21, rue dArtois, F-75008 PARIS http : //www.cigre.

org

A2-105

CIGRE 2008

ANALYSIS ON DIELECTRIC STRENGTH OF SF6 GAS FOR GAS INSULATED TRANSFORMER

I.S. HWANG, H.K. KANG, Y.H. KIM, B.Y. SEOK, K.S. PARK, J.H. WOO Hyundai Heavy Industries Co., LTD. (KOREA)

SUMMARY
Recently, the demand of gas insulated transformer (GIT) using SF6 gas as insulating and cooling medium is being increased continuously in the underground substation of urban area because it has attractive advantages such as non-flammability, non-explosion, and so on. A GIT using SF6 gas is known as relatively reliable compared with an oil-immersed transformer. Hyundai Heavy Industries Co., LTD. (HHI) has developed and commercialized a low pressure type 154kV GIT because it has many economical and reliable advantages. In this paper, dielectric characteristics of various electrode systems in SF6 gas are tested and analyzed to improve the dielectric performance of a low pressure type 154 kV GIT using SF6 gas. The electrode systems which simulate the internal structure of a GIT using SF6 gas are made of stainless steel and dielectric tests according to the gap length between two electrodes and the pressure of SF6 gas are performed. It is found that the dielectric characteristics of electrode system with non-uniform electric field are related to gap length and those of electrode system with quasi-uniform electric field are deeply dependent not only on the gap length but also on the pressure of SF6 gas. Also, barrier effects on lightning impulse dielectric strength test in gaseous condition are experimentally verified. In this investigation, dielectric tests are performed by installing barriers made of pressboard (PB) between electrode system and breakdown voltage according to the pressure of SF6 gas and the number of PBs is measured. Consequently, it is found that barrier effects are effective on lightning impulse dielectric strength test.

KEYWORDS Barrier effect, dielectric characteristics, gas insulated transformer, insulation, SF6 gas, transformer

[email protected]

1. INTRODUCTON
The establishment of underground substation is a general tendency due to environmental and site problems of urban area. Therefore, the installation of power machinery in underground is increasing drastically [1]. Generally, it is well known that a GIT is relatively non-flammable and non-explosive in comparison with an oil-immersed transformer. Also, it is easier to enlarge the capacity of a GIT than that of a cast resin transformer. For these reasons, research and development on GIT is being performed actively in the world [1]. Actually, HHI developed high pressure type 154 kV GIT in 2000 and has been commercializing low pressure type 154 kV GITs known as more effective and economical since 2005. In this paper, lightning impulse dielectric strength of various electrode systems in SF6 gas and barrier effects in SF6 gas are investigated experimentally to improve the insulation performance of a GIT developed by HHI. The breakdown voltage with 2 % probability is calculated by using statistic software, Minitab and electric field strength is simulated by using a finite element method (FEM) tool, Maxwell in order to analyze the dielectric characteristics of electrodes. The experimental results are expected to be applied to reinforce the dielectric strength of HHIs low pressure type 154 kV GITs.

2. EXPERIMENTAL SETUP
The cross sectional view of low pressure type SF6 GIT developed by HHI is shown in Fig. 1 briefly. As shown in Fig. 1, dielectric characteristics at turn-turn and section-section of windings are very important parameters to determine the design parameters of a GIT [2], [3]. Generally, lightning impulse dielectric strength test is known as more severe condition than the AC dielectric breakdown test in a GIT. Therefore, only lightning impulse dielectric strength tests for a GIT are conducted in this study according to the protocol of ANSI C5.21-1990. The test methods of ANSI C5.21-1990 are described in Table 1. The specifications of power supplies used in dielectric test are the maximum lightning impulse of 240 kV and 15 kJ (1.2/50 ). The experimental setup of dielectric tests is shown in Fig. 2. Dielectric tests were performed 10 times in each condition and representative data are calculated by using statistic program, Minitab. The calculated representative data denote the electrical breakdown voltage with 2 % probability.

Side yoke

Tap

HV winding

LV winding

TV winding

Core leg

Shield ring

Shield ring

Fig. 1. Cross sectional view of an SF6 GIT.

Table 1. Conditions of dielectric test on an SF6 GIT Test protocol Test voltage wave front / wave tail Chopping time of CLI ANSI C5.21-1990 - Full Wave Lightning Impulse (FLI): 650 kV - Chopped Wave Lightning Impulse (CLI): 750 kV 1.2 / 50 s 4~6 s

A simulated electrode system is installed in a transparent chamber made of acrylic resin to conduct dielectric tests. A phenomenon of breakdown is observed through the transparent chamber. The upper side of electrode systems is connected with the high voltage terminal of a power supply and the lower side is connected to ground. And then, the test chamber is vacuumized with a rotary pump down to about 10-2 torr in order to minimize the influence of impurities before the injection of SF6 gas. After vacuumizing, SF6 gas is filled from an SF6 cylinder. In this experiment, the breakdown voltage of electrode systems is measured with various pressure ranges from 1 bar to 4 bar and gap lengths from 2 mm to 10 mm.

Fig. 2. Schematic drawing of lightning impulse dielectric strength tests.

3. DIELECTIC TESTS OF ELECTRODE SYSTEMS IN SF6 GAS

3.1. Basic electrode systems Three kinds of electrode systems are made and dielectric tests of the electrode systems are performed. The electrode systems used in this test are classified into three kinds with respect to the distribution of electric field. The pictures of each electrode system are shown in Fig. 3 and the detail specifications of each electrode system are described in Table 2.

(a) #1

(b) #2

(c) #3

Fig. 3. Picture of simulated electrode systems.

3.2. Dielectric test results of electrode systems As a result of analysis by using the FEM, it is found that the distribution of electric field strength takes a quite different complexion according to the shape of electrode systems. The distribution of electric field strength of #1 electrode system is very non-uniform because the upper side of electrode system has sharp edge with 50 m in diameter. On the other hand, the electric field of #2 electrode system is quasi-uniform and that of #3 electrode system is almost uniform because of its even surface and fluent curvature. The experimental results of lightning impulse dielectric strength test on #1 electrode system are shown in Fig. 4 and Fig. 5. Electrode #1 Table 2. Specifications of electrode systems Specifications Upper R: 25m D: 80mm, R: 5mm, H: 30mm D: 25mm D: 80mm, R: 5mm, H: 30mm D: 65mm, R: 7.5mm, H: 15mm

Lower Upper #2 Lower #3 Upper & Lower * R: radius, D: diameter, H: height

Breakdown Voltage [p.u]

5 4 3 2 1 0
0

1 bar 2 bar 3 bar 4 bar 1 bar-Air

4 6 8 Gap length [mm]

10

Fig. 4. Dependence of lightning impulse breakdown voltage on gap length for various pressures. (in case of #1 electrode) (base : @ 2 mm, 1bar-air, #1 electrode system)

Breakdown Voltage [p.u]

5 4 3 2 1 0 0.5 1.5

2mm 3mm 5mm 7mm 10mm

2.5 3.5 4.5 Pressure [bar] Fig. 5. Dependence of lightning impulse breakdown voltage on pressure for various gap lengths. (in case of #1 electrode)

(base : @ 2 mm, 1bar-air, #1 electrode system) Fig. 4 shows the dependence of lightning impulse breakdown voltage on gap length for various pressures and Fig. 5 shows that on the pressure of SF6 gas for various gap lengths. As shown in Fig. 4, the electrical breakdown voltage of #1 electrode system is remarkably improved in SF6 gas compared with that in air. Also, it is found that the electrical breakdown voltage of #1 electrode system in SF6 gas increases logarithmically with according to gap length but it rarely increase according to pressure. In case of #2 electrode system, the electric strength will be distributed more uniformly than in case of #1 electrode system and breakdown voltage acquired by dielectric tests increases logarithmically according to gap length and pressure as shown in Fig. 6.

14 12 10 8 6 4 2 0 0

Breakdown Voltage [p.u]

1 bar 2 bar 3 bar 4 bar 1 bar-Air

4 6 8 Gap length [mm]

10

Fig. 6. Dependence of lightning impulse breakdown voltage on gap length for various pressures. (in case of #2 electrode) (base : @ 2 mm, 1bar-air, #1 electrode system)

Breakdown Voltage [p.u]

25 20 15 10 5 0 0

1 bar 2 bar 3 bar 4 bar 1 bar-Air

4 6 8 Gap length [mm]

10

Fig. 7. Dependence of lightning impulse breakdown voltage on gap length for various pressures. (in case of #3 electrode) (base : @ 2 mm, 1bar-air, #1 electrode system)

The distribution of electric field strength of #3 electrode system will be most uniform among three kinds of electrode systems. Fig. 7 shows the breakdown voltage of #3 electrode system according to various gap lengths and pressures. From these results, it is found that as the distribution of electric field is more uniform, the dependency of breakdown voltage according to pressure and gap length becomes higher. The dielectric characteristics of SF6 gas with various electrode systems are summarized in Fig. 8 and Fig. 9. In case of non-uniform field like #1 electrode system, dielectric characteristics are not dependent on pressure but slightly dependent on gap length. However, in case of quasi-uniform field like #2 and #3 electrode systems, dielectric characteristics are deeply dependent on pressure as well as gap length.

Breakdown Voltage [p.u]

20 16 12 8 4 0 Air

#1-5 mm #2-5 mm #3-5 mm

1.5 2 2.5 3 Pressure [bar]

3.5

Fig. 8. Comparison of lightning impulse breakdown voltage on pressure with various electrode systems. (base : @ 2 mm, 1bar-air, #1 electrode system)

20 Breakdown Voltage [p.u] 16 12 8 4 0 2

#1-3 bar #2-3 bar #3-3 bar

5 7 Gap length [m m ]

10

Fig. 9. Comparison of lightning impulse breakdown voltage on gap length with various electrode systems. (base : @ 2 mm, 1bar-air, #1 electrode system)

4. DIELECTRIC TESTS ACCORDING TO BARRIER EFFECTS IN SF6 GAS

It is known that electrical breakdown mechanism in gaseous state is related to the maximum electric field strength. The maximum electric field strength will increase by solid barriers between #3 electrodes because of high relative permittivity (r=4.6) of solid barriers compared with that of air (r=1). Therefore, solid barriers are not being installed in a GIT for improving dielectric characteristics because the magnitude of maximum electric field increases by installation of solid barriers. However, the effects of solid barriers installed in a cast epoxy resin transformer produced are already proved to enhance the dielectric characteristics. In this study, dielectric tests according to barrier effects in SF6 gas are performed in this paper to analyze barrier effects. 4.1. Structure of electrode systems with solid barriers To analyze electric effects of solid barriers in SF6 gas, solid barriers (125(w) 125(H) 2(t)) whose relative permittivity is 4.6 are used. The simulated electrode system used in barrier effect test is #3 electrode and gap length of it is fixed as 12 mm. The schematic drawing of dielectric test for verifying the barrier effects is shown in Fig. 10. The upper side of electrode system is connected to the high voltage terminal of power supply and the lower side is connected to ground. The number of barriers is varied from 0 to 4 and the pressure of SF6 gas varied between 1 and 2 bar.

1 6 r=7.5 15 12 50 (a) No barrier 6 (b) 2 barriers 2

2 2 1 (c) 3 barriers (d) 4 barriers 1 2

Fig. 10. Schematic view of barrier installation in #3 electrode system.

4.2. Results of electric field analysis The electric field strength of #3 electrode system with barriers shown in Fig. 11 was calculated by using an FEM tool, Maxwell. The electric fields with no barrier and 3 barriers were calculated and compared with each other as shown in Fig. 11. In this calculation, 1 kV is applied to upper side electrode and lower side electrode is grounded. In the right side of Fig. 11, 0 mm denotes the surface of upper side electrode and 12 mm indicates the surface of lower side electrode. As shown in Fig. 11, maximum electric field drastically increases at gaseous region because the permittivity of SF6 gas is relatively small compared with that of barrier. The relation between the number of barrier and the magnitude of electric field is calculated and shown in Fig. 12. From the calculated results, it is known that the maximum electric field is proportional to the number of barriers but the average value is constant irrespective of the number of barriers. It is known that the mechanism of breakdown voltage in gas which is dependent on maximum electric field can be deteriorated by solid barriers [4]. Therefore, solid barriers are not installed in gas insulation system for enhancing dielectric performance in general.

Electric field [kV/mm]

0.25 0.2 0.15 0.1 0.05 0 0 2 4 6 8 Gap length [mm] 10 12 no barrier 3 barriers

Fig. 11. Analysis results of electric field according to barrier effect.

0.25 Electric Field [kV/mm] 0.2 0.15 0.1 0.05 0

E-MAX E-avg.

No 2 3 4 barrier barriers barriers barriers Fig. 12. Variation of electric field with respect to the number of barriers.
4.3. Results of dielectric tests The dielectric test on barrier effects is repeated 5 times in each condition and test results are represented as per unit of breakdown voltage with 2 % probability. In Table 3, the test results of electrical breakdown are classified into creepage and penetration according to the pressure of SF6 gas and the number of barriers. As shown in Table 3, the electrical breakdown proceeds with creepage under the low pressure of 1.5 bar in every condition except in 2 barrier condition. The breakdown proceeds with penetration from the pressure of 1.5 bar in 2 barrier condition. It is verified through experiment that penetration is related to the number of barriers and the pressure of SF6 gas. Generally, it is known that the ionization of gas is in inverse proportion to the magnitude of pressure. Therefore, breakdown has a tendency to proceed with creepage in a barrier system at low pressure because penetration needs more energy than creepage. Actually, breakdown always proceeds with creepage at the pressure of 1 bar irrespective of the number of barriers because of high ionization rate of SF6 gas.

(a)

(b)

(c)

(d)

(e)

(f)

Fig. 13. Procedures of breakdown by creepage under the atmospheric pressure. The behaviour of electrical breakdown is monitored with a high speed camera made by REDLAKE Corporation (MotionXtra HG-100K). Fig. 13 shows the procedures of breakdown by creepage under the atmospheric pressure. For 2 barrier condition, breakdown proceeds with full penetration from the pressure of 1.5 bar because breakdown through solid barrier is easier than along the surface of barrier. Therefore, the magnitude of breakdown voltage at the 2 barriers and 2 bar condition is lower than that at the no barrier and 2 bar condition. In case of 3 barrier condition, full penetration occurs from the pressure of 2 bar and in case of 4 barrier condition, breakdown never occurs without creepage under the pressure of 2 bar. Table 3. Complexion of breakdown with respect to pressure and number of barriers 1 bar 2 barrier 3 barrier 4 barrier creepage creepage creepage 1.5 bar penetration creepage creepage 2 bar penetration penetration creepage

Test results with respect to the number of barriers and pressure are shown in Fig. 14 and Fig. 15. As shown in Fig. 14, it is verified that the adoption of solid barrier is partially effective to enhance the lightning impulse dielectric strength of SF6 gas through experimental tests. However, breakdown voltage of 2 barrier condition is smaller than that of no barrier condition over the pressure of 1.5 bar. For 3 and 4 barrier systems, breakdown voltage is always higher than that of no barrier system independent of pressure and the number of barriers. Through dielectric tests, it is verified that the barrier effects on lightning impulse dielectric strength are definitely effective not only in air but also in SF6 gas. This is a significantly meaningful result to design the internal structure of an HHIs GIT. Actually, the adoption of 2 barrier system can be an adverse condition in an HHIs GIT because the pressure of an HHIs low pressure type GIT is about 2 bar. It will be helpful to use more than 3 barriers to enhance the dielectric characteristics of a HHIs low pressure type GIT.

8 Breakdown Voltage [p.u] 6 4 2 0

No barrier 2 barriers 3 barriers 4 barriers

1 bar-Air

1 bar-SF6

1.5 bar-SF6

2 bar-SF6

Fig. 14. Breakdown voltage of #3 electrode wih barriers according to the pressure of SF6 gas. (base : @ No barrier, 1bar-air)

8 Breakdown Voltage [p.u] 6 4 2 0

1bar-Air 1bar 1.5bar 2bar

No barriers

2 barriers

3 barriers

4 barriers

Fig. 15. Breakdown voltage of #3 electrode with barriers according to the number of barriers. (base : @ No barrier, 1bar-air)

Consequently, several kinds of design parameters such as pressure, safety factor, and economical efficiency as well as the number of barriers are must be prudently taken into consideration to improve the dielectric characteristics of a GIT by using solid barriers.

5. CONCLUSIONS
In this paper, the lightning impulse dielectric characteristics of various electrode systems according to the pressure of SF6 gas and gap length between electrode systems are investigated. Also, dielectric characteristics on barrier effects are experimentally verified and test results are summarized as follows; 1. Dielectric characteristics of electrode systems with even surface increases logarithmically according to gap length as well as pressure. 2. The lightning impulse dielectric strength of SF6 gas is enhanced by using solid barriers in general. However, dielectric characteristics can be degraded at unfavorable conditions. 3. The adoption of solid barriers should be decided according to various design factors such as pressure, safety factor, economical efficiency, and the number of barriers are must be considered to improve the dielectric characteristics of a GIT. Consequently, experimental results will be applied to enhance the dielectric characteristics of an HHIs low pressure type GIT.

BIBLIOGRAPHY
[1] [2] [3] Tsukao.S, Oue.Y and Haseqawa.Y, Application of gas insulated transformer to underground substations in Japan (IEEE, Vol. 1, 2002, pages 517-521) T.Inoue, T.Ternishi, M.Ikeda and S.Yanabu,Dielectric characteristics of turn-to-turn insulation of Gas-Insulated Transformer for various impulse voltage application(Trans.IEE Japan, Vol. 120-B, No.4, 2000, pages 569-575) K.Muramatsu, T.Goda, M.Ikeda, K.Hirai, Y.Ishioka and K.Toda, Study on Application of thermally upgraded and low permittivity dielectrics to Gas Insulated Transformer (Trans.IEE Japan, Vol. 120-B, No.4, 2000, pages 511-517)

[4] Y.Togawa, M.Ikeda, K.Toda and K.Esumi, Progress of Gas-insulated Transformer (IEEE Proceedings of International Conference of Energy Management and Power Delivery, 1995, pages 540-547)

10