Influence of the surface conductivity of a single glass barrier on the breakdown voltage in an air insulated rod plane arrangement 1 Schueller M.1, Blaszczyk A.2, Krivda A.2, Smajic J.1 University of Applied Sciences Rapperswil, CAEM/HVL - Group, Switzerland 2 ABB Switzerland Ltd, Corporate Research, Switzerland Abstract- The effect of barrier systems on electrical breakdown is well known and of wide use in the field of high voltage engineering. As shown in literature, barriers can increase the breakdown voltage of an electrode configuration. The effect of the surface conductivity of the barrier system on the breakdown voltage has not yet been studied in full detail. The aim of this contribution is to show this effect. For that purpose, we investigated a rod plane arrangement with 80mm distance between high voltage electrode (rod) and a grounded plate electrode. A single barrier was placed in the arrangement and its surface resistivity was decreased systematically. The decrease of the breakdown voltage of the arrangement with decreasing surface resistivity has been shown for positive and negative lightning impulse and AC voltage as well. I. INTRODUCTION The barrier effect has been found by E. Marx and H. Roser when they studied air gaps discharges in the 1930’s [1, 2]. In these early works the barrier effect in gases was explained with redistribution of the electric field in the gap because of the space-charge field that was formed by the accumulated charges on the barrier surface originating from impact ionization near the needle tip [1-3]. Nowadays this model from Marx and Roser is still used by the majority of authors to explain the barrier effect in gaseous dielectrics. Some authors even use it as explanation for liquid and solid dielectrics [4, 5]. At present day there are many other experimental hypotheses trying to explain the barrier effect in dielectrics such as inhomogeneous polarization or electrophysical characteristics of the main and barrier materials [6]. According to [6] the Marx-Roser model can only explain the barrier effect in relatively short gas gaps. But it doesn’t hold for large gas gaps, liquids or even solid arrangements as the drift velocity and free path length of the charge carriers are too small to form a significant surface charge layer on the boundary within the for the Marx-Roser model required time [6]. Although this great number of conducted experimental work, the exact mechanism of the barrier effect is still not known [6]. With the measurements presented here the yet not in detailed studied effect of the surface resistance of the barrier on the breakdown voltage was investigated. II. THEORY The Marx-Roser model explains the barrier effect with the creation of space charge due to ionization near the electrode tip. In case of positive tip fast electrons will move into the needle tip and recombine there. The remaining positive space charges drift in direction of the field and accumulate at the barrier. There a quasi-uniform electric field is formed between the opposite plane electrode and the barrier resulting in the increase of the breakdown voltage [1, 2, 6]. In case of negative needle tip the faster electrons move away from the tip also trough the barrier [1] and a positive space charge remains at the tip. These positive charges decrease the field strength at the electrode tip. So the authors of [1, 2] concluded that this leads to a field equalization resulting in a higher breakdown voltage of the arrangement [1, 2, 6]. III. METHODS A. Experimental Set up To measure the effect of decreasing surface resistance of a barrier a vertical single barrier arrangement between a high voltage rod and a grounded plate electrode was chosen, see Fig 1. The grounded plate electrode was made of copper and had the dimensions (height x width) 1100 x 1000 mm. As barrier, a 6 mm thick float glass plate with dimensions 700 x 525 mm was used. The high voltage rod electrode was made of alumina and had a diameter of 7 mm. It was 260 mm long and the tip was rounded with a radius of 3.5 mm. As gap distance between the tip of the high voltage electrode and the ground electrode 80 mm was chosen. The barrier was put in the arrangement at two different positions. Position 1 was 20 mm in front of the high voltage electrode and position 2 was 60 mm in front of the high voltage electrode. The experiments were conducted with lightning impulse voltage and alternating voltage. An eight stage 800 kV, 40 kJ Marx generator was used as source of the positive and negative impulse voltages. A 150 kV, 40 kVA high voltage test transformer was used in the experiments with AC voltage. Fig. 1 Dimensions of the experimental set up. B. Barrier Material and barrier position As barrier material glass was used because collected surface charges are not bound so strong at the surface as it is the case with PVC, acrylic glass or some other polymers. Thus at every discharge the surface charges are more or less removed. This was important as we wanted to have as much as possible the same electric field conditions for every measurement. The used glass for the barrier was a 6 mm float glass plate with polished edges. It had the dimensions (height x width) 700 x 525 mm. It is known from literature that the position of the barrier in the gap has a strong influence on the breakdown voltage of the arrangement [7-6]. Placing the barrier in front of the high voltage electrode at a distance of 20-30 % of the overall length of the gap the positive effect of the barrier is maximal. If the barrier is placed far away from the high voltage electrode the effect of the barrier can even be negative, meaning the breakdown voltage can be lower than without barrier [7-6]. To take this effect into account the barrier was placed at two positions in front of the high voltage electrode. Position 1 was 20 mm in front of the electrode, and Position 2 was 60 mm in front of the high voltage electrode. Considering that the gap was 80 mm these two positions were at 25 % (20 mm) and 75 % (60 mm) respectively of the gap distance. 8 9 10 C. Test Procedure The surface resistance of the barrier was systematically decreased with every experiment. This was realized with a conductive graphite spray. At every series of measurement more graphite spray was applied on the glass barrier and thus the surface resistance was decreased. The surface resistance of the barrier was measured with a Keithley source meter 2410 before every series of measurement. It can provide up to UDC = 1.1 kV and measure a minimal current of 1 pA. Thus it was possible to measure the surface resistance up to a value of = 10 Ω. After measuring the surface resistance of the barrier the experiments were conducted. Like mentioned before the barrier was set once 20 mm and once 60 mm in front of the high voltage electrode. In case of positive and negative impulse voltage the 50% breakdown voltage was obtained using an up&down method according to [11]. This method requires the normal probability distribution of the breakdown voltage, which is fulfilled for tests in gaseous insulation. Rise and decay time of the impulse voltages was used according to the standardized lightning impulse shape with 1.2/50 µs. In case of alternating voltage the breakdown voltage was obtained by applying a voltage ramp with a defined rise time of 9kV/sec. This was done until breakdown occurred. For better comparability with the impulse voltages the peak value of the AC voltage was recorded. A minimum number of 30 measurements was performed for every point and indicated together with the corresponding standard deviation in Fig. 2. All voltages were corrected for pressure, humidity and temperature according to IEC 60060. IV. RESULTS In Fig. 2 the results of all the performed measurements are shown. As dashed-dotted lines the breakdown voltage of the pure air gap without barrier is indicated for positive (blue), negative (violet) impulse voltage and for AC (red). Further positive and negative impulse tests with barrier are indicated by a solid line. AC measurements with barrier are shown by a dashed line. In general a decrease of the breakdown voltage could be shown when the conductivity of the barrier was increased. This effect was measured irrespective of the voltage polarity (impulse) and voltage form (AC and impulse). For impulse tests the breakdown voltage of the system is higher for negative voltage (-142 kV) than for positive voltage (83 kV). The breakdown voltage of the gap for AC is with 92 kV slightly higher than for positive impulse voltage. The polarity effect is reversed when the breakdown voltages of the impulse experiments with 20 mm spacing of the barrier are compared. In that case the breakdown voltage of the arrangement is higher for positive (182 kV) than for negative impulse voltage (-154 kV). The breakdown voltage of the arrangement is higher when the barrier is placed close (20 mm) to the high voltage electrode than if it is placed further away (60 mm). This holds for all measured voltage forms and polarities from a surface resistance of 3.66 1011 Ω down to 3.2 kΩ. At the latter value this behavior changes for all voltage forms and polarities as the breakdown voltage is higher for the barrier further away. For all voltage forms and both polarities the breakdown voltage of the gap was increased when a nonconductive barrier was placed in front of the high voltage electrode. Compared to the values without barrier for positive impulse voltage the increase was the largest as the breakdown voltage was increased by 119 % (20 mm position) and 36 % for the 60 mm position. In case of negative impulse there was more or less no increase of Fig. 2 Breakdown voltage depending on the resistance of the barrier. The dotted lines represent the breakdown voltage of the 80mm gap without barrier. The solid lines are for lightning impulse (LIP) and the dashed lines represent measurements with AC. For each measurement point the standard deviation range is indicated. the breakdown voltage measured. With barrier the breakdown voltage was only 8 % higher than without barrier. For AC the increase of the breakdown voltage was 50 % for the 20 mm position but even slightly lower than the value of the pure gap for the 60 mm position. With decreasing surface resistance the breakdown voltage even increase a little till a measured surface resistance of 11.9 106 Ω. From this value on the breakdown voltage decreases and at a value of 104 Ω the breakdown voltage was lower than without barrier for all measured polarities and voltage forms. V. DISCUSSION If the derived breakdown voltage of the gap only is compared with the value obtained with an insulating barrier (no carbon spray Rsurface = 366 GΩ) in the experiment, the breakdown voltage of the system could be increased in case of positive impulse and AC voltage. This was expected and was already shown in literature for many times [1-6]. In case of negative impulse voltage the barrier had no major effect on the breakdown voltage when it is compared to the value of the gap only stressed with negative impulse. One explanation for that could be the transparency of glass for UV light. So the UV light from the streamer head coming from the high voltage electrode just passes through the glass plate directly leading to further ionization on the other side of the glass plate. Thus the high voltage rod electrode is not shielded by accumulated charges on the barrier like the Marx-Roser model suggests and both formed streamers (the one from the high voltage electrode and the other formed on the other side of the barrier by ionization due to the UV) meet at the corner of the barrier leading to a value of the breakdown voltage close to the value of the gap only. This hypothesis has to be confirmed by conducting positive and negative impulse experiments with a barrier not transparent for UV. Another explanation could be that at a negative tip the electrons “…will penetrate through the barrier into the gap between the barrier and plane as the barrier is transparent for electrons…” [1]. So a positive space charge is left behind at the negative tip due to the slower mobility of the positive ions. This space charge reduces the field strength at the negative tip. This is known as polarity effect. According to [6] this effect is minor than the positive charge accumulation at the barrier in case of positive impulse. So this could explain the difference between positive and negative impulse experiments as well. In general a decrease of the breakdown voltage with decreasing surface resistance could be shown. First compared to the breakdown voltage of the insulating barrier (no carbon spray Rsurface = 366 GΩ) the breakdown voltage increases even a little with decreasing surface resistance. This is very interesting as it leads to the conclusion that the barrier performance may be improved by applying a conductive layer. An improvement up to 10 % could be achieved in our measurements. This can be significant for insulation design. This effect seems to be present for negative, positive impulse and AC. The most interesting region seems to be between a surface resistance of 107 to 104 Ω as a transition of the behavior occurs there and the breakdown voltage of the system is smaller than for gap only. Following the Marx-Roser model [1, 2, 6] this could be because the charge at the barrier is distributed faster at the surface due to the smaller resistance and the barrier behaves as a floating electrode. Based on the stability field for the positive streamer we can estimate that the potential of the floating barrier is during the positive impulse just 10 kV lower ≈ 0.5kV/mm ⋅ 20mm = 10 than the applied voltage ( [12]). Due to low barrier resistance this potential will be transferred to the barrier edge, which creates many inception points around the large barrier circumference. This "enlargement" of inception can lead to the lower withstand voltage than in the no-barrier case where the inception is limited to the small area at the tip of the rod. At a low surface resistance of the barrier the breakdown voltage of the negative impulse voltage is higher than the breakdown voltage for positive impulse. For example for the 20 mm distance at a resistance of 104 Ω the breakdown voltage of positive impulse is 81 kV and for the negative impulse it’s -111 kV. Like explained above with low surface resistance of the barrier the edge of the barrier behaves as electrode and starting point for the streamer. In this case it seems that the polarity effect works like it is known as the breakdown voltage of the negative impulse becomes higher as it is measured for the positive impulse voltage. Following the Marx-Roser model [1, 2, 6] this would also mean that more positive charges are accumulated at the barrier than negative charge. This can be seen as at a high surface resistance of the barrier the breakdown voltage for the positive impulse voltages is much higher than for the negative impulse voltage. Comparing positive impulse voltage and AC the AC breakdown voltage of the gap without barrier is higher than for the impulse voltage. This was expected as with AC in every half circle all the charges are removed from the gap. Thus the probability of a start electron is reduced resulting in a higher breakdown voltage as it was measured for impulse voltage. It was believed that this effect is so explicit because the 9 kV/sec risetime of the AC voltage ramp was rather fast. Repeating the AC measurements of the 80 mm gap with a slower rise time of 1 kV/sec did result in the same breakdown voltage as it was measured for the 9 kV/sec ramp measurements. But if the values for the measurements with barrier are compared this behavior changes as the breakdown voltage of positive impulse is much higher than for AC. Following the Marx-Roser model [1, 2, 6] in case of positive impulse positive charges are accumulated at the barrier resulting in a quasiuniform electric field between the barrier and the plate electrode [1, 6]. Thus the breakdown voltage is increased for a positive needle [1, 6]. Like discussed before one theory [6] is that the positive charging of the barrier guards a positive tip more than a positive space charge guards a negative tip. As with AC the polarity changes at every half circle and also all charges are removed every half circle it could be that the breakdown at AC always happens when the wave is negative. Thus a lower breakdown voltage of AC with barrier was measured as it was measured for positive impulse. But this theory has to be confirmed in further experiments. VI. CONCLUSIONS A decrease of the breakdown voltage of a rod plane arrangement with a single barrier could be shown when the resistance of the barrier was decreased. Interestingly an increase of the breakdown voltage by 10 % could be shown when the barrier was slightly conductive compared to a totally insulating barrier. This might be a significant finding for insulation design. Still many questions of the exact physics in these arrangements remain to be investigated and answered with further experiments and computational models. ACKNOWLEDGEMENTS The authors want to thank Jonas Ekeberg and Sergey Pancheshnyi at ABB, CRC Switzerland, for valuable discussions during this work. REFERENCES [1] Marx, E. ‘‘Der Durchschlag der Luft im unhomogenen elektrischenFelde bei verschiedenen Spannungsarten’’, Electrotech. Z., Vol. H33, pp. 1161-1165, 1930. [2] Roser H, ‘‘Schirme zur Erhöhung der Durchschlagspannung in Luft’’, Electrotech. Z., Vol. H17, pp. 411-412, 1932. [3] Voloschenko N.F., ‘‘About the Barrier Effect Mechanism’’, Russian J. Electricity, No. 3, pp. 21-26, 1946. [4] Yoshino K., ‘‘Electrical Conduction and Dielectric Breakdown in Liquid Dielectrics’’, IEEE Trans. Electr. Insul., Vol. 21, pp. 847-853. [5] Leontev Yu.N., Torbin N.M., ‘‘Influence of the Barrier Position on the Breakdown Voltage of Solids’’, Trans. USSR Universities, Electrical Engineering, No. 12, pp. 34_37, 1961. [6] Lebedev S.M., Gefle O.S., Pokholkov Y.P. “The Barrier Effect in Dielectrics The Role of Interfaces in the Breakdown of Inhomogeneous Dielectrics”, IEEE Trans on Dielect. and Electr. Insul. Vol. 12, No. 3; October 2011. [7] Lebedev S.M., Gefle O.S., Agoris D.P., Pokholkov Y.P. “Barrier effect in dielectrics”, Proc. of ICPADM, Xian, China, pp. 895-898, 2000. [8] Koenig D., Lautenschlaeger H.G. “The barrier effect in air and SF6 insulated nonuniform electrode configurations”, VII Int.Conf. on Gas Discharges and their Application, London, pp. 292-295, 1982. [9] Mauseth F., Jorstad J.S., Pedersen A. “Streamer inception and propagation for air insulated rod-plane gaps with barriers”, IEEE Conf. Electr. Insul. Dielectr. Phenomena (CEIDP), pp. 729-732, 2012. [10] Ming L, Leijon M and Bengtsson T. “Factors influencing barrier effects in air gaps”, 9th ISH Symp. Graz, Austria, paper #2168, 1995. [11] Dixon W.J., Mood A.M. “A method for obtaining and analyzing sensitive data”, J. Am. Stat. Assoc., No. 43, pp. 109-126, 1948 [12] Pedersen A., Christen T., Blaszczyk A., Böhme H. „Streamer inception and propagation models for designing air insulated power devices”, IEEE Conf. Electr. Insul. Dielectr. Phenomena (CEIDP), pp. 604-607, 2009
© Copyright 2026 Paperzz