Академический Документы
Профессиональный Документы
Культура Документы
Search
Collections
Journals
About
Contact us
My IOPscience
This content has been downloaded from IOPscience. Please scroll down to see the full text.
1984 J. Phys. D: Appl. Phys. 17 421
(http://iopscience.iop.org/0022-3727/17/2/027)
View the table of contents for this issue, or go to the journal homepage for more
Download details:
IP Address: 129.173.72.87
This content was downloaded on 09/11/2016 at 18:18
1. Introduction
Insulating systems invariably contain some microcavities
or voids within the insulation
or on boundaries between the insulation and the electrodes.cavities
The are filled with
a medium, either
gas or liquid,which has a lower electric
strength and dielectric constant
than the main dielectric. This can produce stress concentrations that
may lead to electrical discharges and complete destructionof the dielectric. Theoretical considerations
have been limited to the
well known equivalent circuit model for discharges in gaseous
cavities (Kreuger 1964). However, thismodel fails toexplain
theexperimental
observations.
The mechanism of insulation failure when subjected to partial discharges has been
treated by several researchers (Mason 1951, Garton 1962, Kanazashi 1970). Mason
(1951) and Garton (1962) concluded independently that the breakdown
in internal
discharges is not thermal, butit may be due to ultraviolet radiation or ion and electron
bombardment. Again, Garton concluded that internaldischarges are unlikely to be an
effective ultraviolet radiation source. Mason
(1959) attributed the dielectric damage to
ion bombardment, while Dakin and Hewitt
(1963) reported that electron bombardment
may be more significant. The degree
of damage is then dependent upon the number
of
impinging electrons and their kinetic energies. The destructive discharge energy
within
cavities is proportional to the molecularmass of theenclosed gas and the frequency of
the applied voltage.
0022/3727/84/020421
Instituteof Physics
42 1
422
A A Hossam-Eldin
which is negligibly small. The electron energy would be [l - (2m/M)] of the initial
molecular energy. The term (2m/M) accounts for the elastic collision and is equal to
of the initial energy). If the
(i.e.theenergyabsorbed in collisions is 1.09 X
and the
medium is air, the number of space charges per unit volume is 2.7 x 1019
electron cross-section of interception is
cm2 (Mesytas et a1 1969). The MFP would
accordingly be 3.34 X
m.The relaxationtimewouldbe
1.67 X
S.
If we assume the electron drift velocity to be 2 X lo6m S- (Nasser 1971), and the
time for a discharge is in the range 10-100 ns (Massey and Burhop 1962), with 50 ns
taken as the mean value, the number
of collisions per dischargeis 3 X lo4. Consider the
fraction of molecules involved in an ionisation process to be 1% (Massey and Burhop
1962); then the number of collisions per molecule is 3 X lo2. If the energy lost in each
collision is calculatedtobe
of the initial energy,thentheenergy
lost through
collisions in a dischargeis 0.3 of the initial energy. Mostof the kinetic energystill remains
with the electrons andvery little energy can be ascribed to ionic carriers.
In a sampleof solid insulation containing an internal
gas-filled cavity, when the time
( V , ) ,the gas in the
dependent applied voltage is just higher than the inception voltage
cavity is a mixture of charged and neutral particles. The concentration
of charged
particles is small compared to the neutrals, but the kinetic ofenergy
the chargedparticles
is much greater than that
of the neutrals. Therefore
collisions canbe considered asif the
neutrals are at rest.
Suppose the distribution function for electrons
is F ( r , p , l). We can now expressthe
statistical electron densitysuch that the total numberof electrons n is given by
r
- -VV,F-
eE(w)V,F
E ( w ) is the
423
where N is the number of randomly located neutral particles, and Qe is the elastic
cross-section of interception for electrons,
which is givenby the expression
(5 1
6r = ( P / m ) 6t
in a time 6t. A fraction of them ( Q e N 6 r )will experience collisions. When collisions
occur, electrons change direction
of momentum andmay be absorbed, thereby
vanishing
SV, . This results in a rate
of loss of F. The rate
of absorption
from the elemental volume
is ( Q a N 6 r ) .Collisions in other momentum elements will result in electrons scattering
into SV,, and this results in an increase in F. Expressing this increase in spherical
coordinates, the electrons scattered into the momentum element are
~4S.7r F ( r , p , t ) 6 V P s i n 8 6 ~ 6 H
We consider that thereexist
F ( r , p , t ) ( p ) *sin 06q6pf68
electrons in the momentum elementSV;, which is equal to
P* sin
e 6p 6 8 6 q
asp =p.
The total rate
of change of Fdue toelectrons scatteringin and out, and the electrons
absorbed in the momentum elementSV, , is
-F(r,p, t ) =
at
-(Qe
+ QdNGrF(r,p,
+6r
4n
F ( r , p , 8, I@, t ) sin 8 6 8 6 q .
(6)
In steady-state conditions, the expressions in equation (2) and (6) are equal. Furthermore, due to symmetry
of the functionFin thespherical coordinates, the qdependence
can be integrated out in expressions (3), (4) and (6). The resulting expressions for
424
A A Hossam-Eldin
f ( r , p ,t ) = 2 m
F ( r , p , 0, t ) sin 0 8 0
P F ( r , p , 8, c) sin 0 6 0 .
(10)
where V ,F is in the direction of P * , which is the mass flow vector and taken to belying
in the direction of the polar axis. Since Qa 4 Qe, then
Qe
+ Qa
Qe
(13)
where P and P are theinitial and final momentum andais thescattering angle.
If weexpand thelast expression between the brackets
of equation (12) in the Taylor
series, substitute equation (11) into equation (12), then add toit a function to express
the source functionq (which gives the numberof electrons createdby ionising collisions
per second in the elemental volumeSV,,) we obtain the rateof change of (f)and ( P )
due tocollisions, drift, acceleration, absorption and electron creation. Thus the electron
425
df
- eE( W ) COS 8-
JP
(F)f
a( a,p ) (1 - cos a)sin a 6a
meE(w)
aP
cos2 eJP
P
+ qf
meE(w) P cos2 e.
P2
a( a, p ) (1 - cos a) sin a 6 a
N a
( ~ 4 f ) 2 x j ? o ( a , p ) ( l - cos a) sin ada
MP2 ap
0
- --
JP
aP
P '
This can beseparated into two equations
in terms of the directionsp and8:
2 n N P c o s 8 [ / 3 7 o ( a , p ) ( l - c o s a ) s i n a 6 a + e E ( w ) c o s O -d=f O
JP
and
- meE( W )
aP meE( W ) P cos2 e
cos2 e- aP
From equation(16) we obtain:
P
af = -2nNpcos B[j)(a,p)(l
aP
P2
= 0.
3N
~4[2n[+,~)(1mMeE( U )
426
A A Hossam-Eldin
Substituting equation(19) into equation(H),and knowing that the valueof the termin
square bracketsis independent of the electron momentum, then
fb)= A exP(-W4)
(20)
where
3 m N[2nJ$a(a,p ) ( 1 - cos a ) sin a Sa]
2meE( W )
B=-(
M
The constantA can be obtainedby integratingfb) and equatingit to the total number
of electrons n as given inequation ( l ) .The average kinetic energyof an electron can be
calculated by means of equation (20) as
The second term in brackets is the amount of energy gained by an electron from the
external appliedfield as the electron travels a distance equal to the mean free
of thepath
ionised gas inthe cavity.
It can be seen that the energyof an ionisation discharge in a cavity is initially carried
almost entirely by the electrons. Since the time of discharge is very small ( m ) , the
formative time depends on the number
of electrons initiating breakdown by the startof
development of electron avalanches. These avalanches are initiated
by free electronsin
the cavity. It may be assumed that the
positive ions in the discharge never acquire
appreciable energy, and consequently can never be responsible for bombardment damage of insulation.
From the foregoing analysis it is clear tht themain contributing factor to discharge
damage is the average kinetic energyof electrons impinging on thewalls of the cavity.
The energyof the incident particlesis not high enough to permanently change
chemical
a
bond. It is well known that the lifetime of insulation is inversely proportional to the
average kinetic energy of electrons. Since this energy is frequency dependent, the
insulation lifetime is inversely proportional to the frequency of the applied voltage.
These findings agree very well with experimental studies (Dakinet a1 1954).
From equation(21) it can be seen that the
lifetime of insulation can be increasedby
filling the cavity with a gas of low molecular mass, such as hydrogenor helium, instead
of air. Using thefollowing formula:
Lifetime in gasMolecularmass
of air 112
Lifetime in air = (Molecular mass of gas ) .
Table 1 compares the lifetimeof some gases at 2 atm pressure to that
of air. It has been
reported that the inception voltage
is about the same for
all cases at high pressures (Hall
and Russek 1954).
These findings agree well with experimental finding (Okamoto and Ikeda 1965).
Similar results can be obtained by filling the cavity with a dielectric which has a higher
427
2
4
4
2.74
Nitrogen
Oxygen
28
1
32
0.97
electric strength than the insulating medium. This agrees with the established experimental work on impregnated paper insulation.
Experiments have shown that
filling the cavities with impregnating compounds
(Hossam-Eldin andSalvage 1977) increases theeffective life of the insulation.
By correlating these results
with the experimentalfindings (Hossam-Eldin 1977), the
mechanism of breakdown in solid dielectrics by internal discharges can be explained in
the light of the previous analysis. The effect of impinging electrons is a gradual erosion
and localisation of the discharges at a particular site. The rate
of deterioration increases
with an increase in concentration of discharges. This is due to the high energy of the
impinging electrons.Thisresults
in deep localerosionandaninstabilityinthe
dielectric-gas interface. This instability depends upon the discharge channel and may
result in material removal,leaving a submicroscopic channel or pit, and deposits in other
places on the surface of the cavity walls. This phenomenon may lead to subsequent
discharges to the same site. This
is analogous to entering the material
with an extremely
sharp-pointed electrodewhich has a high stress concentration at its tip. With sustained
discharges, and their by-products, these damaged regions become enlarged and even
visible when viewed by an electronscanning microscope (Hossam-Eldin1977).
4. Conclusions
It is concluded that the densityof damage is dependent upon the numberof impinging
electrons, their average kinetic energy and the repetition rateof discharges. The discharge energy is predominantly carried by the moving electrons, which means that the
damage due to electrons bombarding the surface
of a dielectric is greatest.
It has been shown theoretically that
kinetic
the energyof these electronsis dependent
on the frequencyof the applied voltage. Thus the damage
in insulating media is directly
proportional to the frequency and magnitude of applied voltage, and the lifetime is
inversely proportional to the frequency
of the applied voltage.
References
Dakin T W and Hewitt G W 1963 IEEE Trans. Power Appar. Syst. PAS-82 1033-9
Dakin T W, Philofsky H M and Divens W G 1954AIEE Trans. 73 155-9
Garton C G 1962 The Energy of Discharges and Their Interaction with Solid Dielectrics. Gas Discharges and
Electricity Supply Industry (London: Butterworths) pp 412-9
Hall H C and Russek R M 1954 Proc. IEE 101 47-56
Hossam-Eldin A A 1977J . Engng. Sci. (Saudi Arabia)3 33-7
Hossam-Eldin A A and Salvage B 1977 IEEE Trans. Electr. Insul.EI-12 417-23
A A Hossam-Eldin
Kanazashi M 1970 Deterioration of Polymer Films by Corona Discharges: IEEE paper no 32C79-88 pp
272-6
Kreuger F H 1964 Discharge Detection in High Voltage Equipment(London: Heywood)
Mason J H 1951 Proc. IEE 98 44-59
- 1959 Progress in Dielectrics v01 1 (London: Heywood) pp 3-58
Massey H S and BurhopE H 1962Electronic and Ionic Impact Phenomena(Oxford: Oxford University Press)
Mesytas G A. Bychov Yu I and Iskoldskii A I1969 Sou. Phys.-Tech. Phys. 13 1051-5
Nasser E 1971 Fundamentals of Gaseous Ionizationand Plasma Electronics (New York: Wiley Interscience)
Okamoto H and IdekaY 1965 Electr. Engng. Japan85 1-11