XVth International Symposium on High Voltage Engineering

University of Ljubljana, Elektroinštitut Milan Vidmar, Ljubljana, Slovenia, August 27-31, 2007 T9-342.pdf

Influence of surface state and thin dielectric coating on the insulation properties of a coaxial
electrode arrangement in SF6
Christoph Lederle1*, Josef Kindersberger1 and Edelhard Kynast2
1
TU-München, Institute for High Voltage Engineering and Power Transmission, Arcisstr. 21, 80333 München, Germany
2
Siemens AG, PTD H 321H, 13623 Berlin, Germany
*Email: [email protected]

Abstract: The improvement of the dielectric strength of focus was very different. An improvement of 100% but
SF6 insulated systems by using a thin dielectric coating also a reduction of the dielectric strength has been
to change the surface state of electrodes was studied. found. Uncertain results were found about the influence
The paper deals with the breakdown probability of a of the first breakdown and the residual breakdown
coaxial aluminium electrode arrangement with different strength of an insulation system.
surface roughness in pressurized SF6. Thin epoxy resins A proper comparison can be done by a statistical
with different thickness were used as coating material evaluation of the breakdown voltage which requires
on the inner conductor. The outer conductor of the determining the distribution function including
electrode arrangement was polished and uncoated. The confidence region. The damage of coating caused by
surface roughness of the inner electrode was varied. To root points and the influence on consecutive
determine the cumulative frequency function of the breakdowns requires special consideration. Therefore,
breakdown voltage the progressive stress test of charge measurements have to be done to find an
stepwise voltage increase was applied for positive and influence of the surface charge on the coating damage.
negative lightning impulse. Charge measurements have It is also important to investigate electrodes with
been carried out to investigate the influence of the first technical surfaces like they are used in real GIS-
breakdown and of the root points after a breakdown. systems.
With the applied charge measurement the influence of
the rising voltage test on the breakdown probability 2 EXPERIMENT
could be evaluated. The influence of repairing the root
point of the surface by a second coating procedure
(emplatrification) has been investigated. 2.1. Test setup for breakdown tests
The coaxial electrode arrangement used in the
1 INTRODUCTION experiments is shown in Fig 1.

For dimensioning of gas insulated systems surface

irregularities have to be taken into account. Even on uncoated
small irregularities like technical surfaces roughness an s
increase of the electric field increases the probability of
140
80

a breakdown [1]. A thin dielectric coating could

improve the dielectric strength.
Several papers deal with the enhancement of the coating
breakdown strength of gas insulted systems by coating
300
electrodes with an insulating material. The investigated
parameters like insulating gas, gas pressure, electrode
Fig. 1: Drawing of coaxial electrode configuration
configuration and coating material vary in a wide range.
The investigated electrode configurations were rod-
plane, sphere-plane, plane-plane and coaxial
arrangements with different dimensions [7,8,9,14]. The The outer electrode of the coaxial arrangement was
electrode configurations were tested with AC, DC and dimensioned by means of field calculation to avoid
lightning impulse voltage and different testing methods breakdowns on the boarder area. The outer electrode is
[2,8,14]. Air, SF6, N2 and vacuum were used as made of drawn aluminium EN AW 6060. The inner
insulating media. Epoxy-resin, PTFE, silicone, shrink surface of the outer electrode is always polished
hose, Al2O3 and plastics with various thickness of few (Rz = 5 µm) and uncoated. The inner electrode is made
micrometers up to some millimeters have been used as of drawn aluminium EN AW 6060 or cast aluminium
coating material [3,4,5,10,14]. Partially statistically EN AC ALSi10Mg (Fig. 2).
analyses of measurement results have been carried out.
Because of the manifold parameters the experimental

1
 lr

coating

ds

Rt
Z(x)
x

Fig. 3: Parameters of surface texture and coating thickness ds

Fig. 2: Assembly of coaxial electrode arrangement

2.3. Test setup for charge measurements

2.2. Roughness and coating variation A hemisphere-plane-electrode arrangement has been

used to investigate the influence of charge
The mean roughness depth Rz was varied from 5 µm accumulations The grounded hemisphere electrode can
to 50 µm on the inner conductor. To specify the surface be moved by a pneumatic cylinder to stress or
state the mean roughness depth Rz according to [EN measurement position (Fig 4 and Fig5).
ISO 4287] is used. Rz is determined by dividing the
measured trace into 5 equidistant sections lr. For each
section the distance between the lowest depth and the
highest peak Rt is measured (Fig 3). Rz is the mean
value of this five Rt values.
The roughness of the inner conductor is adjusted by
either polishing or vibratory grinded (Table 1). With a
probe
vibratory grinding process, also known as trowalizing, a
surface will be smoothed by constant grinding with a
mixture of emulsion and stones.
cross table
Table 1: Surface treatments of inner electrodes
polishing 4 µm ≤ Rz ≤ 6 µm Fig. 4: Setup for surface potential measurement: coated
trowalized 20 µm ≤ Rz ≤ 35 µm electrode in stress position

The surface texture parameters were measured with

a Surftest 301 from Mitutoyo, which uses the tactile
profile method. For some samples the results were
verified by using a laser profilometer with an auto focus
sensor. The results from both measurement methods
were the same.
Different epoxy-resins (EP2, EP3) with similar
dielectric permittivity and electrical conductivity were
used as coating material. The coating thickness was
varied from 15-300 µm and the surface roughness of the
coating is always between Rz = 3 µm and 6 µm. The
coating thickness was measured with the Elcometer 456
working on the eddy current measurement principle. Fig. 5: Setup for surface potential measurement: coated
After calibrating the coating thickness measurement electrode in measurement position
system to a specific surface roughness the coating
thickness ds can approximately defined as distance The surface potential on the coated hemisphere
between the center line of roughness and the coating electrode is measured by an Electrostatic Voltmeter
edge (Fig. 3). (Trek 341 BH20). The measuring probe is fixed on a
The electrodes were cleaned with Isopropanol and cross table (ISEL C142-4). An A/D converter card
placed in a test vessel filled with SF6 gas pressurized at (PMD-1608FS) is used to automate the pneumatic
0,5 MPa absolute pressure. The gap distance of the switching, the control of the cross table and the
electrode arrangement is 30 mm resulting from the recording of the surface potential data. The cross table
dimensions of the conductors. with measuring probe and the pneumatic control unit is

2
assembled in test vessel and can be pressurized up to 3 INFLUENCE OF SURFACE STATE
0,8 MPa (Fig. 6). Electrical and pneumatically bushings
were used to control the assembly. After each individual Trowalized and polished electrode surfaces have
test of the progressive stress test of stepwise voltage been investigated to determine the influence of the
increase the surface potential of the coated hemisphere surface state on the dielectric strength of the coaxial
is measured till a breakdown occur. Due to protection of electrode arrangement. The surface roughness Rz of
the measurement system, the charge accumulation after uncoated trowalized electrodes is between
a breakdown with lightning impulse voltage was only 20 µm ≤ Rz ≤ 35 µm and of polished electrodes
measured in air. 4 µm ≤ Rz ≤ µm. The electrodes were coated with an
epoxy resign (EP3). No significant first breakdown can
be found for positive and negative lightning impulse
voltage (Fig.7)

800

Breakdown volt. ûd [kV]

700

600

500

Fig. 6: Arrangement for charge measurement: hemisphere 400

electrode in measurement position.
0 5 10 15 20 25 30

For a thin coating the surface charge density can be individual test number
calculated from the surface potential ∆φ, the dielectric Fig. 7: Breakdown voltage over individual test number,
permittivity εr, ε0 and coating thickness d according to negative LI, polished electrodes, EP3 coated d = 65 µm,
(1). (s =30 mm, p = 0,5 MPa)
εε
σ( x , y ) = ∆ϕ( x , y ) 0 r (1)
d The coating thickness can be adjusted with the
number of coating layers. The thickness of one layer is
2.4. Test procedure to evaluate measurements
15 to 20 µm. After the experiment with one coating
The coaxial electrode arrangement is stressed by thickness the next coating layer is placed on the same
standardized lightning impulse voltage (1,2/50µs) which electrodes. New root points occurred from a breakdown
is supplied by a lightning impulse generator with 50 kJ can not be found in previously coated root points. To
at 3 MV charging voltage. The polarity of the lightning repair a punctured surface coating the whole new
impulse voltage (LI) refers to the polarity of the inner coating layer or a single root point can be repainted.
conductor. To determine the cumulative frequency Repairing a dielectric coating by repainting is called
function of the breakdown voltage the progressive stress emplastrification. Fig. 8 shows root points of a coated
test of stepwise voltage increase was applied. The electrode. The darker points are root points of previous
voltages were positive and negative lightning impulse breakdowns which have been recorded.
voltages (LI). From an estimated initial voltage u0 the
voltage was increased with a step amplitude ∆u = 18 kV
until a breakdown occurred. The interval between two
individual tests was ∆tp = 5 min. The number of
individual tests was defined to n = 30 breakdowns. In
contract to a sphere-plane-electrode arrangement no
significant first breakdown can be found [13, 14]
(Fig. 7). The statistical independence of the
measurement values has been controlled by
mathematical methods. The cumulative frequency
function was approximated with the normal distribution
Fig. 8: Root points, EP2 coated (d=315µm)
function and the 95%-confidence region has been
calculated.
3.1. Negative polarity
The breakdown probability of the uncoated
trowalized electrode stressed with negative lightning

3
 impulse is significantly lower than of all coated 3.2. Positive Polarity
electrodes (Fig. 9). With increasing coating thickness
The breakdown probability of the uncoated
(EP3) of the inner electrode, the dielectric strength of
trowalized electrode stressed with positive lightning
the coaxial arrangement increases significantly. After a
impulse is significantly lower than of all coated
test series with 30 breakdowns 17-24 root points can be
electrodes (Fig. 11). The breakdown probability
found on the coated electrode surface.
increased with increasing coating thickness in the range
100 of thickness from 28 to 111 µm except with 111 µm.
breakdown probability [%]

The numbers of root points after a test series with 30

80 breakdowns were between 16 and 21.
uncoated
60 100

breakdown probability [%]

113 µm
40 80
89 µm uncoated
20 60
26 µm
84 µm
0 40
400 500 600 700 800 900 28 µm 111 µm
breakdown voltage ud [kV] 20

Fig. 9: Normal distribution of breakdown voltage with 95% 0

confidence region, negative LI, trowalized electrodes, EP3 400 500 600 700 800 900
coated, (s =30 mm, p = 0,5 MPa)
breakdown voltage ud [kV]
The 10% quantil of the breakdown voltage from the Fig. 11: Normal distribution of breakdown voltage with 95%
uncoated polished electrode is significantly higher confidence region, positive LI, trowalized electrodes, EP3
(~25%) than of the trowalized electrode (Fig. 10). This coated, (s =30 mm, p = 0,5 MPa)
can be explained by the different surface roughness
(polished Rz ≈ 5µm, trowalized Rz ≈ 25µm). With The 10% quantil of the breakdown voltage from the
increasing surface roughness the breakdown voltage uncoated polished electrode is significantly higher
decreases. With a thin layer of dielectric coating the (~20%) than of the trowalized electrode (Fig. 12). This
breakdown voltage increases significantly above the can be explained by the different surface roughness
breakdown voltage of the polished electrode (polished Rz ≈ 5µm, trowalized Rz ≈ 25µm). By coating
independent of the surface roughness of the electrodes. the breakdown voltage increases significantly above the
Equal coating thickness on polished and trowalized breakdown voltage of the polished electrode
electrodes lead to equal 10% breakdown voltages. Only independent of the surface roughness of the electrodes.
with a coating thickness of 90 µm on the trowalized Equal coating thickness on polished an trowalized
electrode a significant difference to the polished electrodes lead to equal 10% breakdown voltages.
electrode with equal coating thickness can be found. Coating of trowalized electrodes lead to higher
Coating of trowalized electrodes lead to higher or at breakdown voltages than uncoated polished electrodes.
least equal breakdown voltage than uncoated polished 800
10% Breakdown volt. ûd [kV]

electrodes.
10% Breakdown volt. ûd [kV]

800 700

700 600
polished
600 polished
500
trowalized trowalized
500
400
400 0 20 40 60 80 100 120
0 20 40 60 80 100 120 coating thickness [µm]
coating thickness [µm] Fig. 12: 10%-quantiles of breakdown voltage with 95%
confidence region, positive LI, polished and trowalized
Fig. 10: 10%-quantiles of breakdown voltage with 95% electrodes, EP3 coated, (s =30 mm, p = 0,5 MPa)
confidence region, negative LI, polished and trowalized
electrodes, EP3 coated, (s =30 mm, p = 0,5 MPa)

4
 4 DIFFERENT COATING MATERIAL 800

10% Breakdown volt. ûd [kV]

Two different coatings made of epoxy-resins with
equal dielectric properties (εr, ρD) and slightly different 700
mechanical properties have been investigated. The root
points of the more brittle EP2-coating are larger than of 600
the EP3-coating
4.1. Negative Polarity 500
The breakdown voltage of the uncoated polished EP3 coating 18-216µm EP2 coating 185-316µm
electrode is significanlyt lower than of all EP2 and EP3 400
coated electrode (Fig. 13). With increasing coating 0 40 80 120 160 200 240 280 320
thickness the dielectric strength of the electrode coating thickness [µm]
arrangement increases. The 10% quantiles of the Fig. 14: 10%-quantiles of breakdown voltage with 95%
breakdown voltage for EP2 and EP3 in the investigated confidence region, positive LI, polished electrodes, EP3 and
thickness range of 175 – 250 µm are equal. The reason EP2 coated, (s =30 mm, p = 0,5 MPa)
could be found in the equal dielectric properties of the
two coating materials. After 30 individual tests 13 to 24
root points can be found on the coated surface.
5 APPLICATION
The knowledge about the improvement of electrode
10% Breakdown volt. ûd [kV]

800
configurations by proper selection of electrode surface
roughness combined with coating can be applied to any
700 SF6 insulated equipment to increase the dielectric
strength of gas gaps and therefore to reach an optimized
600 design. Fig. 15 shows as an example of application a
550 kV gas insulated bushing during high voltage
500 testing where an increase of more than 10 % of the
EP3 coating 18-227µm EP2 coating 175-325µm withstand voltage was reached by coating of the
400 electrodes.
0 40 80 120 160 200 240 280 320

coating thickness [µm]

Fig. 13: 10%-quantiles of breakdown voltage with 95%
confidence region, negative LI, polished electrodes, EP3 and
EP2 coated, (s =30 mm, p = 0,5 MPa)

4.2. Positive Polarity

The 10% quantil of the positive Lightning Impulse
breakdown voltage of the uncoated polished electrode is
lower than of all EP3 coated electrodes (Fig. 14). With
increasing coating thickness also the dielectric strength
A correlation between coating thickness and dielectric
strength of the coaxial electrode arrangement could not Fig. 15: 550 kV gas insulated bushing
be found. The breakdown voltage for EP2 and EP3 is o
in the investigated thickness range of 185 – 316 µm is
only at a thickness of 250µm equal. The high scatter 6 CHARGE MEASUREMENTS
spread of the positive breakdown values cause wide
95% - confidence regions. After 30 individual tests 16 To investigate the influence of the progressive stress
to 21 root points can be found on the coated surface. test of stepwise voltage increase on the surface charge
accumulation of the dielectric coating, charge
measurement after each individual test has been carried
out.
No influence of the stepwise voltage increase on the
surface charge accumulation with a 150 µm coated
polished hemisphere electrode could be found at any

5
 gap distance of the hemisphere and the plane electrode 7.2. Charge measurement
in air (p = 0,1 MPa) and SF6 (p = 0,1 and 0,2 MPa).
- The progressive stress test of stepwise voltage
With all investigated gas conditions the first breakdown
increase has no influence on the surface charge
voltage was significant higher then the following
accumulation on a coated hemisphere electrode (EP3,
breakdown values.
150 µm) in air and SF6.
After a breakdown or a flashover, charges of the
- After a breakdown or flashover, charges of the
same polarity than the applied voltage can be found.
same polarity than the applied voltage can be measured
The measurement of the surface charge density can be
on the coating. An influence of surface charge density
started 30 seconds after the breakdown. The decrease of
for consecutive breakdowns is possible.
the surface charge density over time is very small
(Fig. 17). After 5 minutes a rest charge density of
7 nC/cm2 can be found at the root point. According to 8 REFERENCES
(1) a surface potential of 0,3 kV can be measured. The
[1] A. Pederson , The Effect of Surface Roughness on breakdown in
time between two individual test of the progressive SF6 , no.5 in IEEE Transaction , PAS-94, p. 1749-1754, Sep
stress test of stepwise voltage increase is 1975
∆tp = 5 minutes. After this time an influence of surface [2] A.E. Vlastòs, The influence of thin electrode coating on the AC
charge density for consecutive breakdowns is possible. breakdown in SF6 , London, UK, 4th International Conference on
Gas Discharge and their Application, 1976
10
charge density ‫׀‬σ‫[ ׀‬nC/cm2]

[3] A.E. Vlastòs, Influence of electrode coating on the impulse

8 breakdown of SF6, Milan , Italy, 3rd Int. Symp. on Hight Voltage
Engineering, paper 32.08, 1979
6 [4] A.E. Vlastòs, Influence of the electrode surface state on the
breakdown of pressurized SF6, UK, 6th International Conference
4 on Gas Discharge and their Application, 1980
[5] A. Rein, J.Kulsetâs, Impulse breakdown of SF6/N2 insulation,
2 influence of electrode covering; polarity effects, Gaseous
Dielectrics,Ed. Program Press, pp 315-321, 1982
0
[6] W. Hauschild and W. Mosch, Statistik für Elektrotechniker,
0 100 200 300 400 500 600 Berlin,Germany, VEB Verlag Technik, 1984
time after charging [s]
[7] K.D. Srivastava, R.G. van Heeswijk, Dielectric coating-effect on
Fig. 17: Charge density decrease over time, polished breakdown and particle movement in GITL systems, USA, IEEE
electrodes, 150µm EP3 coated, (s =20 mm, p = 0,1 MPa air) Transactions on Power Apparatus and Systems, 1985
[8] M. Honda, H.Okubo, H.Aoyagi, A. Inui, Impulse breakdown
characteristics of coated electrodes in SF6 gas, Kawasaki, Japan,
IEEE Transaction on Power Delivery, Bol. PWRD-2 No.3, 1987
7 CONCLUSION
[9] M.M. Morcos, K.D. Srivastava, Dielectric coatings and particle
A coated and uncoated coaxial electrode arrangement in movement in GIS/GITL systems, USA, New York; Gaseous.
Dielectrics 8, 1998
SF6 (p = 0,5 MPa) has been stressed with positive and
negative lightning impulse voltage. [10] D. König, R. Heinemeyer, V. Scherb, Breakdown voltages of
dielectric-coated sphere-plane-electrode configurations in high
7.1. Breakdown tests vaccum, USA, New York, IEEE Internatinal Symposium on
Electrical Insulation, 1988
- The dielectric strength of all coated electrodes is
[11] J.H.M. Blennow, Active High Voltage Insulation, Scheden,
significantly higher than of the uncoated electrodes. No Göteborg, Chalmers Univerity of Technology, 2000
influence of the first breakdown could be found. After a
[12] Ch. Lederle, J. Kindersberger, The influence of surface
test series with 30 breakdowns 16-24 root points are on roughness and dielectric coating on AC and DC Breakdown
the coated electrode surface. Voltage in SF6, China, Beijing, ISH, 2005
- The investigated epoxy coatings with equal coating
[13] Ch. Lederle, J. Kindersberger, E. Kynast, Untersuchungen zum
thickness on polished (Rz ≈ 5 µm) and trowalized Einfluss der Oberflächenbeschaffenheit und Leiterbeschichtung
(Rz ≈ 25 µm) surfaces lead to an equal improvement of auf das Isoliervermögen einer Halbkugel-Platte-Anordnung bei
the dielectric strength. Basic requirement it the complete Blitzstoßspannung in SF6, Germany, Hanau, ETG-Fachtagung,
cover of all surface roughness with the coating material. 2005
- Coating of trowalized electrodes lead to a higher [14] Ch. Lederle, J. Kindersberger, The Influence of surface
breakdown voltage than uncoated polished electrodes. roughness on the Impulse Beakdown Voltage of SF6, USA,
- The emplastrification process is well suited to Boulder, CEIDP, pp 522-525, 2004
repair root points and surface defects.