Flashover Behavior of Semiconducting Glazed Insulators Under PositiveLightning Impulse Stress at Different Climatic Conditions

O. Elsässer, K. FeserInstitute of Power Transmission and

High Voltage TechnologyUniversity of Stuttgart

GERMANY

F. MessererInstitute of High Voltage Engineering and

Electric Power TransmissionTechnical University Munich

GERMANY

Abstract

The flashover voltages of a porcelain and asemiconducting glazed insulator are compared. Thetest samples are stressed with positive lightningimpulse (LI) voltage (1.2/50) at four differenttemperatures (10°C, 20°C, 30°C, 40°C) in anenvironmental chamber. In each temperature step therelative humidity is varied between 10% and 90%.The test results are evaluated over the absolutehumidity. A temperature correction is applied to theflashover voltage, to compare the values of the twoinsulator types over the absolute humidity.The potential distribution along the insulators isinvestigated by numerical field calculation. Thedistribution for different frequencies of the drivingvoltage and for impulse voltages is shown. Theflashover behavior of the two different insulator typescan be explained by their potential distribution.The flashover voltages obtained with lightningimpulse stress are compared to results measuredearlier with switching impulse (SI) voltage(250/2500). The differences in the flashover behaviorcan be described with the potential distribution atlightning respectively switching impulse stress.The advantages and limits of the semiconductingglazed insulator, compared to a porcelain insulator,are described for LI and SI stress.

Introduction

Insulators are importent components to guarantee areliable power transmission of electrical energy.Various insulating materials have prooved to besuitable for different climatic conditions and pollutionareas. Composite insulators can reduce service cost insurroundings with heavy pollution or salt fogcontamination. Because of their hydrophobicbehavior, pollution layers can’t build up as quickly ason porcelain insulators. The risc of dry band arcing isalso reduced by the hydrophobicity transfer to thepollution layer. Another possibility to avoid dry band

arcing is the use of semiconducting glazes onporcelain insulators. The flashover performance of different insulatingmaterials under switching impulse stress for severaltemperature and humidity levels has been investigatedearlier [1,2]. For clean conditions no differences in theflashover voltage for porcelain and compositeinsulators could be observed. The semiconductingglazed insulator showed higher values. The tests presented in this paper investigate theflashover voltage of a porcelain and a semiconductingglazed insulator under lightning impulse stress atdifferent climatic conditions. The flashover behavioris described by the potential distribution along theinsulators.

Experimental Setup

The experimental setup is shown in figure 1. Thesame setup has been used in earlier experiments [1,2]with switching impulse stress.

insulator

bushing

environmentalchamber

loadcapacitance

marx−generator

Reext

Rdext

Umea

dividervoltage GND

HV

Fig. 1: Experimental setup

The voltage is generated with a marx generator andshaped to standard lightning impulse voltage (1.2/50)with external resistors. The test samples are located inan environmental chamber. The voltage is appliedthrough a bushing.

The characteristic data of the voltage (Up, Tp, T2 or Tc)is measured with a voltage divider and analysed witha digital impulse analysing system(Haefely DIAS 730).The climatic conditions in the test chamber aremeasured with a psychrometer.

Test Samples and Test Methods

The test samples are shown in figure 2. A porcelainand a semiconducting glazed porcelain insulator withthe same design are used for the tests. The flashoverdistance is about 280 mm.

280 mm

porcelain insulatorsemiconducting glazed porcelain

insulator

Fig. 2: Test samples

The test voltage is applied in steps. About 15 steps areused in each test series, 20 impulses are applied perstep. The results are evaluated in a probability paper.The 50%−flashover voltage is derived from a linearregression through the measured values.

Test Results

All flashover voltages presented in the followingparagraphs are corrected to a pressure of 101,3 kPa.

Influence of the Absolute Humidity on theFlashover Voltage and Temperature Correction

The influence of the absolute humidity on theflashover voltage is presented in figure 3 and 4. Anincrease of the flashover voltage with rising absolutehumidity can be observed for both insulator types andall temperature steps. Differences can be seen in thegradient of the increase.A flashover voltage drop can be recognized at acertain humidity level for rising temperatures.The sole influence of the absolute humidity on theflashover voltage can be seen after applying atemperature correction to the data.

180

190

200

210

220

230

240

250

260

270

280

0 5 10 15 20 25 30 35 40 45 50absolute humidity

Uf5

0

g/m³

kV

30°C

40°C

20°C

10°C

LI pos. 1.2/50

porcelaininsulator

Fig. 3: Dependence of the 50%−flashover voltage (LI pos.) on the absolute humidity at different temperatures for a porcelain insulator

180

190

200

210

220

230

240

250

260

270

280

0 5 10 15 20 25 30 35 40 45 50absolute humidity

Uf5

0

g/m³

kV

30°C

40°C

20°C10°C

semiconductingglazed insulator

LI pos. 1.2/50

Fig. 4: Dependence of the 50%−flashover voltage (LI pos.) on the absolute humidity at different temperatures for a semiconducting glazed insulator

180

190

200

210

220

230

240

250

260

270

280

0 5 10 15 20 25 30 35 40 45 50

absolute humidity

Uf5

0

g/m³

kV

porcelaininsulator

semiconductingglazed insulator

LI pos. 1.2/50

Fig. 5: Influence of the absolute humidity on the50%−

flashover voltage (LI pos.) of a semiconducting

glazed and a porcelain insulator (temperature corrected values)

The values can be corrected with the following

equation: dmeas

m

stdcorr UU ⋅

+°+°=

0C15,273

C15,273

ϑϑ

(1)

The exponent m = 0.7 is determined from theexperiments.Figure 5 shows a relative flashover voltage increase of0.9%/(g/m³) for the porcelain and 0.5%/(g/m³) for thesemiconducting glazed insulator. The absoluteflashover values for the semiconducting glazedinsulator are higher up to a humidity level of 37 g/m³.

Field Calculation

The numerical field calculations are based on theBoundary−Element−Method (BEM). The boundaryconditions are modified to calculate capacitive−resistive fields with surface and volume resistivity[3,4].

Fig. 6: Model of the test setup for the field calculation

For capacitive−resistive field calculation the systemof equations obtained for both known and floatingpotential conductor boundaries is the same as thosefor capacitive field calculation with the onlydifference that the potentials and the charges arecomplex quantities.Fig. 6 shows the test setup as a 3D−model. Numericalcalculations can be made for 2D and 3D−axissymetricgeometries. The environmental chamber is describedas a cylinder which does not have significantinfluence on the field distribution of the insulator.

Potential Distrubution along a SemiconductingGlazed Insulator

Figure 7 shows the potential distribution along thesemiconducting glazed insulator for differentfrequencies of the driving voltage. The values areobtained by the numerical field calculation. For thesemiconducting glazed insulator a value of2.841⋅106 Ω for the surface resistivity is applied (κ =0.352 µS).

0

10

20

30

40

50

60

70

80

90

100

0 50 100 150 200 250 300distance

po

ten

tial

kV

mm

10 kHz

100 kHz

5 kHz50 Hz

50 kHz30 kHz

20 kHz

1 MHz

f

HV GND

distance0 260 mm

Fig. 7: Potential distribution along a semiconducting glazed insulator for different frequencies of the

driving voltage (derived from field calculation)

For low frequencies the potential shows an ohmicdistribution according to the conductivity of thesemiconducting layer. For rising frequencies thepotential distribution shifts over to a capacitivebehavior, which corresponds to the potentialdistribution of a porcelain insulator.

0

10

20

30

40

50

60

70

80

90

100

1,0E+01 1,0E+02 1,0E+03 1,0E+04 1,0E+05 1,0E+06 1,0E+07

frequency

po

ten

tial

Hz

kV HV GND

1 2 3 4 5 6 7

1

76543

2

Fig. 8: Frequency response for the tip of each insulator

shed for a driving voltage of 100 kV

0

10

20

30

40

50

60

70

80

90

100

0 50 100 150 200 250 300distance

po

ten

tial

kV

mm

ohmic distribution and SI

capacitive distribution

LI HV GND

distance0 260 mm

Fig. 9: Potential distribution along a semiconducting glazed insulator under LI and SI stress

Another possibility to illustrate the frequencydependence of the potential distribution is given in

HV

insulator

figure 8. Each curve gives the frequency response onthe tip of the seven insulator sheds. A simpleequivalent RC−circuit for the semiconducting glazedinsulator can be developped, which results in the samefrequency response. Figure 9 presents the results froma PSPICE simulation with impulse voltages on theRC−circuit. For switching impulse voltage thepotential distribution is still ohmic, whereas thedistribution for lightning impulse voltage is betweenthe ohmic and the capacitive distribution.The almost linear potential distribution for switchingimpulse voltage on the semiconducting glazedinsulator results in higher flashover voltage valuescompared to the porcelain insulator. The flashovervoltage is about 20% higher at standard conditions(20°C, 101,3 kPa, 11 g/m³) according to figure 10.Under lightning impulse stress the difference inflashover voltage is about 8%.

160

180

200

220

240

260

280

0 5 10 15 20 25 30 35 40 45 50absolute humidity

Uf5

0

kV

g/m³

porcelain insulator

semiconducting glazedinsulator

SI: 37 kV

LI: 18 kV

11

LI

SI

Fig. 10: Influence of the absolute humidity on the 50%−flashover voltage of a porcelain and a semiconducting glazed insulator under lightning and switching impulse stress

0

10

20

30

40

50

60

70

80

90

100

0 50 100 150 200 250 300distance

po

ten

tial

mm

kV LI 1.2/50layer conductivity κ

0.352 µS0.0352 µS

3.52 µS1.36 µS0.681 µS layer conductivity of

the test sample

Fig. 11: Potential distribution along a semiconducting

glazed insulator for different layer conductivities (LI stress)

A more linear voltage distribution under lightningimpulse stress could be achieved by raising the layerconductivity of the semiconducting glaze. Figure 11gives the influence of the layer conductivity on the

potential distribution. For a more linear potentialdistribution the layer conductivity would have to beincreased by a factor of 4−10.

Discussion and Conclusion

The relative flashover voltage (LI) increase withabsolute humidity is about 0.5%/(g/m³) for asemiconducting glazed and 0.9%/(g/m³) for aporcelain insulator. For lightning impulse stress thesemiconducting glazed insulator shows 8% higherflashover voltage values at standard conditions. Forswitching impulse stress the difference rises up to20%. This can be explained by the linear potentialdistribution along the semiconducting glazed insulatorat SI. The conductivity of this insulator would have tobe raised 4 to 10 times to get a more linear potentialdistribution and higher flashover values at LI. Thiswould cause 4 to 10 times higher leakage currents,which could lead to thermal stability problems,especially if you consider the negative temperaturecoefficient of the semiconducting glaze.For this reason lightning impulse voltage is thelimiting factor for the design of semiconductingglazed insulators. With an 8% higher flashovervoltage compared to a porcelain insulator, theflashover length could be made 8% shorter, whichwouldn’t result in a significantly more compactdesign.

References

[1] O. Elsässer, K. Feser: Flashover behavior ofdifferent insulating materials under positive switchingimpulse voltage stress at different climatic conditions,ISH 1999, London, conference puplication No. 467,Volume 4, pp. 4.119.S27−4.122.S27

[2] O. Elsässer, K. Feser: Flashover behavior ofsemiconducting glazed insulators under positiveswitching impulse stress at different climaticconditions, CEIDP 1999, Austin, annual report,pp. 711−714

[3] S. Charkravorti, H. Steinbigler: Capacitive−Resistive Field Calculation around a HV Insulatorusing Boundary−Element−Method,ISH 1997, Montreal, Vol. 3, pp. 49−52

[4] F. Messerer, W. Boeck: Field Optimization of anHVDC−GIS−Spacer, CEIDP 1998, Atlanta, annualreport, pp.15−18