IBEE Transactions on Electrical Insulation
Vol. 28 No. 1, February 1003
Power Frequency and Impulse Field
Calculation around a HV Insulator
with Uniform or Nonuniform Surface
Pollution
S. Chakravorti and P. K. Mukherjee
Blecticel Enginering Department,
Jndaepar University, Calestta, India
ABSTRACT
HY outdoor insulators are always subjected to pollution of dif-
ferent nature and severity and the performance of a HV insu-
lator under polluted conditions is quite different from that un-
der pollution-free conditions. The present paper is concerned
with the fleld distribution around a post type insulator for dif
ferent severities of surface pollution. The capacitive-resistive
field distribution around the polluted insulator is ealeulated
by a charge simulation method. The fleld calculations are car-
ried out for power frequency voltage and also for lightning and
switching impulse voltages. The effects of uniform or partial
surface pollution and the presence of dry bands on the field dis
tribution around a polluted post insulator are studied in detai
and are presented.
1. INTRODUCTION
V ingulators usually serve as support or spacer of
clectrodes with respect to the grounded frames or
ground plane, Generally, in practice, HV insulators are
rounded by gaseous dielectrics and are stressed be-
tween two electrodes. In all the applications, a knowledge
of the electric field distribution around the inzulater is
vety important for the evaluation of the expected perfor-
‘mance of the ineulator. Moreover, a better understanding.
‘of the behavior ofthe insulators in use is necessary to as
sure a high degree of relisbility in operation of the HV
stems
Mukherjee and Roy [1] calculated the feld diatribu-
tion in and around a dise insulator of simple geometry
by charge simulation method (CSM) [2]. Subsequently,
Khan and Alexander (2) did very accurate charge sim.
lation modeling of a practical disc insulator geometry.
and Iravani etal [8] applied CSMf for
the field calculation around a string of diee insulators of
f transmission line in a practical environment. Keana-
[Nkusi et al. [6] accurately determined the field distribu-
tion around a section of a practical port type insulator
bby CSM. However, all these researchers computed the ca-
pacitive field distribution under pollution-ree conditions
But in outdoor applications, the HV insulator surfaces
are exposed to pollution of different nature and severity,
[T]. The performance of a HV insulator under pollution-
free condition is quite diferent from that under pollut-
ed conditions. ‘This is because for a polluted insulator
the field distribution is capacitive, capacitive-resstive,
or resistive, depending upon the severity of the pollu:
tion, while for @ pollution-free insulator the field i pure-
ly capacitive, ‘Therefore, itis very important to know
the changes in the field distribution around an insulator
caused by different severities of surface pollution.
Abdel-Salam and Stanek [8,9] computed the field die-
tribution around e HV insulator having simple geometry,
considering a uniform pollution layer on the insulator aur-
face. Singer (10] determined the effect of uniform surface
pollution on the field distribution around a porcelain sup-
port insulator.
‘The present paper examines the effects of uniform and
nonuniform surface pollution onthe field distribution around
0018-9367 83.00 @ 1993 IEEE.“4 Chakravorti et al Power Frequency and Impulse Field Insulator with Pollution
‘2 port-type porcelain insulator. In the present work, the
technique proposed by Takuma etal. (1, 12] is employed
for the field caleulation including surface resistance by
CSM. Since the capacitive.resatve feld distribution is
dependent on the frequency and waveform of the applied
voltage, the field calculations are carried out not only for
power frequency voltage but alo for lightning and switch.
ing impulse voltages. The potential distribution and the
clectrie stresses for such conditions ate studied in detail
In addition tothe effects of uniform surface pollution, the
‘effects of partial surface pollution and alao the effects of
the presence of dry bands on the field distribution ate
‘computed under both power frequency and impulse volt-
‘ages and are presented in this paper.
ALL OMENSIONS 1H mm
Figure
Poat-type insulator stresed between tno electrodes.
2. POST-TYPE INSULATOR
NHB ingulstor-electrode arrangement under considers
tion for field computation is shown in Figure 1. The
post insulator is made of porcelain, having a relative per-
rittivity of 6. The insulator is surrounded by air and is
stressed between a pair of electrodes. The system under
study isa two-dielectric, rotationally symmetric configu:
ration, The bottom electrode is taken as grounded while
the applied voltage to the top electrode is considered to
bbe of unity magnitude, The uniform pollution layer is
simulated by a uniform surface resistivity along the ineu-
lator, while nonuniform surface pollution is simulated by
considering different resistivities at different locations on,
the insulator surface.
The resistivity of the pollution layer is mainly depen.
dent upon the type of pollutants, e.g. industrial pollution,
‘marine pollution etc. and the weiness of the pollution
layer. Purther, the resistivity of dry pollution layers very
high, while the resistivity decreases heavily when the pol-
lution layer is wetted (13). Hence, the field calculations
are cartied out for different values of surface resistivity,
ranging fom 10"? to 10° 0, to simulate different degrees
of severity of surface pollution, from low to very heavy
pollution
3. SIMULATION TECHNIQUE
1HE algorithm developed for ealevlating the field dis-
tribution around the post-type insulator i based on
the CSM [2]. Several papers [11,12,14-16] have been
published on the analyses of eapsctive-resiative fields. It
had been observed (8, 11,12] that CSM is more euted to
such problems then other methods of numerical field cal:
culation,
Since the insulator-electrode assembly isa rotationally
symmetsic configuration, ring charges (2}are employed for
simulation, Each electrode ie simulated by a ret of ting
changes placed within the electrodes. ‘The porcelain-air
interface is simulated by two sets of ring charges, one set
placed in porcelain and air, respectively. Images of the
fictitious ring charges with respect to the ground plane
are also considered in the present algorithm. The set of
linear equations to be solved for the unknown fictitious
changes, under power frequency and impulee voltages, ace
obtained from the boundary conditions as given below.
‘These equations are then solved for the unknown fictitious
charges. ‘The potential and the electric stresses at any.
point can then be ealeulated from the combined elfet of
the appropriate fictitious charges.
3.1 POWER FREQUENCY
ALGORITHM
For poner frequency capacitive-resistive field caleula-
tion, complex fictitious charges varying sinusoidally withtime are employed to give the instantaneous feld distribu.
1 in the steady state 1,12). ‘The boundary conditions
with the complex quantities are as follows.
Diichet’s condition onthe electrode surface
4) = a)
where 4(i) is the calculated potential at any point é, and
V is the known electrode potential. A bar on a variable
represents a complex quantity
Potential continuity condition on the dielectric inter
face
Boi) = $468) @)
where the subscripts D and A represent porcelain and
Continuity condition of Dy on the dielectric interface
Deo(i) ~ Deals) = (8) @
where Dz is the normal component of the electric fux
density and o is the surface charge density.
In Equation (2)
1_[a-1)
1)
28) = FasTy LRG
RED
6)
where isthe angular frequency, I the surface resistance
‘and 5 is the surface area. They are given as follows
n= [ 6)
Sti) = joa (6)
where 1 ie the radial coordinate of dl along the surface
and p, isthe surface resistivity.
3.2 IMPULSE ALGORITHM
For capacitive-resiative field calculation under transient
voltages, wo methods had been proposed by Singer (10
‘and Takuma et a. (11), respectively. Of these two meth:
‘ods, the algorithm proposed in [11] ean be extended read-
ily to any transient voltage waveform and hence is used
in this paper for impulse feld calculation, In this meth-
‘od {11}, real fictitious charges ate employed and the feld
distribution is calculated by dividing the total time spen
into short time intervals, Hence, for impulse field caleu-
Tation, Equation (1) and (2) can be applied without the
Vol. 28 No. 1, February 1903 46
complex treatment. However, Equation (3) is modified
as follows
Dyp(i)s— Deals =o(i)s at £=Aty (7)
where
oti = 24 Dr Ai): a4
Y= 5 RG RGD
(6)
"The subscript 1 represents the time instant ¢ = ty. The
unknown charges can be obtained from Equations (1),
(2), (7) and (8) at €= Aty. Subsequently, at t = At, +
Ale,
ofile
x RG RU+ 1)
where o(i)s is known from the previous time step. Thi
iterative sequence gives the capacitive-esstive field die-
tribution for the impulse voltages at any time instant
‘The values of At ean be chosen differently for different
time steps. However, for simplicity these may be made
the same forall the time sepa,
3.3 NATURE OF FIELD
DISTRIBUTION
From Equations (8) and (4), it is evident that the field
is capacitive when the surface charge density @ is low
and is resistive when o is high. For intermediate values
of o, the field i capacitive-resttive. Again, the value of
1 depends upon the product of frequency and surface re-
sintvity ps an given by Bquations (4) and (5). The value
of @ decreases with increasing frequency for a given
Hence, for low frequency voltages, eg. power frequency
voltage, the values of ps for which the field will be ea-
pacitive or capacitive-resistive or resistive will be higher,
compared to that at high frequency voltages.
Since the effective frequency of impulse voltage is much
higher than the power frequency, the values of py for
‘which the fld will be capacitive or capacitive-resstive or
resistive will be much lower for impulse voltage. Simi
1y, for lightning impulse voltage the corresponding values
of py will be lower than thooe for switching impulse volt-
age, because the effective frequency of lightaing impulse
is higher than that of ewitching impulee voltage
Field calculations are carted out to identify the range
of g, within which the field distribution changes from
capacitive to capacitivesesistive and from capacitive.
resistive to resistive for power frequency, lightning and40 Chakravorti et al: Power Frequency and Impulse Field Insulator with Pollution
switching impulse voltages, The field distribution for dif.
ferent values of py within such identified ranges are de-
termined for different types of applied voltages.
4. POWER FREQUENCY FIELD
DISTRIBUTION
(ompuTarions ate cartied out for sinusoidal 50 He
jpplied voltage. ‘The potential distribution and the
electri stresses are calculated along the insulator surface
from the HV electrode to the ground electrode for uni-
form as well as nonuniform surface pollution. The electric
stresses are calculated on the ait side of the porcelain-
air dielectric interface from the point of view of surface
flashover and are presented in a normalized formal, Le
V/m/V, for ready reference
10! 9,
4 curve 1 - 10° 5 9
Praia
i areas
|
g
j
OO ep
cited $80
‘long the insulator surface =
Figure 2
Potential distribution along the insulator surface
with uniform pollution lnyer under power frequen
cy voltage
4.1 EFFECT OF UNIFORM
POLLUTION
Computations are carried out for different values of
i to determine the effect of eeverity of surface pollue
tion on the feld distribution around the insulator. From
the results of computations, it has been found that for
2 > 10D, the field is capacitive, while for p, < 10°R,
i ia resistive. For intermediate values of ps, the field
is capacitive-resistive, Figure 2 shows the potential dis-
tribution along the insulator surface for the above men-
tioned three types of field distribution. Figure 3 shows
Resultant stress in V/m/¥ —~
ae
‘Mong the insulator surface =
Figure 3.
Vasintion of By along the inealator surface with
‘uniform pollution layer under power frequency
voltage
Table 2.
Highest values ofthe stzeses and their location
con the invaltor surface for uiform pllotion wn
der power frequency voltage
Ps | potential | Stress | Highest Yalue| Location]
ina distribution in av | in Fig
15.15 3
Ht | capacitive wn
11.78 7
Tesistive | 21.78 7
e109 | capacitive | & 1.83 1
& aar 7
Te 7
P| Resistive |e 545 a
5 7% 1
the variation of the resultant stress Bip along the insu
lator surface for capacitive and resistive fields. Table 1
prevents the highest values of Bp, the tangential stress
By, the normal stress Ey, and their locations on the in-
sulator surface for the three types of field distribut
From ‘Table I it can be observed that on the insulator
turface the highest values of Eq and Ey occur in the re-
tistive field while the highest value of Ey occurs in the
capacitive field, It can be observed also that the highestIBEE Transactions on Electrical Insulation
values of Bp and By for the resistive field is nearly twice
the corresponding values for the capacitive field and these
ttreses occur neat the tip of the topmost shed.
‘The normal stress at the insulator surface is respon-
ible forthe deporition of the surface charges, while the
tangential stres is responsible for the movement of the
turface charges in the direction ofthe electrodes. Further,
the resultant electric stress adjacent to the insulator tu
face is responsible for the production of charged particles
by ionization processes. Hence, if the resultant stress is
increased, in the region near the electrode, as inthe case
of polluted insulators, the production of charged pacti-
cles also is increased. It is known that the onset of the
tutface fashover of the insulator occurs at a crtical leak-
tage current [17]. Hence, a higher Ep near the electrodes
will requite lower applied voltage to cause that critical
leakage current, resulting in lower withstand voltages in
the case of polluted insulators.
Hy 3
\ curve 2 ~ Uniform g
\ pollution with p, = 10°,
\\ Curve 2 ~ pollution free.
\ curve 3 - Partial
‘pollution from ¢ to 7,
os}
Potential in V —=
phitthente tet
‘along the insulator surface —~
Figure 4
fect of patil polation on potential dstibu-
tion along the intulator surface ender power fre
quency voltage
4.2 EFFECT OF PARTIAL
POLLUTION
For the field calculation with pastial pollution, ps isnot
taken constant throughout the intulator surface but
considered diferent at different locations. In the present
‘work, in the polluted section ofthe insulator surface 2,
taken at 10°, which corresponds to resistive field, while
in the pollution-free section py i taken as 1019, which
corresponds to capacitive field. Several cases of pactial
Vol. 28 No. 1, February 1003
a
“3
curve 1 - Partial
g pollution from ¢ to 7.
£ curve 2 - Uniform
e
> pollution with, = 10°
s
pe
A
i
FR et
‘Along the insulator surface >
Figures.
[ect of partial pollution on variation of By along
‘the insulator sorface under power fequency vlt=
fon the naalatr surface for pail pollution under
poner fequency voltage
Stresses va
Polluted
one & | & | &
HtoT | Highest value] 24.2 Bat
Location 7 7
Tite | Highest Value] 15.2 | ie | ie
Locatica sf [ou
8 to 2 | Wighest Value! m5
Location 5
CteT | Mghest valu 78
Location 7
TtoTE| Highest Value] zat | 19.2 | 25.05
tok] Location 713 7
18 to 2
surface pollution have been studied and are presented in
‘Table 2. Table 2 shows that if only the top portion of the
uppermost petticoat is polluted, ie, from 4 to 7 as shown
in Piguee 1, the highest values of Bp and Eq and the48 Chakravorti et al Power Frequency and Impulse Field Insulator with Pollution
highest value of Bare slightly lower and nearly 4% high-
er, respectively, compared to the corresponding values for
resistive field with uniform surface pollution, Such high
value of By occurs in the vicinity ofthe live electrode, he
location $ of Figure 1. However, the worst case is found
to be when the pollution of the section of the insulator
surface from C to 7 takes place. ‘Then the highest value
‘of By is more than twice and the highest values of En
and By age about 1.6% the corresponding values for the
resistive field. Figure 4 showe the potential distribution,
for partial pollution of the insulator aurfece from C to 7
in comparison with the capacitive and the resistive po=
tential distribution, while Figure 5 shows the variation
‘of Ep along the insulator surface for this case of partial
pollution, along with the resistive variation. The results
clearly indicate that partial pollution ofthe insulator su.
face near the eleczodes is more hartnful to the polluted
ingulators than the uniform surface pollution
4.3 EFFECT OF DRY BAND
{tin generally recognized that very often failure of pol-
luted outdoor inaulatore occurs due to the formation of
dry bands in the pollution layer on the insulator surfaces,
Gellert ef al, [18} did some work on finite element mod.
ling of dry bands on polluted insulators. In the present
paper, an attempt has been made to study in detail the
effects of the dry bands on the field distribution around a
polluted insulator, ‘The dry banc is simulated by a one
of very high resistivity on the uniformly polluted surface
‘of a low resistivity insulator. ‘The width of simulated dry
bhend can be varied easly by varying the width of the
zone of high resistivity. Since a dry band has two edges,
the edges towards the live electrode and the ground elec
trode are called the ‘leading’ edge and the ‘trailing edge’,
respectively. Studies have been carried out for narrow at
well as wide dry bands. ‘The effect of locations of ry
‘bands is also studied
Table 3 shows the values of Bp and By at the edges
and also the percentage of the applied voltage appearing
‘across the narrow (5 mm) dry bands for different locations
ofthe dry bands. Table 4 presents similar results for wide
dry bands (10 mm) at different locations. The results
clearly show that exceptionally high stresses occur at the
erdges of the dry bands and that the electric stresses at the
edges are dependent on the width of the dry bands. The
wider the dry band, the lower are the stresses at the edges
of the dry band. It ie alao found that the location of the
dry band doce not have a strong influence on the electric
stresses at the edges. Table 3 and 4 also show that the
potential appearing across the dry band is dependent on
the width and the location of the dry band. ‘The wider
Table 3
Strona at the edges of 6 mim dry bends under
Power frequency voltage,
ary band] 2 ot. | Type of] Ragitude in Vial
Location} across | stress
ary ban ‘eating ee] Trailing eta]
2 a | & 9.8 18.9
q oo 115.3
+ a | & 705.5 omy
a 1745 a3
’ wl oe 180.2 88
a 29 BS
8 a7 | 1.8 154
ft 128.2 13
w w | of 12 AS
& MBS 48
Table 4
‘Stress the edgen of 10 mam dey bands under
power fequency voltage
ary sane] x fot. | type off Magitute in Waar
Lecation| across | stress
fry band Leading ete Trailing edge
zn |& a5
& ray
+ fe le 7 ry
& wa we
7 te | & ws rm)
B 3.2 M1
w fam | & 8.7 We
& ma BS
w |e | & 5B
& m9
the dey band, the higher the potential difference across
the dry band. Previous researchere, while working on the
sromth of the dry bands, had assumed that nearly the
‘entire applied voltage appears across the dry band (18,
18), Bub, from the present study it can be inferred thatIEEE Transactions on Electrical Insulation
the growth of the dry band is dependent on ite width and
location.
Curve 1 - Uniform pollution
10 with P= 1088,
curve 2- 5mm dry band at
N, location 4.
[| As curve 3 ~ simultaneous
|S SS"am ary bands at
\,, locations 4 and 10,
| <
9]
Pat paste ey
‘Along the insulator surface —~
Figure 6
fect of dry bands on potential distribation slong
the insulator surface under power frequency volt
{curve 1- Smm dry band at
‘ location 4,
curve 2 - Simultaneous Sam
i dry bands at
‘" ocations 4 and 10.
Resultant stress in V/m/V
5
3
TH SHE oe te
‘Mong the insulator surface >
Figure 7
fect of dry bands on variation of Bx along the
inguletor surface under power fequency voltage
Studies have also been carried out to determine the
‘effect of simultaneous occurrence of dry bands, ‘The sim-
ulation algorithm does not put any limit on the locations
Vol. 28 No. 1, February 1903 40
Table 5,
Stresses at the edges of simultaneously occuring
‘S mm dey bande under pomer frequency roltage
Simul | Ory band] Type Wageitude in Via¥
taneous} Location] stress
ory bane Leading ete] Trailing etal
7 | & 106.7
E 16.5
zee
. | & at
& 28
2 Met
a2
& 8.8 2.8
& 8.8 a4
2 | &
&
2am
~ | & ra oI
& 45 8
- 7 &
&
vee
wo | & 4 Th
B 68 a7
2 | & oy
a mm 33
zenl om | & ry oT
v2 & wu 3.
mu | 5 a4
na Bul
E
Ee
‘or on the number of dry bands tobe simulated simmlta-
neously. Hence, fld calculations are carted out with 2
5 or more dry bands occurring simultaneously at differ
feat locations. Table 5 shows thin effect of 2 and 3 dry
bands. Comparison of Table § with Table 3 shows that
the strees at the edges of simultaneously occuring dry
bands are lower than those for single dry band, Since
itis found that the simultaneous occurrence of dry bands
is less onerous for the polluted insulator, the renuts for
larger numberof dry bands occurring simultaneously ate
not presented inthis paper. Figures 6 and T show the po
{ential distribution andthe variation of By along the i-50
sulator surface for a narrow dry band at location 4 of Fig-
le Land also for simultaneous occurrence of dry bands at
locations 4 and 10 of Figure 1, Figure 6 clearly indicates
the percentage of the applied voltage appearing across the
dry bands, while Figure 7 vividly shows the much lower
clectriestreses at the edges of simultaneously occurring
Ary bands compared to a single dry band.
5. IMPULSE FIELD
DISTRIBUTION
(oMPUTATIONS ate extended to investigate the field
distribution around a polluted insulator under both
lightning and switching impulse voltages. The impulse
voltages are represented by a double exponential wave:
form as given below
¥ (4 = Alex(-at) ~ exp(-8)]
(an)
For a 1,2/50 ye lightning impulee voltage of unity peak
‘magnitude, A, a and b are 1,037, 0.0147 and 2.47, respec.
tively, while for « 250/2500 us switching impulse voltage
cof unity peak magnitude, these values are 1.0947, 0.00032
and 0.075,
For the field eslevlation, the lightning and switching
impulse voltages as given by Equation (10) are divided
into @ number of suitable discrete time intervals. The
time slepe selected for the wavefront and the wave tail
fare nat equal but are selected judiciously to give fast but
accurate field calculation over the entire time span. As in
the case of power frequency field distribution, effects of
uniform and nonuniform pollution under impulse voltages,
are studied
5.1 EFFECT OF UNIFORM
POLLUTION
‘Computations are performed for wide ranging values of
1, vader both lightning and sitehing impulse voltages
For lightning impulee voltage it has been found that the
field is capacitive for p, > 10%, while the fed in resi.
tive for pr < 108. Por intermediate values of ps, the
field is capacitiveresistive. Figure 8 shows the capac
tive, capacitiveesstive and resistive potential distribu
tions along the insulator surface at 2 us under lightning
impulse voltage. Similar potential distributions at 240 us
Under switehing impulse voltage are shown in Figure 8
Figure 9 shows thatthe eld distribution changes fom
pacitive to resistive for gs = 109 to 107 under switch-
ing impulse voltage. Figures 10 and 11 show the time
Chakravorti et al: Power Frequeney and Impulse Field Insulator with Pollution
29)
pried as ef
‘Along the insulator surface =
Figure &
Potential datibution slong the insulator surface
‘vith uniform ollation layer under lightning im-
pal
voltage
°
curve 1 - 10° 9
2-10° 9
3-108
t= 240 us
CTs seh
‘Along the insulator surface
Figure 9
Potentil distribation slong the insulator surface
with uniform pollution Inger under switching i
Pale voltage.
variation of By at location § of Figure 1 fo diferent val-
‘ues of p, corresponding to capacitive, capacitive resistive
‘and resistive fields under lightning and ewitching impulse
voltage.IEEE Transactions on Electrical Insulation
' ‘
:
g
&
geo a
fic
an Van
i om
30 08
Time in us —=
Figure 10.
Vasiation of Ex with surface resistivity at lo
tion 8 onder ghtning impsee vleage.
12-0} Aina
60
Resultant stress in V/m/V —=
od]
300
time in
7006
we
Figure 21
Vasiston of Bn with surface resistivity at loca:
tion 8 under switching impulse voltage
Comparison of Figure 10 with Figure 11 shows that
the nature of change of the stresses with py are more or
Jess the same for both lightning and switching impulse
voltages. Figures 12 and 13 show Bq at some select-
ed locations for capacitive field and resistive field under
lightning impulse voltage. Figure 13 clesrly shows exces-
Vol. 28 No. 1, February 1003 31
99 0 ong
A108 o
Resultant stress in v/n/v~
ool
oo 30 100
Time in We —e
Figure 12.
By. at some selected locations on the ineulator
ssrface for capacitive feld distribution ‘under
lightning impulse vltege
|
7
s
4160
4 e
8 | conptenceransa= §
@ trae S50
od
oo 30 "00
tine in le
Figure 13.
En at some selected locations on the insulator
turface for reintiv eld dieteibution under
ting impulse rltage.
sively high strese near location 7 of Figure 1 fo32 Chakravorti et al: Power Frequency and Impulse Field Insulator with Pollution
field, similar to the case of power frequency field. More or
less the same peak Bq occurs near location 7 for resistive
field under switching impulse voltage also
‘The analyses of the results show that for moderate pol-
lution severity, eg. = 1070, the feld is nearly resistive
under switching impulse voltage, while it is nearly ca
pacitive under lightning impulee voltage. In other words,
‘witching impulse voltage is more harmful tothe polluted
insulators compared to lightning impulse voltage. Fur-
ther, for low pollution severity, es = 100 the field
in nearly resistive for power frequency voltage, while it is,
nearly capacitive for both impulse voltages. Hence, pow-
cr frequency voltage can be said to be mest harmful to
the polluted insulators from the point of view of surface
flashover
8
Resultant stress ih v/m/V —=
30
Time in
100
use
Figure 16.
Bp at tome selected Locations on the inelator
surface for partial surface pollution from C to T
under lightning impulse voltage.
5.2 EFFECT OF PARTIAL
POLLUTION.
AAs in the ease of power frequency field, the effects of
partial surfece pollution on the field distribution around a
polluted ingulator are studied under lightning and switeh-
ing impulee voltages. It has been found that the electrie
stresses due to partial pollution under the impulse volt.
fages ate more or less the same as those for power fre.
‘quency field. Similar to the case of poveer frequency feld,
the worst case is found to be the pollution of the insu
lator surface from locations C to 7 of Figure 1 and the
7
4
3
5
i
oo x0 ‘oo
tine in ue =
Faw 8
By atthe edges of 6 mm dry band et location 4
fon the insulator surface under Kightning impolse
voltage.
highest value of Bip occurs near the location 7 of Fig-
ure 1. Figure 1d shows Bq at some selected locations un-
der lightning impulte voltage for this worst case of partial
pollution
5.3 EFFECT OF DRY BAND
Studies have also been carted out considering narrow
and wide dry bands on the uniformly polluted surface of
the insulator under lightning and switching impulse volt-
ages. Study of the results of computations shows that
the electric stresses at the edges of dry bands due to im-
pulse voltages ate slightly higher than thoee due to power
frequency voltage under identical conditions and, hence,
oes not merit any special attention. Figure 15 shows Ex.
at the edges of a narrow (5 mm) dry band at location 4 of
Figure I under lightning impulse voltage. This particular
case is presented here as thie is found Lo be the worst case
fs far as stremes are concerned.
6. CONCLUSIONS
ROM the results of computation, the following conclu
jone can be artived at.IEEE Transactions on Electrical Insulation
1. The changes of the field distribution from capacitive
to capacitive-resistive and from cepacitive-reistive to
tesintve oceur within a nazrow range of surface ress.
tivity under both power frequency and impulse volt-
ages,
Resistive field distribution causes unfavorable elect
cal sttestes around a polluted insulator, which in turn
result in lower withstand voltage
For nonuniform sutface pollution, excessively high
stremen occur at the junction of two different a
resistivities,
Partial pollution ofthe insulator surface near the ele
trodes caures higher stresses than does uniform polli-
ion of the insulator surface
Bxtremely high stresses occur at the edges of dry
bands and these stresses are mainly dependent upon
the width, but not much upon the location, of dry
bands.
6, For a given surface resistivity, switching impulse volt-
‘age causes higher stresses than lightning impulse volt-
‘age. However, power frequency voltage of the same
amplitude gives rise to the highest stresses.
REFERENCES
(i) P, K. Mukberjee and C. K. Roy, “Computation of
Fields in and Around Ineulatore by Fictitious Point
Charges", IBEB ‘rans. Blectr. Insul., Vol. 13, pp.
24-31, 1978,
H. Singer, H. Steinbigler and P. Weiss, “A Charge
Simulation Method for the Calculation of High Volt-
age Fields", IBEE Trans. on PAS, Vol. 93, xp. 1860-
1668, 1974,
el
(3) M. J. Khan and PH. Alexander, “Charge Simu-
lation Modeling of Practical Ineulator Geometries",
TEBE Trans. Blectr. Insul., Vol. 17, pp. 325-832,
1982,
[4] 2. Hazander, S. Milojkovie, and I. Kanenica, “Nu-
metieal Field Caleulation of Insulator Chains for
HV Transmission Lines”, paper No. 12.08, 3rd Int.
Symp. on HV Bng., Milan, Aug. 1979.
(s] M. R. Iravani, M. R. Ragbuveer, "Numerical Com-
putation of Potential Distribution Along a Trans-
mission Line Insulator Chain", IEEE Trans. Elect.
Insul, Vol. 18, pp. 167-170, 1983,
(6] S. Kaana-Nkusi, P. H. Alexander, R. Hackam, "Po.
tential and Electric Field Distribution at a HV Ineu-
lator Shed”, IEEE Trans, Electr. Insul., Vol. 23, pp.
307-317, 1988,
[7] P. Clavatie and Y. Procheron, “How to Choose In.
sulators for Polluted Areas", IEEE ‘Trans, on PAS,
Vol. 92, pp. 1121-1131, 1973
Vol. 28 No. 1, February 1903 83
(8) M. Abdel-Salam and E. K. Stanek, “Optimising
Field Stress on HV Insulators”, IEE Trans. Elects
Insul., Vol. 22, pp. 47-56, 1987.
9
M. Abdel-Salam and B. K. Stanek, “Field Optimiza-
tion of HV Insulators”, IEEE Trans. 1A, Vol. 22, pp.
504-601, 1986
(10) H. Singer, “Impulse Stresses of Conductive Di-
electrics", paper no. 11.02, 4th Int. Symp. on HV
Eng., Athens, 5-9 Sept, 1983,
(01) ‘7. Takuma, and T. Kawamoto, “Field Calculation
Including Surface Resistance by Charge Simulation
Method”, paper no. 12.01, 3ed Int. Symp. on HV
Eng, Milan, 28-31 Aug, 1978.
(12) T. Takuma, T, Kawamoto and H. Fujinami, “Charge
Simulation Method with Complex Fictitious Charges
for Caleulating Capacitive-Resitive Fields”, IEEE
‘Trans. on PAS, Vol. 100, pp. 4865-4672, 1981.
[13] K. Cheean, Z. Pobl and ‘T. Kowalak, “Hygrotcopic
Properties of Pollutants on HV Insulators”, IEEE
‘Trans. Blectr. Insul., Vol. 24, pp. 107-112, 1989
(14] B, Bachmann, “Models for Computation Mixed
Fields with the Charge Method”, paper no. 12.05,
rd Int, Symp. on HV Eng, Milan, 28-31 Aug.,
1979,
(15] 0. W, Andersen, “Finite Element Solution of Com-
plex Potential Electric Fields", IEEE Trans. on PAS,
Vol. 86, pp. 1156-1161, 1977.
{16} A. DiNapoli and C. Masretti, “Electrostatic and
Electromagnetic Field Computation for the HV Re-
sitive Divider”, IEEE ‘Trans, on PAS, Vol. 98, pp.
197-206, 1979.
(17] M.S. Risk and R. Hackam, “Performance Improve.
‘ment of Insulators in a Gas Insulated System”, IEEE
‘Trans. Blectr. Insul., Vol. 22, pp. 439-446, 1987.
[18] B. C. Gellert ond J. K, Rasmussen, “Finite Element
Modeling of Dry Zone Formation on Polluted Out-
door High Voltage Insulators”, paper no. 24.07, 6th
Int. Symp. on HV Eng., New Orleans, 28 Aug-sept,
1989.
[U9] J. 0, Loberg and E. C. Selthause, “Dry Band
Growth on Polluted Insulations”, IBEE Trans. Blee-
tr, Inul., Vol. 6, pp. 138-141, 1971.
‘Manuscript was received on I Mey 1992, in revised form 30,
‘August 1992,