238 NON-UNIFORM FIELD BREAKDOWN OF GASES J.-F. Drapeau, G. Olivier, Y. Gervais Department of Electrical Engineering Ecole Polytechnique de Montreal P.O. Box 6079, Station A , Montreal, Canada, H3C 3A7 INTRODUCTION In 1977, a new law was proposed to describe the uniform field breakdown of gases: a simple empirical relation enabled the prediction of breakdown voltage of pure gases. Later, this approach was successfully applied to gas mixtures. This paper presents an attempt to extend this method for evaluation of breakdown voltage under non-uniform field conditions. THE EMPIRICAL BREAKDOWN L A W Past works have clearly shown that, under uniform field condition, breakdown voltage of gases can be satisfactorily described by a simple empirical law: U - fiG* do* t where: U is the dc or peak ac breakdown voltage in kV, d is the gap length in millimeters, 6 is the molecular density in mole per liter, and G, 0 , t constitute a set of experimentally determined coefficients which characterizes each gas. In the case of binary gaz mixtures, the knowledge of the coefficients of individual gases is sufficient to predict the behavior of the mixture [3]: where U1 and U 1 are the breakdown voltages of the individual gases which may be calculated from equation 1, and n2 is the concentration of gas 2 in p.u. Up to now, as long as the field uniformity is maintained, these equations appear to apply over an extremely 239 wide range of experimental conditions and for all the gases tested in laboratory or for which data was available in literature. Furthermore, it was recently shown that equation 1 is consistent with Townsend's theory [ 4 ] . EXTENSION TO NON-UNIFORM FIELD CONDITIONS Based on the knowledge of uniform field breakdown, two assumptions were postulated for non-uniform field behavior. First, it was suggested that the configuration of the electric field, if not too distorted, does not affect the molecular density term (coefficient G) of equation 1. This hypothesis was based on the fact that the breakdown mechanism does not change as long as the maximum stress to the mean stress ratio does not exceed five [5]. Secondly, it was proposed that all gases should have a similar behavior under identical non-uniform field configuration. Unlike the first hypothesis, the second one was not sustained by experimental evidence. Two different electrode configurations where chosen for the experimental work: small, 1.25 inch in diameter sphere electrodes and sphere-to-plane electrodes consisting of a 4 inch diameter disk and a 1.25 inch sphere. The cell used, described in [l], enables tests to be carried out with pressure ranging from 5 to 700 kPa and gaps up to 60 mm. The high voltage generator is limited to 150 kV rms. Seven gases were tested: nitrogen, carbon dioxide, helium, argon, freons 116, 115 and 12. Figures 1 to 4 and table 2 present a partial but representative sample of the data recorded. EFFECT OF PRESSURE To infer the hypothesis that the electric field configuration does not modify the exponential relation existing in uniform field between breakdown voltage and molecular density (pressure), the basic aspect of the disruptive voltage plots must be preserved. Therefore, the breakdown voltages, for a given gap length, must remain straight lines when plotted on logarithmic scales, and the slope of these lines must conserve the same value than in uniform field conditions. It has to be remembered that this slope corresponds to the G coefficient. Experimental results confirmed this hypothesis. Figures 1 and 3 clearly show that the first condition is well 240 respected. In these figures, there is no visible differences between data obtained in uniform field (curves 1 and 2) and data obtained with gradually distorted fields. Table 1 summaries the value of the G coefficient obtained in uniform field conditions [ 1 , 2 ] and in non-uniform field with sphere-to-sphere and sphere-to-plane electrodes for various gases. This table clearly indicates that G is constant and, therefore, is not affected by the field configuration. Except for freon 12, experimental dispersion easily explains the minute differences found in this table. In the case of freon 12, severe decomposition occurred after only several breakdowns. TABLE 1: Values of coefficient G in uniform and nonuniform field configurations EFFECT OF GAP LENGTH The second hypothesis suggested that, for a given field configuration, the reduction of breakdown voltage, expressed as a ratio of the actual breakdown voltage to the uniform field value, should be independent of the nature of the gas. Figure 5 clearly shows that this assumption was no supported by experimental evidence. The different gases behave very differently. The first group of gases (nitrogen, carbon dioxide, helium and argon) shows a similar pattern of breakdown voltage reduction with gap length and almost supports our hypothesis. However, further tests with several freons shown a completely different pattern. These results indicate that other factors than molecular density and gap spacing affect dielectric withstand in non-uniform field . 24 1 Nevertheless, data obtained in non-uniform field, can still be described by equation 1: where G has the same value than in uniform field, and 0' and t' have different values that are valid only for only one electrode configuration. Table 2 gives values of 0' and t' for the two electrode configurations studied. These coefficients are valid only for the straight lines appearing on right half-planes of figures 2 and 4, where the field configuration is non-uniform. In the fuzzy area between uniform and distorted fields, both equations 1 and 3 do not apply; the breakdown voltage evolves smoothly from one mode to the other. In that area of quasi-uniform fields, equation 1 and 3 overestimate the disruptive voltage for the various gases with a maximum error varying from 4 to 12 percent. CONCLUSION In non-uniform field conditions, the exponential relation existing between breakdown voltage and molecular density (pressure) is not affected. However, it was found that the different gases d o not have the same behavior under non-uniform field configuration. The breakdown voltage is affected by the nature of the gas. Particularly, the disruptive voltage of freons is greatly reduced by the non-uniformity of the field. REFERENCES [l] OLIVIER G., GERVAIS Y., MUKHEDKAR D., "A New Approach to Compute Uniform Field Breakdown of Gases", IEEE Trans. on Pow. App. Syst., Vol. PAS-97, No.3, pp. 969-976. [2] OLIVIER G., DAIGNEAULT G., GERVAIS Y., "Uniform Field Breakdown of Fluocarbons and Other Gases", IEEE 1984 CEIDP Annual Report, pp. 358-363, 84CH1994-3. [3] DAIGNEAULT G., OLIVIER G., GERVAIS Y., GALARNEAU A . , "Uniform Field Breakdown of Gas Mixtures", IEEE 1985 CEIDP Annual Report, pp. 130-136, {I 85CH2165-9. [ 4 ] DAIGNEAULT G., OLIVIER G., GERVAIS Y., "Uniform Field Breakdown of Gases, Revisited", IEEE 1986 CEIDP Annual Report, pp. 316-320, {/ 86CH2315-0. [ 51 ALSTON L.L. , "High Voltage Technology", Oxford University Press, London, 1968, pp. 17-54. 242 I I hV 917054 NITROGEN J z 1 zw CI 0 6 v 0) m 100 m -0 c C z 50 1: 10 mm 2 15 mm 3: 20 mm 4: 25 mm 5: 30 mm 6: 35 mm 7: 40 mm 0 U Y m 9) L m 8: 45 mm 9: 60 mm 10 / 01 I L 0.1 1 Molecular density (rnole/liter) Fig. 1 : Constant electrode spacing curves for nitrogen LV P 200 I 7 5 0 CI 0 6 W 9) w 100 m -0 c C % 50 0 U Y m f m NITROGEN / " Q 10 6: 7: 8: A D: 1.25' 9 0.020 0.041 0.061 0.082 0.102 0.123 0.143 0.163 0.184 m/l m/l m/\ m/l m/l m/l m/l m/l m/l L 0 50 Gap length (rnrn) Fig. 2 : Constant density curves for nitrogen 10 243 kV FREON 116 7 9 1 6 5 PW 4 a 2 1 50 5rnrn 7.5 rnrn 1: 2 3: 10 rnrn 4: 15rnrn 20 rnm 5 6 30 rnrn 7 40 rnrn 8: 50 rnrn 9 60 rnrn 10 I 0.1 0.05 aoi Molecular density (mole/liter) Fig. 3 : Constant electrode spacing curves for freon 116 LV . . . . . .., . 6 200 . 5 4 loo 50 10 10 Gap length (mm) Fig. 4 : Constant density curves for freon 116 100 244 TABLE 2 : Coefficients G.0.T in uniform and non-uniform fields 1.0 I 0.8 I 10 I 50 Gap length (mm) Fig. 5 : Non-uniformity breakdown voltage ratio 100