
Am 2 z
776V

5mm o~~~~~~~~
4o.2K Paint Shapes

180 10 mm 180 77.6 KV3230 320


~~~~~77.6 KV 32

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~a77.6 KV
160 160
~~~~~~~ ~~.77.6KV
60280
160 280

a.< 69KV t 7.K


:.140 .6 140 60.4
7.K

~~~~~~~~~~~
E-
~69
KV24
240OK
'E 6*240 z
~~~~~77.6KV
120
a .69KV
0 120
:3
~A 69.OKV~~~~~~~~~~~~~~~9.K 6.K

KV0 ~~~~~~~~~~~~~200
60.4 KV A
60.4KV 776KV 200-
1co a 69KV a100 ~~~~~~60.4
51.8 KV. A u,60.4KV o a-a 77.6 KV

u A a6.K
.60.4KV 0 a69 KV 160 160
80 a80 a ~ ~ ~
x776K
-a776KV a aI 77.6 KV..
a a
a a60.4KV 51.8 KV
6.K

B-eatv 60 a
polarit * 60 ~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~120 -120
B-Tre -oi-A a- a69.0 KV
a a a a~~~~-51.8 KV a 60.4KV
a 690KV
X~~~~~~~~~~~ ~~~~~~~~~....--VX- a 69.0 KV
40 a4 2V 40 a80 aa "80 a60.4 KV
a A 518KV
~604K
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~7'

20 20 8a 51.8 KV
5
1aaKV51 .8 KV
* a ~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~40
aKV a 403
~~.a...~~~-aa43 2KV ~ ~ ~ ~~
43.2

Authorized licensed use limited to: Tallinn University of Technology. Downloaded on April 08,2021 at 12:29:16 UTC from IEEE Xplore. Restrictions apply.
because the first point, which is free of any infringement, causes
only an "offset" and because more than two points are subject x
to the same degree of interaction with the 9-inch spacing.
Thus, the total loss current increases proportionally with n,
initial "offset" is
I
and any
neglected.
RIV: Figure 11 shows RIV under the same conditions and 400
point dimensions as Fig. 9, but for positive polarity only. As 4
in the case of corona loss, with 9-inch spacing, RIV is relatively 77.6 KV
free of interaction, but again the effect of the increased voltage 300
and sharpness is to require greater spacing to avoid interaction. 69.OKV
Figure 12 shows, for positive polarity and with the same
physical conditions as Fig. 11, that RIV (with quasi-peak 200 77.6KV
value at 1 Me/s) increases approximately as the square root of 60.4 KV
n. Lippert, et al. [12] have found this in their laboratory
tests.
x 77.6KV
6.K
According to Stone, Gens, and Gehrig [13], this relation- >
ship holds only when the corona pulses are nearly coincident l00 69.0 KV
in time. If the pulses are separated by a time comparable to a 90
the measuring meter's charging time, then the square root rule 2 KV
80
does not hold. Their tests were made with the use of com- O 70
mercial conductors energized by alternating voltages. Al- > 60 x 60.4 KV
though the square root rule holds in the case of a laboratory , 60 .8KV
setup energized by a dc source, more detailed investigation 50
must be done, using an HV dc test line in the field.
Fig. 13 indicates that, for positive polarity, RIV from the c 40
shorter metal points is larger than that from the longer 43.2 KV
metal points at all frequencies. The points are 12 inches 32
0 30:
apart. Frequency is given in megacycles. 0

Conclusions 20
1. The shapes of individual corona pulses from water-drop
points hanging from a conductor are the same as those from RIV OC
metal points.
2. When water is dropping on a conductor, two kinds of
audible noise are recognizable, "hissing" and "crackling." I l I
Both kinds of corona discharge are caused by individual 3 2 4
pulses of identical shape and of many different amplitudes, . ....
but with different pulse repetition rates and different rates of Number of Points n
recurrence for each burst pulse group. The recurrence rates Fig. 12. RIV from differently shaped metal points at 9-inch
of these bursts differ because hissing corona comes from rela- spacing as a function of number of points
tively stable small water-drop points, while crackling corona
comes from water drops which form into long thin points
which break off quickly. 100
3. English [6] reports that the differences in corona current 70 A
waveform depend on the time constant of the measuring cir- >- 50 / +65KV lA l
cuit. Although he stated that there are various shapes of I
corona-discharge current, the author has found that different 30|
corona patterns are caused by many individual pulses which 20
all have the shape previously described. 0>
4. The amplitude of individual pulses from water-drop 10
points, as well as from metal points, is believed to depend on a 7 , \ L
the field distribution in the vicinity of the points. When the
gradient at the tip of the point is very high but falls rapidly
5 aW 7

c
with distance from the point (as for sharp points), the pulse 3
amplitude is low because total impetus from the field is low. *n 2 Typicol
Repetition rate is high, since "onset" is easy as a result of pokron
RIV orints se Fig. 10
the high initial gradient. { RVIonsseFgl
5. The longer the water or metal points are, the smaller in 0.1 0.3 0.5 1.0 3.0 5.0 10
amplitude are the individual corona pulses. However, long Fig. 13. Frequency spectrum of RIV from metal points of two
points emit many more pulses than shorter ones, under both different shapes
positive and negative polarity.
6. It is clear that, for positive polarity, the amplitude of
Dotted Iines-RIV from three sharp short brass points
Solid lines-RIV from three long sharp brass points
individual corona pulses becomes larger and pulse repetition Conductor-Smooth aluminum, 3/4 inch in diameter
JANUJARY 1965 Akazaki-Cjorona Phenomena from Waler Drops on Smooth Conductors 7

Authorized licensed use limited to: Tallinn University of Technology. Downloaded on April 08,2021 at 12:29:16 UTC from IEEE Xplore. Restrictions apply.
rate decreases with time during the conductor-drying period, high direct voltages, IEEE Trans. on Power Apparatus and
because the stable water-drop points become shorter. Systems, vol 83, Jan 1964, pp 5-16
[2] Nigol, O., and J. G. Cassan, Corona research at Ontario Hydro-
7. Various tests have demonstrated that, at a given posi- Coldwater project, Trans. AIEE (Power Apparatus and Systems),
tive polarity voltage, RIV is a function more dependent on Vol 80, Jun 1961, pp 304-313
pulse mpolaitudevolthage,
pulse amplitude than on the
tion
the repetition
r orae ofpendividual
rate of individual
[3] Griscom, S. B., A theory regarding the generation of radio noise
on high-voltage transmission lines, Proc. American Power Conf.
pulses (when a quasi-peak meter is used under laboratory vol 23, 1961, pp 745-753
conditions). Loss current, however, is a function more de- [4] Sato, Y., Y. Tsunoda, and K. Arai, Corona pulses from a water
pendent on repetition rate than on the amplitude of in- drop on a cylindrical conductor surface, Electrotech. J. (Japan),
vol 8, no. 1/2, 1963, pp 34-38
dividual pulses. [5] Pauthenier, M., G. Duhaut, and L. Demon, Contribution a
8. For positive polarity, RIV goes up with time during the l'etude des pertes electriques dans les lignes A courant continu A
haute tension. Etude particuli6re des gouttes d'eau en pointe
conductor-drying period because of the higher pulse ampli- fixe, Rev. Gen. Elec., vol 59, 1950, pp 133-136 (Nat'l Research
tude, but loss current decreases because of the decreasing Council of Canada Tech Translation TT-1084).
repetition rate. [6] English, W. N., Corona from a water drop, Phys. Rev., vol 74,
1948, pp 179-189
9. In the case of metal points on a smooth conductor in a [7] Hylten-Cavallius, N., et al, Insulation requirements, corona
laboratory setup, the distance for no interaction between losses and corona radio interference for high voltage direct
current lines, Study Committee No. 10, CIGRE, Paris, France,
corona discharges from two neighboring points seems to be 1963
above 9 inches. To some extent, it is dependent on voltage [8] Lippert, D. G., W. E. Pakala, S. C. Bartlett, and B. J. Sparlin,
and the sharpness of points. The radio-influence characteristics of bundle and single con-
ductors-500-kv test project of the American Gas and Electric
10. RIV from metal points on the conductor is propor- Company, Trans. AIEE (Power Apparatus and Systems),
tional to the square root of the number of corona points, if the vol 76, 1957 (Feb 1958 section), pp 1302-1310
points are spaced far enough apart to each be in corona. [9] LaForest, J. J., C. B. Lindh, D. D. MacCarthy, F. Olsen, and
Loss current is linearly proportional to the number of corona M. W. Schultz, Jr., Radio noise and corona loss results from
project EHV, IEEE Trans. on Power Apparatus and Systems,
points. vol 82, Oct 1963, pp 735-750
[10] Rakoshdas, B., Pulses and radio-influence voltage of direct-
Acknowledgment voltage corona, ibid., May 1964, pp 483-491
[11] Liao, T. W., W. A. Keen, Jr., and D. R. Powell, Relationship
The author would like to thank P. Robert who assisted between corona and radio influence on transnmission lines,
him with the measurements; R. M. Morris, A. R. Morse, N. L. laboratory studies. I-Point and conductor corona, Trans.
Kusters, and F. and guidane inthe onperaon
C. Creed for their comments performing 530-540(Power Apparatus and Systems), vol 76, Aug 1957, pp
AIEE
the experimental work and guidance in the preparation of [12] Lippert, D. G., W. E. Pakala, S. C. Bartlett, and C. E. Fahrn-
the manuscript; and Mrs. I. M. Wood and Miss L. E. Chap- kopf, Radio influence tests in field and laborator- 500-kv
man for typing and drafting the paper. test project of the American Gas and Electric Company, Trans.
AIEE, vol. 70, pt I, 1951, pp 251-269
References [13] Stone, L. N., R. S. Gens, and E. H. Gehrig, Radio interference
attenuation on energized high-voltage transmission lines:
[1] Morris, R. M., and B. Rakoshdas, An investigation of corona measurement and application, Trans. AIEE (Power Apparatus
loss and radio interference from transmission line conductors at and Systems), vol 78, Dec 1959, pp 1238-1247

A Guide to the Application of Capacitors without


Induction Motor Self-Excitation
C. F. DeSieno, Senior Member IEEE B. J. Beaudoin, Student Memiber IEEE

Abstract: This paper develops a straightforward method for static load is to increase the amount of capacitance which can be
determining the maximum amount of capacitance which can be used safely without induction motor self-excitation.
connected in parallel with an induction motor and static load com-
bination without danger of self-excitation. Special consideration
is given to the effect of the static load. Graphical results are When an induction motor is operating in lparallel with a source
presented which can be readily applied to a wide range of induc- of capacitive kilovars, there is the well-recognized danger of
tion motor size and static load combinations. The effect of the self-excitation if the motor-capacitance combination is dis-
connected from its voltage supply. One kilovar source which
Paper 64-59, recommended by the IEEE Transmission and Distribu- is often switched with the motor load is static capacitors,
tion Committee and approved by the IEEE Technical Operations which are used principally to correct the load power factor.-
Committee for presentation at the IEEE Winter Power Meeting, Another source is the capacitance of atransmission line supply-
New York, N. Y., February 2-7, 1964. Manuscript submitted
October 30, 1963; made available for printing December 23, 1963. ing the motor load which may be de-energized at the other
C. F. DESIENO iS with the American Electric Power Service Corpora- end. When self-excited (more specifically, capacitor-excited),
tion, New York, N. Y., and B. J. BEAuTDoINis with the Massachusetts the motor operates temporarily as an induction generator
Institute of Technology, Cambridge, Mass. while its armature continues to rotate because of the kinetic
This paper is based in part on a thesis submitted to the Department of energy stored in the rotating masses. Under this condition,
Electrical Engineering, Massachusetts Institute of Technology, by th moo temnlvlaecnrset aaigvle e
B. J. Beaudoin, in partial fulfillment of the requirements for the B.S. temtrtrmlvlaecnrs odmggvle e
degree. cause it is no longer regulated by the voltage source. In
8 DeSieno, Beaudoin-Capacitors without Self-Excitation JANUARY 1965

Authorized licensed use limited to: Tallinn University of Technology. Downloaded on April 08,2021 at 12:29:16 UTC from IEEE Xplore. Restrictions apply.

You might also like