You are on page 1of 13

810

IEEE Transactions on Power Delivery, Vol. 12, No. 2, April 1997

'

~ o d ~ lofi Alt~tude
~ g
Effects on AC lashover of ~ ~ l l u t e digh Voltage ~nsulators
Farouk A.M. Rizk
Fellow IEEE
Institut de recherche d'Hydro-QuCbec (IREQ)
Varennes, QuCbec, Canada J3X 1S1

Abstract - The Paper i n t l " z j a new Physical


approach to account for the effect of reduced air density
on the flashover voltage and critical leakage current of
polluted high voltage insulators. The analysis starts by
updating the mathematical model, previously established, of power frequency flashover of polluted insulators at normal atmospheric pressure. It then proceeds to
introduce the effect of ambient pressure on the physical
parameters of the dielectric recovery equation. The
effect of reduced pressure on the arc boundary radius is
investigated. The combined effect of humidity and
reduced air density on the dielectric strength at ambient
temperature is also accounted for. The above analysis
results in a new expression for the reignition voltage
which includes ambient pressure effects.
The analytical findings are then used to investigate
the effect of reduced air density on the critical leakage
current and flashover voltage of simple-shaped polluted
insulators. The effect of more complex profiles is subsequently introduced.
The model results are compared with experiments
and the agreement established is quite satisfactory.
Finally simple practical altitude correction factors for
polluted insulators are proposed.

Several investigations? mostly experimental, have


already addressed the problem of altitude effects on
pollution flashover of high voltage insulators [l-91. For
a given severity, measured in terms of the specific layer
conductivity os and ambient pressure p atm, the

A.Q. Rezazada
Non-member
General Motors of Canada Ltd.
London, Ontario, Canada N6A 4N5

critical flashover voltage U,(.,,

1)
terms of the corresponding value at 1 atm, U, (os,
by:

U c ( o s , p ) = K d Uc ( 0 ~ 7 1 )

(1)

where Kd is an air density correction factor which, in


general, varies with the type of voltage stress, insulator
profile and pollution severity.
Kd has often been expressed in the form [8991:

where p is the ambient pressure in atm and the


exponent m depends in general on the type of voltage
stress, insulator design and pollution severity.
During laboratory tests, it appears that m also
depends on the mode of voltage application (e.g. gradual rise, constant application, etc.) [9].
Rudakova and Tikhodeev [8] reviewed the Russian
literature on the subject, including field, laboratory and
vacuum chamber tests. It was found that in mountainous regions there is a general tendency for lower
pollution severity the higher the altitude. Furthermore,
it was reported that the dependence of the critical
flashover voltage on severity in the range 2-14pS is
practically independent of ambient pressure, which
leads to the conclusion that the exponent m above is
rather insensitive to severity. It was also found that for
a standard insulator, in the majority of studies, the
exponent m above could be taken as 0.5. The relationship between ambient pressure in atmospheres and the
altitude H in km was taken as [SI:

96
104-0 PWRD A paper recommended and approved by the IEEE
Transmission and Distribution Committee of the IEEE Power Engineering
Society for presentatlon at the 1996 IEEE/PES Winter Meeting, January 2125, 1996, Baltimore, MD. Manuscript submitted August 1, 1995; made
available for printing December 5, 1995.

p ) can be expressed in

p = (1 - H I 44.3)5.25

(3)

This yields an altitude correction formula [8]:

Kd = 1- 0.059 * H

(4)

For insulators with deep ribs and small rib spacing,


the exponent m increases to reach 0.8, as the insulator
performance is influenced by air breakdown between
ribs [SI.

0885-8977/97/$10.00 0 1996 IEEE

E111

Mercure [9] analyzed the results of several


investigations of the dependence of the flashover
voltage of cap-and-pin insulators on ambient pressure.
Ishiis AC results [6] yielded m = 0.50 for standard and
m = 0.55 for antifog insulators. The 50% flashover voltage was determined by the wet contaminant technique,
using the up-and-down method.
AC pollution tests [7] were performed on four types
of cap-and-pin insulators in 3-unit strings in the pressure range 48-101 H a . It was found that m varied in the
range 0.28-0.50 for different pollution levels and insulator types with an average of 0.44. It was also found
the critical AC voltage gradient and the critical current
can be related by [7]:

Ec=C1,a-0.67 p 0.65

(V/cm, A, atm)

(5)

where the constant C varied according to the insulator


type in the narrow range of 345-376 VA0.67/cm.
AC artificial pollution tests [lo] were carried out in
an evacuated chamber in the pressure range 0.671.0 atm and confirmed that for simple shaped insulator
(standard IEEE disc) the exponent m can be taken as
0.5. For a more complex shaped pin type insulator
(NEMA 56-1) the exponent was as high as 0.8.
AC solid layer tests were carried out [ l l ] on a
smooth cylindrical insulator 11.9 cm diameter and
85 cm length as well as 5 other porcelain supporting
insulators in a fog chamber in the pressure range 50101 kPa and ESDD in the range 0.03-0.40 mg/cm2. For
the simple insulator profile m was found to be about
0.40 independently of severity. For other insulators m
varied with severity, being low at low severity and low
again at high severity, exhibiting a maximum in
between. The above tests, however, were carried out
with gradual voltage rise to flashover.
Despite the above experimental work and a few
empirical attempts [ I l l , there is no physically based
mathematical model available to account for altitude
effects on AC flashover of polluted insulators.
A physically based model should be able to:
- determine the dependence of reignition voltage of
an AC arc on ambient pressure, for the current
range of interest to pollution flashover,
- account for the variation of critical distance and
leakage current with air density,
- derive an expression for the critical AC flashover
voltage, at a given pollution severity, as a function
of altitude,
- quantify the dependence of altitude effects on
insulator profile.

A model meeting the above requirements lis


introduced below.

BASIC MODEL
In the model introduced by Rizk [12, 13, 141, the
minimum voltage necessary for AC flashover of a polluted insulator is determined by the reignition criterion
of the residual arc following a current zero of the critical leakage current. The criterion for arc motion lis
taken as necessary but not sufficient and assumed to be
satisfied at a lower voltage as in the DC case. The
problem becomes essentially to solve the energy
balance equation [12] of the residual hot gas, which
starts with a temperature of typically 3000 K at current
zero. Cooling of the residual hot gas takes place by
generalized thermal conduction involving kinetic and
dissociation energies and the effect of convection is
indirectly expressed by the arc boundary radius. Thie
latter is a function of the peak arc current and the quasistatic arc E-I characteristics used as one of the boundary conditions.
Introducing the thermal flux function S due to
Maecker [ 151:
T

S = j 0x d t
where K is the coefficient of thermal conductivity and T
the axial gas temperature, assuming cylindrical symmetry, S can, in the range 300 K-3000 K, be expressed as

S = const T P

(7)

where p is a constant.
The thermal diffusivity k = rdscp ( 6 is the gas
density and Cp is the specific heAt) can also be approximated, again in the range 300-3000 K, by:
k=a*S

(8)

where a* is a constant.
Solution of the energy balance equation [12] yielded
the following expression for the variation of the
minimum breakdown voltage ud of the residual arc
column with time t subsequent to current zero:
U,

= U&

where

{ 1 + (so/ s,

udu

So

- l)/[l

+ 4a*(So - s),

21 r1P

t / r,

is the breakdown voltage at ambient


temperature
is the thermal flux function at t = 0
(T = 3000 K)

812

Sb
rb

is the thermal flux function at ambient


temperature (T = 300 K)
is the arc boundary radius.

The boundary radius was obtained from Maecker's


solution of the energy balance equation of the static arc
1151, at the peak of the AC leakage current.
The critical situation is reached a quarter cycle from
current zero ( t = d2a)when the circuit voltage reaches
the dielectric recovery voltage u d .
For an arc of length x the critical reignition voltage
resulting from the above analysis takes the form:
ucx = x Eda f < i m >

+ rp ( L- x ),i

q, = 0.497 Ie0.663
,
Substituting for U&,
(13) into (12):

(11)

where No and n are constants from the static arc characteristic E = N o , rp is the average pollution resistance
per unit length and L is the leakage path and searching
for the critical point results in the critical arc length x,,
critical current i, and critical voltage U,. As is well
known rp is related to the specific layer conductivity os
by the relationship rp = f /(Lo,), where f is the insulator
form factor.

Revised thermal conductivity of air at temperatures


of 2000 K and above were taken from Ref. [16], while
the more common values in the range 300-1500 K were
obtained from [17].
Numerical integration of (6) yielded the following
values:
at T = 3000 K, SO= 350.8 J1m.s;
at T = 300K, S b = 5.34Jlm.s.
Regression analysis resulted in the exponent in (7):

p = 1.778 and in (8) above U *

= 3.78 x 10-6 m3/J.

Substituting the numerical values in (9):

with t in s and rb in cm.

(13)

5 ms (50 Hz) and

rb from

(14)
Within the range 0.05-1.00 A, regression analysis
shows that (14) could be expressed as

U,, = 716 xli,.OS26

(Vpeak, cm, Ape&)

(15)

This yields values very close to the following expression given by Claverie and Porcheron [18] based on
experimental results:

(16)

U,, = 800 x/&

as shown in Fig. 1 representing the dielectric gradient


Ed = U,, / X .

ik

UPDATING OF MODEL
PARAMETERS

t =

(cm, Ape&>

(Vpeak, Ape&, cm)

(10)

where ,i is the peak leakage current and Eda is the


dielectric strength at ambient temperature.
Combining the reignition equation (10) with the
circuit equation:
U,, = x No / ig

The dependence of the boundary radius on current


was found as:

3000
2500 -

8
.
-3

2000 -

& 1500 -

>

p, 1000-

,\

500 - x MODEL
.
Empirical [18]
I

Fig. 1 Variation of the dielectric reignition voltage


gradient with arc current at atmospheric
pressure in a resistive circuit.

EFFECT
OF PRESSURE ON BASICPARAMETERS
For the limited pressure range of 0.6-1 atm and for
the temperature range of 300-3000K, the effect of
pressure on thermal conductivity of air can be
neglected. Also the initial temperature of the dielectric
recovery period was assumed practically constant at
3000 K. Therefore, the values of SO, Sb and p defined

813

above will remain the same, independent of pressure in


the range of interest.
The specific heat is also practically constant in that
range, while the density is proportional to ambient
pressure. This means that the thermal diffusivity k is
inversely proportional to ambient pressure and the same
follows for the parameter a * which represents the slope
of the k-S line.
The variation of the quasi-static voltage gradient with
pressure is needed both for the circuit equation at critical conditions and for determination of the arc boundary radius, required for the dielectric recovery equation.
With a static characteristic of the form:

At 20C, H, = 17 g/m3 and (20) can be approximated


as :
Ed,(P) = Ed,(l) * p *

BASICEQUATIONS
AT VARIABLE PRESSURE
Introducing the above pressure dependency of thle
different parameters into (9), the dielectric reignition
equation takes the form:

E in = No pmo
there is some uncertainty about the variation of m, and
n with pressure [2, 6, 191.
With m, = 0.2, which is roughly a mean value of the
measurements for arcs in air in the current range
20mA-4A and pressure range 20-150kPa [19], our
calculations showed that:
% Oc P-0.465

The exponent of -0.465 compares with -0.38 determined by Suits [20] for arc currents in the range 110 A.
The variation with pressure of the dielectric gradient
at ambient temperature, Ed,, comprises two factors:
- a simple proportionality to the ambient pressure
- a factor that has to account for the effect of
humidity, which during laboratory tests in a fog
chamber is at saturation.
From IEC Publication 60-1, it follows that the
humidity correction factor Kh for power frequency
voltage, at constant temperature:

:(

Kh = 1 0.012

where

Hu
P

(7

* p0.07))]0'562

(22)
Within the range 0.05-1.OO A and ambient pressurle
range 0.6-1 .O atm, regression analysis yields the muclh
simpler, though approximate, reignition equation:
0.77 -0.526
Ucx(p)= 716 x p
I I,,
(Vpeak, cm, Atm, Ape&)
(23)
At any pressure, the circuit equation takes the form:
Ucx(p)= x No pmoI

ik + r- ( L - x) im

(24.)

Solving (23) and (24) and searching for the criticall


point

(%

= 0) yields the critical arc length xc,

critical current i, and critical voltage U,.

NUMERICAL RESULTS AND


COMPARISON WITH EXPERIMENTS

For simple insulator shapes, where the ratio of the


insulator leakage path L to height h is not high, it is
assumed that at any pressure, within the range of practical interest, the arc follows the leakage path, without
bridging of adjacent ribs or consecutive sheds.

It follows that:

* 1+ 0.012

[1+64.69/(1+ 1.057 I

SIMPLE INSULATOR SHAPES

- 11)

is the absolute humidity, g/m3


is the ambient pressure, atm

Ed,(P) = E&(l) * p

(:
I1
(iL326

UCx(p)=5233x*p* 1+0.2 --1

I)

Critical Voltage

-- 11

[1+0.012 ( H , - 1l)]

(20)

Solution of (23) and (24) above gives the dependence


of the critical voltage per unit leakage length U J L on
ambient atmospheric pressure p , for different pollution

814

severities as expressed by the average pollution resistance per unit leakage length rp as shown in Fig. 2.
For a cap-and-pin insulator with L = 38 cm and form
factor f = 0.75, the above rp range (1000-6000 Rlcm)
corresponds to a specific layer conductivity range of 320 ,US.For a long rod insulator with L = 180 cm andf=
6.3 the corresponding range is 6-35 ,US.
loo0

Regression analysis shows that U c L can be


expressed as:

U, I L = const. rpa p m

(25)

Within the rp range of 1000-6000 Wcm and ambient


pressure range of 0.6-1.0 atm, the exponents a and m
amounted to 0.355 and 0.466 respectively. The model
value of m = 0.466 is in excellent agreement with the
value of m = 0.50 most quoted from experiments on
simple shaped insulators [8, 91.

800 -

Critical current

600 -

>

i
.
5

Fig. 4 shows model results of the variation of the


critical current i, i.e. the maximum current that can
flow without flashover, as function of ambient pressure
in the range 0.6-1.0 Atm, at different severities. It is
shown that the critical current is more sensitive to
ambient pressure than the critical voltage. For the same
severity, the critical current decreases significantly with
altitude.

400 -

200 - x r p = 1000 o h d c m
U rp = 3000 ohmkm
rp = 6000 o h d c m

Fig. 2 Dependence of critical voltage per unit leakage


length on atmospheric pressure, for different
pollution severities expressed in average
pollution resistance per unit leakage length.
Fig. 3 shows the variation of U J L with rp for different values of atmospheric pressure in the range 0.61 atm.

-3a
d

x r p = 1000 o h d c m

0.6

+- rp = 3000 o h d c m

+ m = 6000 o h d c m

.3

0.4

> C

0.2

loo0

0.5

0.6

0.7
0.8
P , Atm

0.9

1.o

Fig. 4 Critical leakage current as function of atmospheric pressure, for different pollution
severities.

+ p = 1.0Atm
+ p = 0.8 Atm
* p=0.6Atm
0
0

1000

2000

3000 4000
rp , ohm/ cm

5000

6000

Fig. 3 Variation of the critical voltage per unit leakage


length with average pollution resistance per
unit leakage length, for different values of the
ambient atmospheric pressure.

Regression analysis shows that for a fixed severity,


the critical current can be expressed as:

i, = const. pmc

(26)

where m, is a constant, which varies slightly with


severity. For example m, = 0.576 at rp = 1000 Wcm
and reaches 0.635 at rp = 6000Rlcm, so that an

815

approximate mean value of m, = 0.6 can be used within


that range.

Critical Distance
Experimental results and mathematical models show
that if the arc burning on a polluted insulator surface
bridges about 2/3 of the leakage path, flashover is practically assured [14]. The effect of ambient pressure on
such critical distance x, is presented in Fig. 5.

In the present analysis it is suggested that the phenomenon of arc bridging and accordingly the efficiency
of utilization of the leakage will depend on:
- the insulator geometry as explained above
- the pollution severity; as it is expected that with
lower severity and accordingly higher voltage
stress arc bridging will be more likely
- ambient pressure; since as shown above the
streamer gradient in air is more sensitive to
ambient pressure than the pollution gradient.
Guided by the empirical expression (28), it is proposed that for an insulator of height h and leakage
length L:

q=1-e
where E,

ED
C

0.541

;
cl

+
0.50
0.5

rp; 1 0 0 C l o h y l
rp = 3000 ohdcm
rp = 6000 ohmkm

0.6

0.7
0.8
P , Atm

,
0.9

,
1 .o

Fig. 5 Critical distance per unit leakage length as


function of atmospheric pressure, for different
pollution severities.
It is shown that, practically independent of the pollution severity or ambient pressure, the critical distance:
x,

0.65 L

(27)

LONGLEAKAGE
PATHINSULATORS
For long rod insulators with closely spaced sheds or
cap-and-pin insulators with deep and close ribs, there is
a possibility of arc bridging by sparkover in air across
some highly stressed gaps, instead of following the
leakage path, as indicated in Ref. [21].
Cheng and Nour [22] have shown experimentally that
for an insulator with rib width w and depth d the efficiency of utilization of the leakage path can be
expressed as:

q=l-e

4wld

where c' depends on severity.

-cES hlEpL

(29)

is the streamer gradient in air


is the pollution flashover gradient
is a constant

i.e. the variable that determines bridging is not only


geometry but is the ratio of the streamer sparkover
voltage of the insulator to its pollution flashover voltage. (29) shows that at high values of EsWEpL, the utilization factor will approach unity, which is physically
sound. Such high value can be due to, for example,,
high pollution severity resulting in low Ep values.
Introducing the dependence of Es and Ep on pressure,
(29) takes the form:
q=l-e

-cE,1 hpo.331Epl L

(30)

where Esl and Epl refer top = 1 atm.


For any insulator design at a given pollution severity.,
the leakage path utilization efficient ql at p = 1 atm
follows from (30):

q1=1-e

-C

Esl hl EP1L

(31)

It follows that the efficiency at any pressure p , q(p).,


is related to q1 by:
(32)

q1 can be measured at p = 1 atm, at any requiredl


severity, by comparing the insulator flashover voltage:
with that of a smooth insulator of the same leakage pathL
and form factor. (32) remains valid even if h in (29) is
replaced by another parameter such as the insulator
d i m eter .
Introducing q(p)in (25):

816

U J P ) = rl(P). Pm .

/ rll

Table 1

(33)

which for a given insulator shape can be expressed as:

U&> = pme . UJ1)

(34)

where m, is an effective pressure exponent.


Fig. 6 shows the variation of m, with 771 in the range
0.6-1.0. Here m, increases from 0.47 at
= 1.0
(smooth cylinder) to 0.68 at q1 = 0.6. The latter value
of q1 is indicative of a poor insulator design with too
deep ribs or too close sheds.
In experimental results on antifog insulators the pressure exponent varied from 0.55 [6] to 0.8 [8] in good
agreement with the above model results.

0.5

0.6

0.7
0.8
ETA1 ,pu

Altitude Correction Factor for Polluted


Insulators with Different 771

0.9

1.o

Fig. 6 Variation of the effective pressure exponent me


with 771 characterizing the insulator profile.

A L T I T U ~ ECORRECT1
From the above analysis, it is clear that the altitude
correction factor Kd will be a function of not only
altitude but also the insulator design represented above
by the parameter 171.
Table 1 shows calculations of Kd from proposed
values of m,.
Although antifog insulators (low ql) are more
severely derated at high altitude as shown in the Table,
such insulators can still be used advantageously when
warranted by pollution severity.

1.oo
0.95
0.85
0.70
0.60

CONCLUSIONS
1. A new, physically-based method is introduced to
account for the effect of altitude on AC flashover
voltages of polluted insulators.
2. The parameters of a previously introduced model
for flashover of polluted insulators at sea level
have been updated, based on revised values for
thermal properties of air at high temperature and
the resulting reignition voltage calculations are in
excellent agreement with experiment.
3. The critical AC withstand voltages of polluted
simple-shaped insulators vary approximately with
the square root of ambient pressure.
4. The critical current is somewhat more sensitive to
ambient pressure than the critical voltage, the
pressure exponent being 0.6 for insulators of
simple shapes.
5. The critical arc length amounts to 65% of the leakage path, practically independent of pollution
severity or ambient pressure.
6. A new formula for the efficiency of leakage path
utilization has been derived, which includes the
effects of insulator geometry, pollution severity
and ambient pressure, and which is in good
agreement with experimentally obtained flashover
voltages at different altitudes.
7 . New altitude derating factors of polluted insulator
performance are presented and account, for the
first time, for insulator geometry.

REFERENCES
[ l ] J. Fryxell and' A. Schei, "Influence of High
Altitude on the Flashover Voltage of Insulators,
Elteknik, Vol. 9, No. 1, 1966, pp. 1-3.

817

[2] R. Wilkins, "Characteristics of Low Current Arcs


Relevant to Pollution Flashover", Second Int.
Conf. on Gas Discharges, IEE, 1972, pp. 206-208.
[3] G.A. Lebedev, E.I. Ostapenko, "Effect of Air
Pressure on Dielectric Strength of Wet
Contaminated Insulators", Elektrotekhnik, No. 1,
1972, pp. 56-58.
[4] V.I. Brobroskii and 1.0.Ordokov, "A Study of the
Electrical Strength of External Insulation under
Mountain Conditions", Soviet Power Engineering,
NO.7, July 1978, pp. 428-430.
[5] T. Kawamura, M. Ishii, M.Akbar and K. Nagai,
"Pressure Dependence of DC Breakdown of
Contaminated Insulators", IEEE Trans. on
Electrical Insulation, Vol. EI-17, No. 1, 1982,
pp. 39-45.
[6] M. Ishii, K. Shimada, T. Kawamura and
T. Matsumoto, "Flashover of Contaminated
Surface under Low Atmospheric Pressure", 4th
ISH, Athens, 1983, Paper No. 46.02.
[7] Z. Tiebin, Z. Renyu and X. Jiaqi, "The Influence
of Pressure on AC Flashover Characteristics of
Contaminated Insulators", IEEEKSEE Joint
Conference on High Voltage Transmission
Systems in China, Beijing, Oct. 17-22, 1987,
pp. 291-294.
[8] V.M. Rudakova and N.N. Tikhodeev, "Influence
of Low Air Pressure on Flashover Voltages of
Polluted Insulators: Test Data, Generalization
Attempts and Some Recommendations", IEEE
Trans. on Power Delivery, Vol. 4, No. 1, January
1989, pp. 607-613.
[9] H.P. Mercure, "Insulator Pollution Performance at
High Altitude: Major Trends", IEEE Trans. on
Power Delivery, Vol. 4, 1989, pp. 1461-1468.
[lo] Anibal de la O., Jorge Glez de la Vega,
"Performance of AC Insulators under Low
Pressure Fog Chamber Tests", 7th ISH, Dresden,
1991, Paper No. 44.19.
[111 H. Chaofeng, G. Zhicheng and Z. Renyu,
"Influence of Air Pressure on AC Flashover
Voltage of Polluted Post Insulators", 8th ISH,
Yokohama, Japan, 1993, Paper No. 46.01.
[ 121 F.A.M. Rizk, "Analysis of Dielectric Recovery
with Reference to Dry-Zone Arcs on Polluted
Insulators", IEEE Conference Paper No. 7 lCPl34PWR, presented at the Winter Power Meeting,
New York, N.Y., 1971.

[ 131 F.A.M. Rizk, "A Criterion for AC Flashover of

Polluted Insulators", IEEE Conference Paper N~D.


71CP135-PWR, presented at the Winter Power
Meeting, New York, N.Y., 1971.
[ 141 F.A.M. Rizk, "Mathematical Models for Pollution
Flashover", Electra, No. 78, 1981, pp. 71-103.
[ 151 H. Maecker, "Uber die Charakteristiken
zylindrischer Bogen", Zeitschrift fur Physik, 15'7,
1959, pp. 1-29.
[16] J. Yos, "Revised Transport Properties for High
Temperature Air and its Components", AVCO
Report, Lowell, Mass., 1967.
[17] V. Isachenko, V. Osipova, A. Sukomel, "Heiit
Transfer", Book, Moscow, 1974, p. 562.
[18] P. Claverie, Y. Porcheron, "How to Choose
Insulators for Polluted Areas", IEEE Trans., Vol.
PAS-92, No. 3, May/June 1973, pp. 1121-1131.
[ 191 J.P. Novak and G. Ellena, "Arc Field Measurement
with a Simple Experimental Arrangement", .I.
Phys. D: Appl. Phys. 20 (1987), pp. 462-467.
[20] C.G. Suits, H. Poritsky, "Application of Heat
Transfer Data to Arc Characteristics", Phys. Rev.,
Vol. 55, 1939, pp. 1184-1191.
[21] D.A. Hoch, D.A. Swift, "Flashover Performance of
Polluted Insulation: An Assessment of the
Influence of Air Density", AFRICON 9Z!,
Swaziland, 22-24 September 1992.
[22]T.C.Cheng and H.I.M. Nour, "A Study on the
Profile of HVDC Insulators", IEEE Trans. on
Electrical Insulation, Vol. 24, No. 1, February
1989, pp. 113-117.

BIOGRAPHY
Farouk A.M. Rizk (Fellow 82) is Fellow Research
Scientist at IREQ, Chairman of E C Technical
Committee 28: Insulation Coordination and Convener
CIG& WG 33.04: Insulator Contamination.

A.Q. Rezazada holds a B.Sc. Eng. degree from


Kabul University in Afganistan and an M.Sc. degree
from McGill University, Montrkal, Canada. Mr.
Rezazada is presently a reliability engineer at General
Motors Canada.

818

Discussion
Gu Leguan (Chongqing University, PRC): I
should like to congratulate the authors on a very
valuable paper which should of interest to utilities
in high altitude regions. This paper for the first
time makes a systematic study on modeling of

M.Moreno and M.Ramirez (LAPEM-CFE, Irapuato, Mexico) :


Mathematical Models for pollution insulators has been a subject of
much interest. The authors should be congratulated for their timely
paper which introduces a new physical approach to account for the
effect of reduced aidensity on the flashover voltage of polluted
insulators. This subject is important for Mkxico and especially for
CFE with transmission sistems at high altitude.
In Mexico we have done some experimental work at LAPEM
(Irapuato 1710 ma.s.1.)and at Topilejo (H.V.Experimenta1Station
at 3000 m.a.s.1.)and part of this data was presented in ref. [l].

altitude effects on AC flashover of polluted high


voltage insulators. I would pay my respect to the
effort of the authors and add the following
comments.

Is the critical distance x,

0.65L (27) independent

of the arc bridging and the types of specimens, that

is, the standard polluted insulator, and the polluted


insulators with deep ribs and small rib spacing?
When discussing "Long leakage path insulators"
the authors state that "-the pollution severity; as it
is

expected

that

with

lower

severity

and

accordingly higher voltage stress arc bridging will


be more likely". Clarifying this point would be
appreciated.

We have made a comparison between our results and your


Mathematical Model results (fig.2) and we have comments and
questions as follow:
The tendency (low U,(p)/L values with reduced p) for Standard
and Fog insulators with rp middle values is quite acceptable,
however, for Standard insulator with rp=9600 Q/cm and with fog
type insulators with rp=500 !U cm the slope or tendency is less
pronounced. Could you comment on this?.
The method used for the solution of eqs. (23) and (24) and
searching for x, , i, ,U, is not clear enough, could you put the
description of the procedure into words ? . How is the fbntion U,,
(p) around the point x,?. dU, /dx=O means a maximun or a
minimun point ? and how would it be interpreted ?.
In eq. (30) there is an exponent equal to 0.33 for the pressure,
where does it come from ?.
Would you be so kind as to answer the above questions and make
additional comments ?.
REFERENCE

D. A. Swift (University of Natal, R.S.A.): Dr. Rizk and his


colleague are to be congratulated on extending the
mathematical model of AC flashover across a polluted and
wetted hydrophilic insulator to take air density into account.
One wonders if this new model can be of more general use for
coping with problems in this important research area.

[ 11 D. Serrano, M. Ramirez and M. Moreno " High Altitude A.C.


Standard Tests on Polluted Insulators " CIGRE 33-94 (WG-

04) 28 IWD. Ludvika, Sweden, 1994.Manuscript received February 12, 1996.

N.N. Tikhodeev, E.A. Solomonik, NIIPT, St.


Peterburg,
Russia: T h e paper is of great scientific and
For various reasons, especially cost, there is a growing need to
practical
interest
for selection of line insulation of HV
be able to calculate - more accurately than is currently the
and
EHV
AC
overhead power transmission lines
case - the flashover voltage of practical shapes of high

voltage insulators. To achieve this aim, one should note that it


seems likely that air-gap discharges between the ribs of a shed
and between sheds play a not insignificant role in this overall
process. Therefore, the next generation of flashover models
need to address this problem. As the authors have shown that
air density affects air gap discharges to a greater extent than it
does surface discharges, we have a variable that could be
employed to good effect for getting the weightings that have
to be apportioned to these two components of the flashover
mechanism.

running at high altitudes of 1OOOm to 4000m above


the sea level. A number of experimental studies,
including that by NIIPT authors [8], used comparative
tests of wet polluted insulator units of different types at
varying relative air pressures p to derive a correction

factor K~ J p o r K~ p m , m tending to be higher for


units of more elaborate shapes. These findings had no
theoretical explanation; now this paper bridges the gap
successfully.

Here are some comments and questions:


Therefore, I ask the authors if they have given any thought to
this broader scope that may now be opened up.

1.The authors calculate U, from Eq. (24), whose

819

second term takes into account the resistance Ri of the


polluted unit's wet zone 3 after the dry zone 2, with the
current applied to zone 3 pointwise by the partial arc 1
of length x (see the insert). The authors of the paper are
indubitably aware that the resistance can be found from
the equation Ri(x) = rp(L - x ) only roughly, which
makes the U, calculation very approximate.
The resistance of the
wet zone at p = 1 was
studied in detail in two
NIIPT papers, one of
which dealt with cap.and-pin
insulators
(E.A. Solomonik, Resistance of polluted
insulator units with
and without partial arc
on their surface, NIIPT
i
Proceedings, No. 1 1,
1965, pp. 74-104) and the other, with rod units (J. Yu.
Gutman, Methodology of calculation of flashover
voltages of polluted rod insulators 7th ISH, R. 43.18,
1991). In these papers the magnitude of RzT was
determined more stringently on the basis of calculations
and simulation of respective electrostatic fields (a ring
with one pointwise current application and a cylinder
with two such inputs). In both cases it was found that
with the current applied pointwise the resistance Rz* ( x )
is 1.5 to 2.5 times as high as Ri ( x ) determined from the
linear model. In addition &* ( x ) depends non-linearly
on x , especially at d L c 0 . 3 and x/L>O.8. In our
opinion, use of the simplistic linear model yields lower
flashover voltages which substantially different from
those found experimentally. Calculation of the ratio
U, ( p ) / U,(1)
may partly compensate the errors
resulting from the neglect of non-linearity of Rz* ( x )
and the pointwise current application to the wet zone in
its critical point (XJL= 213). At any rate, use of the
reported linear model instead of the Rt: ( x ) approach
should be additionally validated.
2. The allowance for the absolute humidity in
Eqs. (19) through (21) is not supported by experimental
findings. Why is the reference temperature taken to be
20C? As a flashover develops the ambient temperature
near the insulator surface and in the partial arc zone
rises much higher.
3. Summarized in Fig. 6 and Table 1 are the basic
recommendations of the reported study. On the whole
they are corroborated by USSR test findings. It could
be to a greater advantage of the paper if the authors
related the parameter
to os more rigidly and

correlated it with the geometry of insulators, including


porcelain and composite rod units.
Manuscript received February 22, 1996.

Farouk A.M. Rizk: I would like to thank the


discussers for their compliments, valuable interest in
the paper and for many pertinent questions. Particularly
noteworthy is the fact that all the discussers have:
previously contributed to the subject matter of the:
paper.
To Prof. N.N. Tikhodeev and E.A. Solomonik
The discussers request validation of the use of the
linear model Rp(x) = rp(L-x) and I am pleased to
respond.
Let us approximate the exponent of i, in (23) by 0.5
to express U,,@) as:

U,, ( p )= A x p0'77 I &

(23)'

For simplification let us also take n=OS in the static


arc characteristic so that:

U,, = x No pmo I

&

This simplification, as will be shown later, will have


insignificant effect on the result.
Let us express the pollution layer resistance in the
general form Rp(x), so that the circuit equation
becomes:
V , , ( p ) = x No pmo I&+

Rp(x)i,

(24)'

Equating (23)' and (24)', with m, = 0.2:


A x p0.77 I & = x

No po.2 I&+

Rp(x)i,

(35)

= ( A p 0.77 - N o p0.2)2'3 x2l3 R,-2/3(x)

(36)

From which:

( A p0.77 - N o p o e 2 ) x

Jm
:. ,i

= R , ( x ),i

Substitute for ,i in (23)':

820

U , ( P )=

~ ~ 2 P10.77
3
p0.77

R2./3(x)

- No p~.2)"3

(37)

The critical distance xc is found from solving:

-(x
d
213 R,113 (x))=O

dx

(38)

and has nothing to do with ambient pressure as long as


no significant arc bridging takes place (see response to
Gu Leguan below). Substitute x=xc in (37):

From (39), independent of whether Rp(x)is linear or


nonlinear, it is obvious that the dependence of U,@) on
pressure can be expressed by:

uc (PI 1 UC(1) =

p0.77 (1- No / A y f 3
0.77 - p0.2 No /

(40)

which from regression analysis in the range 0.6 I


p 5

where m = 0.478, very close to the value of 0.5 most


quoted from experiments on simple shaped insulators
and to the value of m = 0.466 obtained above from the
linear model.
Note that the static arc characteristics have
insignificant effect on the ambient pressure dependence
of the insulator critical withstand voltage. For example
if N,P'.~/A is completely neglected in (40), the
exponent m would be 0.513.
As for the second point brought by the discussers, the
assumption made is that the dielectric withstand
voltage of the residual arc gap at any temperature T is
related to that at ambient temperature T, by Ud(T) =
Ud(T,) * T,IT, independently of the prevailing ambient
humidity at T,. This relationship is supported by
experiments in ref. [23]. [24]. U& = Ud(T,) = xEd, is
defined as the dielectric withstand voltage of the gap
concerned at ambient temperature i.e. either without
any arcing or after very long time following arcing to

permit full recovery of dielectric strength. Therefore


the humidity corrections were applied as for any other
gap at ambient temperature, taken as 20C, using IEC
Standard 60- 1. Experimental support and explanation
of this correction approach are given in ref. [25],[26].
However in the vicinity of arc current zero, the
temperature of the column is obviously much higher
than ambient and the effect of temperature on dielectric
strength is accounted for by the thermodynamic
function multiplied by U,, in (9) or byf(i,) in (10).
Finally at the request of the discussers, 1 will
elaborate further on the proposed expression for the

Substituting in (31) and recognizing that Esl is a


constant:

Substituting for 'i,in terms of q,it follows that:


q = l - - e-const.

0
: hl f

" L1-"

(43)

As previously mentioned, for cap and pin insulator


units, the outside diameter D may replace the spacing h
in the formulae for q. It may even be more appropriate
to replace L by the protected leakage path L, and h by
the minimum distance in air between two consecutive
discs. Further comments on 71 are given in the
response to Prof. Swift.

To Prof. Gu Leguan
Let the arc length x, be related to the length x of
bridged part of the polluted insulator by:

where A ( x , p ) is in general a function of x and the


ambient pressure p .
With the resistance of the unbridged part expressed
by the genera' expression R ~ ( x ) it, can be shown,
following the simplifications introduced above, that the
critical voltage corresponding to any x:

82 1

,ycx (P ) = ~ ~ 2 x2/3
/ 3 P 0.77 R,113 ( x > l

(A p0.77 - N o p0.2)1/3

(45)

The critical voltage U,@) corresponds to a critical


distance x, obtained from:
d (il2
(x,
l3
P) *

213

dx

R ,1/3

=o

(46)

This means that in general the critical length x, will


depend on both the distribution (nonuniformity) of the
pollution layer surface resistance as well as on the
bridging factor il and may therefore differ from

xc

Details of the derivation of the recovery and circuit


equations and their solution are given in Ref. [12-141
and need not be repeated here.
The exponent of 0.33 in (30) is justified as follows.
From (21) Eda or E, is proportional to po'80. From
(25) E, is proportional to p0*47.It follows that E,/Ep is
proportional to p0.33.

0.65 L .

On the other hand for insulators with insignificant


arc bridging of the ribs (ilZ1) or insulator designs
where the bridging factor il can be considered
reasonably independent of x (A = q p ) ) the linear
model then predicts: x, 2 0.65 L independently of
pollution severity or ambient pressure as given in
conclusion ( 5 ) of the paper.
The statement concerning voltage stress is clarified
as follows. At lower pollution severity, it is clear that
the insulator withstands higher voltage. This means that
there will be more voltage available for bridging of the
air gaps i.e. the air gaps will be more stressed at lower
pollution severity, making bridging more likely.

To M. Moreno and M. Ramirez


As mentioned in the paper, the model results of
Figures 2 to 5 apply to insulators without bridging of
adjacent ribs or consecutive sheds. The discussers
mention that their experimental results confirm the
findings of Fig. 2 within the range of r, indicated (1000
- 6000 Wcm). The statement concerning the results of
r,= 9600Wcm and 500SZ/cm is not clear and it is
difficult to comment without seeing the actual results.
My guess is that the discussers compared a point
corresponding to rp = 500 Wcm obtained on an antifog
insulator with r, = 9600 CYcm obtained from a standard
disc and found the tendency less pronounced than in
mean that
@less is correct then this
Fig. 3. If
for the very low Pollution severity (rp=9600wcm7
0,2 2 @ ) , at an altitude of 1710m, Some arc
bridging was already taking place on the standard
insulator. This would account for the less pronounced
variation.

To Prof. D.A. Swift


Yes, I gave some thought to the broader problem of
the effect of insulator shape on its pollution
performance. I agree that the model described in the
paper may serve as a starting point for more detailed
future investigations of that subject. I will include here
only some brief comments.
Previous investigations on the efficiency of
utilisation of the leakage path of cap and pin insulators
[27] resulted in the empirical formula (using symbols
of the present paper):
771

=1/[1+0.5(L/D-l)]

(47)

for U D I 1.55.
For long rod insulators ref. [28] gives the relationship
(using the symbols of this paper)
vi = 2.86 / [1+ L / h]

(48)

Comparison of these relationships with (42), (43)


shows that our formulae go farther than (47) and (48) in
that they indicate that
is a function of not only
insulator geometry but decreases at low pollution
severity, as can be concluded from experiments of
Fig. 5 in Ref. [29]. Furthermore, they show that for
constant r,, 771 is a function WL for long rod insulators
or their replacement quantities for cap and pin units.
On the other hand, for constant q,as is frequently the
case in experimental comparisons, the form factor f
also plays a role in determining 771, for the same h and
LFinally in (31) the assumption that c is independent
of h/L may be valid only for a limited range of that
parameter. If wL however varies within a wide range,
e.g. from WL = 1/3 (antifog rod) to WL = 1 (smooth
cylinder), the nature of the parameter c should be
further
investigated
both
theoretically
and
experimentally. This comment however has no effect
on (32), which remains valid even if c becomes a
function of WL.

822

eferences
[23] L.L. Alston, "High Temperature Effects on
Flashover in Air", Proceedings IEE, Vol. 105, Part
A, December 1958, pp. 549-553.
[24]A.H. Sharabaugh, P.K. Watson, D.R. White,
T.H. Lee, A. Greenwood, "An Investigation of the
Breakdown of Nitrogen at High Temperature with
Use of a Shock Tube", AJEE Trans., Vol. 80, Part
111, 1961, pp. 333-344.
[25] K. Feser, A. Pigini, "Influence of Atmospheric
Conditions on the Dielectric Strength of External
Insulation", Electra, No. 112, 1987, pp. 83-95.
[26] N.L. Allen,
J.R. Fonseca,
H.J. Geldenhuys,
J.C. Zheng, "Influence of Air Humidity on the
Dielectric Strength of External Insulation",
Chapter 8, CIGRE Monograph "Guidelines for the

Evaluation of the Dielectric Strength of External


Insulation".
[27] G.N. Alexandrov, J.M. Gutman, V.L. Ivanov,
V.E. Kiesewetter, A.S. Maikopar, S .D. Merkhalev,
A.A. Philippov, V.S. Rashkes, N.N. Tikhodeev,
"Dielectric Strength of Line Insulation", CIGRE
1966, Paper No. 417.
[28] S.D. Merkhalev, E.A. Solomonik, "Selection and
Operation of Insulators in Regions with Polluted
Atmospheres", Book, Energoatomizdat, Leningrad,
1983, p. 65.
[29] F.A.M. Rizk,
A.El-Sarky,
A.A. Assaad,
M.M. Awad, "Comparative Tests on Contaminated
Insulators with Reference to Desert Conditions",
CIGRE 1972, Paper No. 33.03.
Manuscript received May 29, 1996.

You might also like