Vous êtes sur la page 1sur 4

H igl -Pressure Gas

as a

Dielectric

GORDON C. NONKEN
ASSOCIATE AIEE

Synopsis: Impulse and 60-cycle breakdown


strength of sphere gaps, rod gaps, and f
samples built up of solid and fluid dielectrics
arc reported. The tests were made in nitrogen up to 200 pounds per square inch gauge
pressure and in Freon up to 70 pounds per
square inch gauge pressure. Dielectric
strength values were compared with the
breakdown of transformer oil between the
same electrodes. Gap conditions were found
for which the 60 cycle breakdown was twice
its impulse breakdown.

ECENT interest in the dielectric

R strength of gases ulder pressure has

been stimulated by the steadily increasing voltages used in power transmissioii


as well as in X-ray and atomic physics
work anid bv the shortcomings of atmos-

a*
air d of oil insulation. Althougl
pheric
scores of articles have been published oii
high pressure gas dielectrics during e
last forty vears, they colitain practicallv
no iniforniation on the impulse dielectric
strength of gases at higliressure. This
is a seriouts omission frotn the point of

lasth poressuyegars theleycricsta uingpracti thie

view of those interested in generating


and distributing apparatus since the kind

and amounit of insulatiotn in these apparatus are influenced greatly by impulse


p
a
strength. The theoretical background
related to the breakdown of gaps in gases
is much too incomplete to permit the
extrapolation of available data much be-

yond theexactconditionsunderwhichthe
data were obtained. Therefore, numerous general as well as specific breakdown
data must be accumulated in order to
make the proper utilization of high pressure gases as dielectrics,
This paper will present impulse and 60cycle breakdown data between gaps in
nitrogen up to 200 pounds pressure and
in dichlorodifluorometliane (dichlorodifluoromethane, commercially known as
"Freon" will be referred to in this paper
as CC12F2) up to 70 pounds gauge pressure. In each case, a comparison will be
made with the breakdown of transformer
oil at atmospheric pressure between the

same gaps

Recent Work
Several valuable contributions have
been made to our fund of engineering
knowledge of high pressure gas dielectrics
during the last decade. Paschen's law,
which was looked upon by many as a
basic law of the dielectric strength of gaps
at different spacing and pressure, has
been shown to diverge far from the data
at pressures above about six atmospheres. This law states that the sparkover voltage of a uniform field gap in a
gas is directly proportional to the product
of the gap spacing and the absolute pressure. The breakdown of a fixed gap is
lower than the breakdown predicted by
Paschen's law at pressures above 100
pounds per square inch. In fact the
data 1,2 indicate that for air tested between spheres at a pressure of 600
pounds per square inch the actual sparkover is 55 percent of that predicted from
atmospheric pressure data by Paschen's
law.

dielectric strength of nitrogen when tested

in a uniform field at atmospheric pres-

sure. Notable among them are CCIlF


with a dielectric strength of three times
that of nitrogen and CC12F2 the gas commercially known as Freon and in common
use in mechanical refrigerators, with a
dielectric strength of 2.4 times that of
nitrogen at atmospherio pressure.
Most of the earlier dielectric strength
tests were made at low pressure or small
spacings. Recently d-c breakdown tests2
on spheres have been made at pressures
up to 600 pounds per square inch and
450 kv breakdown. These data show also
how slight roughnesses of the electrodes
reduce the breakdown by as much as 50
per cent at 600 pounds pressure. Other

investigators4 made test in air and nitrogen up to 1,700 pounds per square inch.
They indicate that the rate of increase of
dielectric strength of air with increasing
pressure continues to decrease at least up
to 1,700 pounds per square inch pressure.
This and other work5 indicate that although the dielectric strength of air and
nitrogen are practically the same at at-

mospheric pressure, the dielectric strength


of nitrogen increases less rapidly than
that of air with increased pressure, particularly above 100 pounds persquareinch.

Apparatus
The experimental data presented in this
paper were obtained by tests in the
special pressure tank shown in figure 1.
The pressure tank is four feet in diameter
by seven feet high. Although it was
tested hydrostatically up to 400 pounds
per square inch pressure, it has not been
used above 200 pounds gas pressure for
safety reasons. It has three five-inch
diameter sight windows made of discs of

iO
o

Paper 41-159, recommended by the AIEE com-

mittee on basic sciences, and presented at the AIEE


South West District meeting, St. Louis, Mo., October
8-10, 1941. Manuscript submitted August 5, 1941;
made available for preprinting September 8, 1941.
GORDON C. NONKEN is with the works laboratory of

-_ _i
icm

so__
To0

___

__

40

__+ __

__

30

Some gases have been found2 which


have as much as three times the 60 cycle

" 2

-_
_
300 _
E_ E
500
400

,0

20

30 40

SO

100

200 300

ABSOWTE PRESSURE IN POUNDS PER SQUARE INCH

Figure 2. 60-cycle and impulse breakdown

between 6.25 centimeter spheres


and 8=CCI2F2, impulse and 60-cycle
CurvesCandD=nitrogen,impulseand60-cycle
General Electric Company, Pittsfield, Mass.
Curve
cpe
E=ol, mpulseda
1C
For all numbered references, see list at end ofE=oil,impulse at atmospheric pressure
CurveF=oil,60-cycleatatmosphericpressure
pFaper.
Curves A

toshr

Figure 1. High-pressure electric test tank

DECEMBER 1941, VOL. 60

Nonken-Gas as a Dielectric

TRANSACTIONS 1017

methyl-methacrylate and an 18 inch steel


door. A single high voltage bushing
was used to carry the voltage into the
pressure tank. This bushing was of the
oil-filled type with a porcelain shell. The
bushing was the same size both inside and

outside the tank so that it could be used at X_

240 - -

160

- -

{/-r-

=_

/
maximum voltage with atmospheric pres..120
- - _
sure in the tank.
0 - 7t --~---->
Two devices were used to adjust or
change the electrodes inside the tank
_o
without opening the tank. One consisted
of a screw which was operated by a shaft
- - - -_
40
passing through the tank wall to a cali_ _ ll
_
_
_
brated hand wheel. The gap electrode
was mounted on the end of the screw and
200
16 O
0
120
8
SO
60
40
100
20
40
could be spaced accurately from the outof
consisted
other
The
of
the
tank.
side
PER SQUARE INCH
reolig ltfristlediteho-ABSOLUTE PRESSURE IN POUNDS
a revrolving platform installed in theabottom of the tank. The test samples which by a 350 kv, 1,000 kva testing trans- Figure 4. 60-cycle and impulse breakdown
were made up of solid and gaseous di- former. The voltage was varied by con- between one-half-inch square rod gaps in
nitrogen
electrics were good for only one break- trolling the field on the alternator conbreakdown
Curves
and
C=60-cycle
B,
A,
The
the
testing
transformer.
to
nected
as
twelve
As
many
down determination.
of these solid dielectric samples were impulse voltage was supplied from a 500 Curves D and E=60-cycle corona starting
voltage
placed on the revolving platform. These kv impulse generator. The impulse voltbreakdown
and
Curves
H=impulse
G,
F,
samples were connected, one at a time, age wave shape was checked and the
a
caths
with
were
observed
to the voltage source by turning the turn breakdowns
zed in figure 2. These data show
table in the proper position with a driving ode ray oscillograph connected across the summari
mechanism which ran through a packing test electrodes through a resistance di- the dielectric strength of nitrogen to ingland in the tank wall. These mecha- vider. All impulse data presented herein crease as the 0.86 power of the pressure
nisms saved a large amount of time by were obtained with a 11/2-40 microsecond instead of increasing as the first power of
the pressure in accordance with Paschen's
avoiding the necessity of opening the impulse wave,
The dielectric strength of CC12F2
law.
tank after each test.
increases at a slightly smaller power of the
Gases used were of a commercial grade Test Results
absolute pressure than does the dielectric
obtained in high pressure tanks. The
Breakdown tests were made with the strength of nitrogen.
nitrogen was dried by passing it through
These gas breakdown curves represent
a trap surrounded by liquid air. The following three types of electrodes:
the crest value of the 60 cycle breakboth
CC12F2 with high liquefying temperature
1. 6.25 centmeter diameter brass spheres
down voltage and the crest value of the
e
was dried in columns of calcium chloride.
Standard 1/2 inch square brass rod gaps impulse voltage breakdown with a 1 /2The CC12F2 was drawn from the test 2.
and
40 microsecond voltage applied. The 60
tank through a compressor which reconwhich included cycle tests were made by increasing the
3. Electrode
densed the gas into its liquid phase and solid
dielectrics,arrangements
voltage at a rate such that the voltage
a
into
storage
it
ped
pressure
high
PUMPe
The results of experiments made be- increased from approximately half of
tank for future use.
The 60 cycle test voltage was supplied tween the 6.25 centimeter spheres are breakdown voltage to breakdown in one
minute. An elapsed time of one minute
was allowed between successive voltage
3
280?
l
280
applications. The average of ten break.
8CM
- - -\1
....

itcrintodafhighappressurelystorage

240 200 -

6CM

- 1/ - 8

1 1 1 | |

\ \1

down determinations was used to estab-

Li

' 16C - - - -

roo

Is
O
1018

I-L 1
IFI

five impulses of the same voltage at

- -

51
-1 1 I
$<-1
%.-;
%
<
o 20| 1 ' 1 S
|
144
<6C> |- T
J | |
- - I
L -1- I e 12 -

A-.11IIII

~_ loI~~
->t||1
401~._

III1

60

70

ABSOLUTE PRESSURE IN POUNDS PER SQUARE INCH

Figure 3. 60cycle ing


e
impulsbreak|and
five or more
steps two per cent apart in
above and below the breakdown
down between onehalf-inch square rod
~~~~gaps in CCI2F2

80

voltage,

voltage. The voltage was lowered in two


per cent steps until none of the five im-

the gap.CentThe
broke downtWOPer
B,dNdC= ~~pulses
CUrVeSA
StePSvoltag
WaS alSOraiSedin
Unti1
ACuresA,adC
60cyl breakdown each of the five impulses at a particular
11 11
Curves D and E= voltage broke down the gap. The statis60-cycle

| ;,'1 5 11 1 11 1 11 1111_

IEI11 40III50
Er'1tI
30
20
10

TRANSACTIONS

lish a single datum point. The maximum


spread of individual values was rarely
over five per cent for the 60 cycle tests.
t\]<|1[< 1 |The| |/@
impulse tests were made by apply-

corona

tical voltage at which half of the impulses

starting voltage resulted in breakdown was takQen as the

Curves F,G, and H

impulse breakdown

Nonken-Gas as a Dielectric

breakdown value of the gaps. The maxi-

mum spread of individual values was with


ELECTRICAL ENGINEERING

_ - - -_ .
_
few exceptions not greater than eight per
_ A
cent of the breakdown value.
240 - _ __
t
- -1
Observations showed that the breakdown of a gap was lower immediately
200
after a breakdown. Therefore a proL
cedure was followed such that a five
_
- - 60 - - minute interval elapsed between a flashover and the succeeding impulse applica- __
0
tion. Only one-minute interval was used
-i---.
after a voltage application that did not j 120 l
result in breakdown of the gap before
>
0
J 8
another impulse was applied. This spread
in the impulse data was probably in- - - - _ creased by the fact that there was no ef- - -I
_
40 -_
fort to increase the free ion concentration
_
_l l
l
within the tank by radioactive salts or
]
1
II
an ultra-violet light source.
0-160
i80
120
60
80
40
20
0
Breakdown of 6.25 centimeter spheres
140
i00
200
ABSOLUTE PRESSURE IN POUNDS PER SQUARE INCH
spaced one centimeter apart in 10-C
transformer oil is plotted here for comparison. The oil breakdown tests were 1/2 inch square rod gap was therefore Figure 5. 60-cycle and impulse breakdown
made at atmospheric pressure only. The chosen for test electrodes to characterize between one-half-inch square rod gaps spaced
transformer oil had a 60 cycle breakdown the more severe conditions of non-uni- six centimeters in nitrogen, CCI2F2, and transformer oil
of 33 kv in a standard oil test cup with form field encountered in high voltage
Curve A= impulse breakdown in oil
0.1 inch gap spacing. The impulse apparatus.
60 cycle and impulse breakdown tests Curve B=60-cycle breakdown in oil
strength of transformer oil with a 11/2-40
microsecond wave is slightly over two were made on this 1/2 inch rod gap Curve C=60-cycle breakdown in nitrogen
times its 60 cycle dielectric strength, mounted in a vertical plane in the high Curve D=impulse breakdown in nitrogen
while the impulse and 60 cycle break- pressure test tank, figure 1. Tests were Curve E=60-cycle breakdown in CCI2F2
down strength of spheres in the high pres- made with both 60 cycle and 11/2-40 Curve F=impulse breakdown in CCI2F2
microsecond impulse voltages in nitrogen
sure gases are the same.
and CC12F2 at various pressures and gap
spacings. Figures 3 and 4 give the results
Rod-Gap Tests
Figures 3 and 4 show how the breakdown
of these tests. The corona starting volt- Figu
reand wih thesbreakdown
Requirements of commercial high volt- age shown on figures 3 and 4 was taken voltage increases with pressure up to a
age apparatus other than the electrical by means of a cathode ray oscillograph criticaliprease an thensereasstithe
insulation strength often dictate that the which received its voltage from a high futherdncrease in pressure until the
conductors adjacent to the insulation be frequency amplifier placed across a shunt breakdown of the gap is only slightly
above the corona starting voltage of the
made of some other shapes than spheres or in series with the test gap
The 60 cycle breakdown of the rod gaps gap. This citical pressure in the case of
curvatures of large radii. The standard
nitrogen is aboult 100 pounds per square
inch gauge, while for CCl2F2 the critical
Table 1. Relative Dielectric Strength of Solid and Fluid Dielectric Arrangements in Transpressure is only a few pounds above at. .
.
former Oil and in Gases Under Pressure
r
mospheric pressure. The critical pres____ _T
CREST BREAKDOWN IN PERtCENTOF N!1 SAMPLE
stines are lower than those found by other
OIL BREAKDOWN
INSULATION SET- UP TYPE OF VOLTAGE
investigatorsl4 on tests with point-plain
10-R
C Ct2 F2
N ITROGEN
PREyIBAR
electrodes.
Goldman and Wul found
/ 0 IN.
200 IS. /SO. I4. 70 LBS. I
| 10-C. OIL GAUGE
E DUCT
d
,@PRESSEOARD
pressure
for nitrogen to be
GAUGE
critical
the
PRESSURE
______E__ll_
PRESSURE.
about
150 pounds per square inch as comPRESSBOARD
TERMINALS
|
CYCLE PUNCTURE
132
100

-7

rO6

FILLPRESSBOARI 5CM.ERMNAL

8(tI0'BY103'

72

77

120

DUCT
JMETAL PLATE
iPRESSIBIAIRI
DUCT
PR.ESSBylOA
|
@
32
| S / S |60 CYCLE PUNCTURE
1
~~5CM.TERMWA^LsIll
I
L
60 CYCLE
@
| PR7EDRD P^Cw |

142

115

143

1.5/40iS. IMPULSE

pared with 100 pounds as shown in this

|60 CYCLE CREAPAGE

.0s18 CPAPERWR

, .O7
OTN
007"tOTTON,
] .018'PAPER*

| Q
1

-ED
OLEL I
DIA.BY
|ELECTRODE
I"

T@

1.5/4O0

+ 4"+
11

WALL

l
|

56

I
|

41

S. IMPULSE

The impulse
breakdown for rod gaps in nitrogen spaced
37
at six and ten centimeters has a negative
between 70 and 120 pounds per
|
I
square inlch, while the impulse breakdown
26
ll
|~~~~~~~~~~~~~~~of
|
rod gaps in CCl2F2 increases steadily
*|
|

|
|

31

||

_ l_
__l_l
l

11
02
100
60 CYCLE
|
|________ i_____ l___ l__

DECEMBER 1941, VOL. 60

The impulse breakdown of rod gaps in


nitrogen and CC12F2 under pressure has
only a remote resemblance to the 60

cycle breakdown

curves.

~~~~~~~~~~~~~slope

|#|ll

48

report.

|
78

Nonken-Gaes as a Dielectric

pressue. Figure 5gives


withincreasing
and 60 cycle breakdown
the

impulse

spaced
~~~~~~~~~~gaps

ofrod

six centimeters apart in


CCl2F2, nitrogen, and in transformer oil.
The 60 cycle breakdown of CCl2F2 iS

about 2.2 times the impulse breakdown at

TRANSACTIONS 1019

6
- -';9- - - - - F atmospheric pressure. It is very unusual
,
400
,
impulse
the
where
condition
a
have
to
I
__ _MaI
/,
less than the 60 o' soo
300 - _
breakdown of a gap is*
cycle breakdown. For example, the im- (< ao
pulse breakdown with a 11/2-40 micro- z
_
second wave for a rod gap in air at
0
.
atmospheric pressure is about 1.8 times 80
704
_
60
the 60 cycle breakdown.
The fact that the im ulse breakQdown is a'lI50 1 tt -l
Igreater than the 60 cycle breakidown for ,,
_0
rod gaps in air has been explained on the ~3
b g h- 2
ATMOSPHERIC RESSURE
basis of time lag before the gap is ionized
in such a manner as to make it highly conducting. The condition in CC12F2 gas o
20 30 40 50 60 70 80 90I-100
-10 0 10 _________
where the 60 cycle breakdown gTeatly > -30-20 TEMPERATURE -DEGREES CENTIGRADE
exceeds the impulse breakdown should Figure 6. Vapor-pressure temperature curve
be worthy of additional study. The
of
reason for this unusual increase in the 60
cycle breakdown is not clear. There is
evidence that the space charge in the gap fluid dielectrics had higher dielectric
is responsible for this unusually high strengths in transformer oil than they did
in nitrogen at 200 pounds per square inch
breakdown.
pressure or CC12F2 at 70 pounds per
Solid-Fluid Dielectric Combinations square inch pressure.

-_

rod

wb

CCIF2

solid and
different
Five
arrangeents of sold
arrangements
dlffernt
Flve
fluid dielectrics were tested for breakdown when immersed in 10-C oil, nitrogen, or CCI12F2. Schematic arrangements
of the dielectrics and a comparison of
voltage breakdowns are made in table I.
Extreme care was taken in preparing the
samples for these tests to see that the oil
samples were completely impregnated and
free from bubbles and that the as-filled

v gas-filed
were thoroughly yvacuum-dried
and kept dry until they were tested. All
of these dielectric arrangements have very
samples

ec sgaps
empea ure Effects
Temperature

High voltage apparatus . filled with com.


pressed gases may be subjected to a wide
by virtue of
range of temperature either
or by the heat

their ambient conditions


liberated within them.
This change in temperature will change

th.islchgin

tee rature will cag


the dielectric strength of the gaseous dielectric in different ways according to the
manner in which the gas is supplied to the

vessel. The following are generalizations


upon theoretical considerations and
based
bechecked experimentally for

CCl2F2

non-homogeneous fields. The dielectric


constants of the paper, pressboard, and tween spheres.
porcelain used in these tests are all two (a). If gas were supplied in a manner such
that a constant gas pressure was maintained,
or more times as large as the dielectric
of the gas would
the dielectric strength
constant of transformer oil. Therefore,
inversely as the absolute temperachange
where the oil and solid insulation are in ture.
series, the oil must stand a dispropor- (b). If the
the total quantity of gas
gas in the sysstrssstrength of
tintl large par
the dielectric
part of the voltage stress.
tionately
tem is unchanged,voltage
Since the dielectric constant of the com- the gas should remain unchanged with
in temperature. This follows from
pressed gases is substantially unity as change
fact that the dielectric strength of a gas
compared with 2.2 for that of transformer the
is proportional to its density.
oil, the voltage across the gas dielectric
similar
a closed system
in the
If the gas isbut
in series with solid insulation is considera- (c).condition
gas is saturated and
(b)
an excess of liquefied gas is present in the
bly more than the voltage across theblyoil tto
if it were made to replace the gas. This system, the gas pressure will change with
condition alone puts the gas dielectric to temperature according to the vapor pressure
of the gas at the particular temperature.
in sesewhen used
used in
a considerable handicap when
strength of the gas will
'PThe dielectric
Wlthsolld
ries dleectrlcs.change
in such a system nearly the same as
Imabeseen from table I that in the vapor pressure of the gas. Vapor presgeneral these combinations of solid and sure of CCl2F2 is shown in figure 6. It is

tota

more

rcoieswithrsold handieleics.

1020

TRANSACTIONS

Nonken-as as a Dielectric

obvious that such a system has much greater


change in dielectric strength of the gas with
than
change
( ang in
a do thee systems
i temperature
r
d (b).
Summary
1. The impulse and the 60 cycle crest
breakdown strength of nitrogen are substantially the same when tested between
spacing; the imspheres at less than radius
and 60 cycle crest breakdown strength
pulse
usad0ylcetradwsrnt
CC12F2 are likewise the same. Nitrogen
of
at 175 pounds per square inch absolute pressure, CCI2F2 at 62 pounds pressure and 10-C
transformer oil at atmospheric pressure
at the same impulse voltage
break
when tested between 6.25 centimeter
spheres spaced at one centimeter. Under
the same conditions the 60 cycle breakdown
voltages of the three fluids are the same with
nitrogen at 80 pounds absolute pressure,
at 25 pounds absolute pressure, and
CC12F2
10-C oil at atmospheric pressure.
2. The fact that there is a condition where
the 60 cycle breakdown of a rod gap in
CC12F2 is over two times the impulse break-

down

down of the same gap under the same condition is disturbing to the orthodox theory
surrounding
in gases.the electrical behavior of spark
The 60 cycle breakdown between rod gaps
is substantially higher in nitrogen above 60
pounds per square inch absolute or in CC12F2
transabove atmospheric pressure than inimpulse
former oil. On the contrary, the

breakdown between rod gaps is considerably

lower in nitrogen under 215 pounds per


square inch absolute pressure and CC12F2
below 85 pounds per square inch absolute
pressure than in transformer oil.
3. The breakdown strength of solid insulations tested in nitrogen at 215 pounds per
square inch absolute pressure and in CC12F2
at 85 pounds per square inch absolute pressure as compared to its breakdown in transformer oil indicates that with but one exception the tests made in transformer oil
showed higher strength than when made in
the gases at the above pressures.

References

COPESEys-OENI
BRADW
NONUNIFORM
FIELD-II, I. M. Goldman
ELECTRIC
and B. M. Wul. Journal of Technical Physics,
USSR, volume 3,1935, pages 16-27.
the

2. BREAKDOWN STUDIES

IN

COMPRESSED GASES

A. H. Howell. AIEE TRANSACTIONS, volume 58

1939 (May section), pages 193-204.

3. DIELECTRIC STRENGTH OF INSULATING FLUIDS


E. E. Charlton and F. S. Cooper. General Electric

Review, volume 40, 1937, pages 438-42.


4. DURCHSCELAGUNTERSUCHUNGEN IN

KOM-

PRIMIERTEN GASEN UND IN FLUSSIGER KOELENSAUR1 p, a Zeer-. Annalen der Physik, volume 14,
,

pages

5. DEE

ELE;KTEISCEE DURCESCELAG VE;ESCEIIEDE;-

HlktONC#EM DIRUCE, E. Finkelmann.


NE ASHv UNTER
282-6.

ELECTRICAL ENGINEERING

Vous aimerez peut-être aussi