HUMIDITY INFLUENCES ON THE BREAKDOWN OF ROD-PLANE GAPS UNDER POSITIVE IMPULSES IN ATMOSPHERIC AIR P. N. Mikropoulos and C. A. Stassinopoulos Department of Electrical and Computer Engineering, Aristotelian University of Thessaloniki, Thessaloniki, Greece Introduction For many years there has been a growing interest in understanding the influence of the atmospheric conditions and particularly of the absolute humidity on the breakdown mechanism of air-insulated high voltage systems. It has been recognised that the previous correction procedure IEC-60 [1], allowing measurements of breakdown voltages to be referred to a standard humidity level had shortcomings [2 6]. For that reason new IEC standards [7] emerged based on a semi-physical model as was suggested by Feser and Pigini [4]. Although the new standard has more applicability it was proposed that the short gap clearances should be regarded as exceptions from the new correction procedure [4, 8] or that a different correction procedure should be more appropriate for short gaps [9, 10]. Besides, it is believed [5, 11& 12] that there is a lack of data regarding the effect of humidity on air breakdown for short gaps. This paper contributes such data referring to the variation of the 50% breakdown voltage and of the humidity correction factor for rod-plane gaps with the spacing, the geometry of the tip of the rod and with the wavefront duration of the positive applied impulse voltages. Experimental arrangement A 4-stage, 560kV, 1kJ Marx generator produced impulse voltages with wavefront varying from 2 to 250ìs and wavetail around 2000ìs, including the "normal" switching waveshape 250/2500ìs [7]. The gap consisted of a cylindrical brass rod with a diameter of 23mm hanging over an earthed aluminium 2x2m plane placed 80cm above the floor of the laboratory. Two gap lengths, 25 & 50cm, and three types of interchangeable rod tips have been used: a conical tip with a 30 o end, a square-cut tip and a hemispherical tip. The atmospheric conditions were not artificially controlled as humidity was found to vary normally achieving values between 2.4 and 21g/m 3. Class 1 (multiple-level) tests [7] were performed in order to obtain breakdown probability curves "p(u)" and thus to determine the values of the 50% breakdown voltages "U 50 " and of the standard deviation. These tests were executed for breakdown probabilities from 0 to 1 with groups of twenty applied impulses per voltage level and for six, at least, values of absolute humidity per gap-impulse configuration. Each voltage level differed by about 2 3% from
the previous one. All the voltages were corrected for air density according to the standards [7]. Information on the discharge was obtained by the oscillographic monitoring of the electric field at the centre of the plane. Results and discussion The p(u) curves were compared to both the Gauss and Weibull cumulative distributions. It appeared that the fit with the Gaussian cumulative was best for probabilities from 0 0.85 whereas the fit with the Weibull cumulative was best for probabilities between 0.85 1. However, in some cases, nongaussian probability distributions were seen to exist. Typical examples of this can be seen in Fig. 1. 1 0.75 0.5 0.25 0 190 210 230 250 270 290 Ucr(kV), 40/1900ìs, 50cm, square-cut tip. Fig. 1: Fitting with the Gauss (solid) and Weibull (dashed) cumulative distribution of the p(u) curve for two values of absolute humidity. 17.9gm -3 5.9gm-3 30 25 20 15 10 5 0 tf (ì s) 25C 25S 25H 2 10 18 40 100 150 250 Fig. 2: Frequency(%) of deviation from the Gauss cumulative distribution of the p(u) curves vs front duration. 25cm gaps. The relative frequency of the non Gaussian probability distributions for the 25cm gaps is shown in Fig. 2 with the tip of the rod and the front duration "t f " as parameters. From this figure it can be seen that the deviation from the Gauss distribution becomes more frequent for intermediate wavefronts and for gaps with square-cut tip of the rod. Besides, it was found that the non-gaussian distributions increase in number with increasing humidity (Fig. 1) and with decreasing gap spacing. Since the breakdown mechanism for short gaps is governed by the pre-discharge phenomena it is the authors' opinion that the deviations of the p(u) curves from the Gauss distribution are related with the amount and the position of the positive space charge injected by the initiation of the first corona in conjunction with the rate of drifting of the latter away from the vicinity of the rod towards to the plane. For short wavefronts the large amount of the positive space charge of the first corona due to the high inception voltages inhibits the development of a leader resulting in breakdown to occur mainly with streamers under higher voltages. For long front durations the injected space charge of the first corona is significant less thus allowing breakdown to take place always through a leader channel and thus under lower voltages.
In the case of intermediate wavefronts a transition exists from a streamer to a leader dominated breakdown mechanism resulting in deviations from the Gauss distribution. This transition is depended also on the electrode profile, the gap length and on humidity. All the U 50 breakdown voltages, obtained through the Gaussian cumulative distributions, were linked with the corresponding values of absolute humidity and a linear correlation was found to exist. Besides breakdown voltages always attained bigger values with increasing humidity. Typical examples of such linear regressions of U 50 over humidity involving the best correlation coefficient and the poorest one can be seen in Fig. 3. In order to estimate the influence of humidity and the dependence of the latter on the Fig. 3: Linear regression of U 50 over humidity. The vertical bars represent the corresponding standard deviations. gap configuration and on the front duration the percentage of correction of the U 50 per g/m 3 of absolute humidity has been computed by finding a per cent correction factor "k s " from the equation: k s s x 100 U n50. In this equation "s" is the slope of the straight lines produced via the linear regression of U 50 over humidity and "U n50 " represents the values of the 50% breakdown voltage for a normal humidity (11g/m 3 ) obtained from these. Figs. 4 and 5 relate the humidity correction factors k s to the front duration for the 25 and 50cm gaps respectively with the tip of the rod as parameter. Fig. 4: k s (%) vs front duration. 25cm gaps. Fig. 5: k s (%) vs front duration. 50cm gaps. Since leader channel has a bigger conductivity ( 1kV/cm) as compared with the streamers ( 5kV/cm) the important parameter at breakdown for the gaps tested becomes the proportion of the gap that is bridged by either type of discharge. Also, it was established that the
bigger the part of the gap occupied by streamers the larger the influence of humidity on U 50 [12, 13]. For short wavefronts due to the large inhibitory action of the first corona the gap is mainly traversed by streamers. For longer wavefronts due to the smaller injected space charge of the first corona breakdown occurs through a leader propagating with a large amount of steps (leader coronas). Thus in both these categories the main proportion of the gap is bridged by streamers either directly (short t f, with no leader formation) or cumulative (longer t f, due to leader coronas) resulting in high values of k s (Figs. 4 & 5). For intermediate front durations the amount of the charge of the first corona, depending on the tip of the rod and on the gap length, either allows the development of a leader of some length propagating continuously (50cm conical tip) or causes the initiation of the second corona to occur near the crest of the impulse thus resulting in breakdown to take place without any subsequent coronas. This results in smaller values of k s (Figs. 4 & 5). With decreasing gap spacing the influence of humidity on the breakdown voltage becomes more strictly related with the influence of the latter on the pre-discharge phenomena. Thus the influence of the tip of the rod on k s becomes more marked as the gap length reduces and also the dependence of k s on the electrode profile can be explained by the same observed influence of humidity on the inception voltage of the first corona. Conclusions For short gaps the transition from a streamer to a leader dominated breakdown mechanism results in non-gaussian distribution functions for the breakdown voltages. The dependence of k s on front duration and on gap configuration and its variability, its values ranging between 1.3 and 0.2, contrast sharply with I.E.C. [7] which supposes a constant value equal to 1. Thus this paper confirms that the short clearances should be regarded as exceptions from the new correction procedure and that a different correction procedure should be more appropriate for short gaps. References [1] IEC Publication 60-1, 1973 [2] AIHARA, Y., HARADA, T., AOSHIMA, Y., ITO, Y.: "Impulse flashover characteristics of long air gaps and atmospheric correction". IEE Trans., 1978, PAS-97, No. 2, pp. 342-348 [3] AIHARA, Y., WATANABE, Y., KISHIZINA, J.: "Analysis of new phenomena regarding effects of humidity on flashover characteristics for long air gaps", IEEE Trans., 1983, PAS-102, No. 12, pp. 3778-3782 [4] PIGINI, A., SARTORIO, G., MORENO, M., RAMIREZ, M., CORTINA, R., GARBAGNATI, E.: "Influence of air density on the impulse strength of external insulation". IEEE Trans. 1985, PAS-104, No. 10, pp. 2888-2900. [5] ALLEN, N.L.: "Corona, breakdown and humidity in the rod-plane gap". IEE Proc., 1986, 133, No 8, pp. 562-568 [6] FESER, K., PIGINI, A.:"Influence of atmospheric conditions on the dielectric strength of external insulation". Electra, 1987, 112, pp. 83-95.
[7] IEC Publication 60-1, 1989 [8] STASSINOPOULOS, C.A., ANDREADOU, K, and SERGAKI, A.: "Influence of humidity on the breakdown mechanism of small rod-plane gaps stressed by positive impulse voltages". etzarchiv, 1990, 12, pp. 273-277. [9] MATALLAH, M., TURRI, R., DAVIS, A.J., and WATERS, R.T.: "Atmospheric correction factor for positive switching impulse breakdown voltages". 7th International Symposium on High Voltage Engineering, Dresden, 1991, paper 42.07. [10] GELDENHUYS, H.J.: "The breakdown voltage of air in a 50cm rod-plane gap over a practical range of air density and humidity". 5th International Symposium on High Voltage Engineering, Braunschweig, 1987, paper 14.02. [11] DAVIS, A.J., MATALLAH, M., and WATERS, R.T.: "The effect of humidity on the positive impulse breakdown characteristics of rod/plane and sphere/plane air gaps". Proc. 10th Int. Conf. on Gas Discharges and Their Applications, Swansea, 1992, 1, pp. 310-313. [12] MIKROPOULOS, P.N., and STASSINOPOULOS, C.A.: "On the influence of humidity on the breakdown mechanism of medium length rod-plane gaps stressed by positive impulse voltages". IEE Proc. -Sci. Meas. Tecnol., 1994, 141, No 5, pp. 407-417. [13] FESER, K., and SCHMID, J.: "Influence of atmospheric conditions on the impulse breakdown of rodplane gaps". Proc. 5th Int. Symp. on High Voltage Engineering, Braunschweig, 1987, paper 11.01