Düzgün olmayan alanda SF6, N2 ve SF6+N2 gazlarında boşalma gerilimlerine elektrod yüzey pürüzlülüğünün etkileri

Abstract

ÖZET Bu çalışmada, küre-düzlem ve çubuk-düzlem elektrod sistemlerinde, SF,, N` ve SF,+N2 gazlarında, elektrod yüzey pürüzlülüğünün olması ve olmaması durumlarında, (0,5- 4) bar basınç aralığında delinme ve korona başlangıç gerilimlerinin değişimi incelenmiştir. Deneyler, doğru ve 50 Hz alternatif gerilimde yapılmıştır. Elektrod yüzey pürüzlülüğü makroskopik yönden ele alınmıştır. U,=f(p) eğrileri, SF, ve SF,+N2 gaz karışımında bir maksımum-minimum davranışı gösterdiği, N~ gazında ise 0` etkisinden dolayı sözkonusu davranışın çok az ortaya çıktığı saptanmıştır. SF,+N2 gaz karışımlarında kritik basıncın N2 oranı ile arttığı ve dolayısı ile korona bölgesinin genişlediği gözlenmiştir. Çubuk-düzlem elektrod sisteminde, (0,5-4) bar basınç aralığında, U,(-)>U,(+) olduğundan sözkonusu elektrod sistemi bir sivri uç-düzlem eleketrod sistemi gibi davranmıştır. Deneylerde çalışılan basınç aralığında, korona başlangıç gerilimleri hemen hemen basınçla lineer olarak artmıştır. Ayrıca, alanın düzgünsüzlük derecesi azaldıkça, SF, gazının, SF,+N~ gaz karışımına oranla, yüzey pürüzlülüğüne çok hassas olduğu ortaya çıkmıştır. Elektronegatif etkisi gösteren gazlarda, U,=f(p) eğrisinin negatif olduğu bölgede, yüzey pürüzlülüğünün delinme gerilimine hemen hemen etki etmediği, bu bölgenin dışında ise etkisinin var olduğu saptanmıştır. Ayrıca SF,+N2 gaz karışımlarında sözkonusu etkinin N` gazının oranı ile değiştiği gözlenmiştir. SF,+N? gaz karışımlarında, 50 Hz alternatif ve pozitif doğru gerilimde, minimum delinme gerilimi elektrod sistemlerine bağlı olmaksızın hemen hemen sabit kaldığından, bu gerilimin hesaplanması için bir ampirik formül elde edilmiştir. Bu formülün, 50 Hz alternatif gerilimde SF, oranının %50 den büyük olduğu SF,+N2 gaz karışımlarında % 1-2 hata ile geçerli olduğu gözlenmiştir. Diğer yandan az düzgün alanlarda U,=f(p) eğrisinde maksimum-minimum davranışı bulunmadığından, sözkonusu formülün, sadece düzgün olmayan alanlar için geçerli olduğu ortaya çıkmıştır. Kullanılan elektrod sistemlerinde %40 SF,+%60 N` gaz karışımı durumunda, delinme gerilimi değerlerinin negatif doğru gerilimde %60 SF6+%40 N gaz karışımınınki ile yaklaşık aynı olduğu, 50 Hz alternatif ve pozitif doğru geriliminde ise, %60 SFg+%40 N2 gaz karışımınınkinden daha yüksek olduğu görülmüştür. vı

SUMMARY EFFECTS OF ELECTRODE SURFACE ROUGHNESS ON CORONA INCEPTION AND BREAKDOWN VOLTAGES IN SF., N` AND SF,+N0 IN NON-UNIFORM FIELDS 6 2 6 2 Sulphur-hexaf luoride has good dielectric and heat transfer properties and it is extensively being used in power apparatus as a dielectric medium. Besides being expensive, it has a relatively high boiling temperature. It is very sensetive to strong localized fields often encountered in practical systems due to electrode surface imperfections and the presence of free particles. There fore in practical systems, the ideal Paschen's law break down strength is not achived. Furthermore, long non uniform field gaps insulated with compressed SF,. show surprisingly low breakdown strongth under the applica tions of impulse voltages [l]. Measurements of uniform quasi-uniform and non uniform field breakdown voltages have shown that the addition of small amounts, a few percent of SF, to common gases like air and nitrogen, etc., results in an appreci able increase in the breakdown strength of these gases. From the existing information, it appears that SF,+air mixtures show relatively less degree of saturation as compared to SF,+N` mixtures. This is probably due to the presence of electronegative 0~ in the air. Because of the presence of chemically active oxygen in air, SF,+air mixtures are technically less important as compared to SF^+Np mixtures. Although the excellent insulation and arc interrup tion properties of SF, have lead to its widespread use in circuit breakers, recent studies have shown the possibi lity of further enhancing these properties by using SF, mixed with lighter gases such as N~. Garzon [2] have studied the comparative interruption properties of SF,+N` mixture. By measuring the rate of rise of recovery or voltage (RRRV) for a synchronous interrupter, Garzon has shown that the performance of SF,+N` mixtures having %50 SF, by volume at pressures of 1300 to 1900 kpa is approximately 1,39 times better than that of pure SF,. He also found that the recovery capability of a non- synchronous breaker using this gas mixture was at least as good as when pure SF,. was used. Other advantages such as shorter times for pneumatic operations and the use of higher total pressures without liquif ication make mixtures of SF, more attractive and useful for further applications in circuit breakers. Synchronous breakers using SF,+N2 viimixtures are already in operation. Recent studies of SF,+N` mixtures have revealed that such mixtures are less sensitive to the presence of contamination and electrode surface roughness as compared to pure SF, [3]. Furthermore, SF,+N2 mixtures have the advantages of lower boiling points and are less expensive. The lower boiling point of the mixtures is rather impor tant in colder climates. Previous investigations show that in a highly non-uniform field, breakdown voltage versus pressure curve for compressed SF, and SF,+N` gas mixture, exhibits a maximum-minimum characteristic. Such a field distribu tion can be obtained by making use of the point-plane or rod-plane electrode configurations. Experimental data show that in a highly divergent field, within a given pressure range, breakdown is always preceded by corona discharge. However corona may be avoided if the gas pressure is increased above a critical value. In literature the above phenomena is mentioned as `corona breakdown` and `direct breakdown` respectively. In fact, for a given electrode configuration, there is a transition between `direct breakdown` and `corona breakdown`. In transition region, the pressure beyond which the breakdown is direct breakdown and where corona inception and breakdown voltages coincide is called `critical pressure`. Effect of electrode surface roughness on breakdown in SF, and SF,+N` mixtures in uniform-field and coaxial electrode systems have been investigated by Pederson and Farish respectively [3], [4]. All of them used protrusions with microscopic sizes, but there is not available data about macroscopic sizes of protrusion, that can be obtained by production imperfections and presence of metallic particles. According to above facts, the main purpose of this work is therefore to investigate the effect of polarity, field non-uniformity, gap spacing and pressure on corona inception and breakdown voltages of compressed N«, SF, and SFft+N« gas mixtures, with and without presence of macroscopic sizes of electrode surface roughnesses. Experiments are carried out with 50 Hz AC and DC voltages up to the peak value of 280 kV in both polari ties, using rod-plane, sphere-plane electrode systems. To obtain various degrees of non-uniformity, the sphere radius is changed and also to reach higher degree of non-uniformity, radius of 1 mm rod electrode is selected. To obtain desired degree of roughness hemisphere with heights equal to 3 mm and 6 mm are mounted on the plane electrodes for each configurations. THe electrodes are mounted in a pressure vessel made of polymethyl metacrylate (plexiglass). All electrodes are made of Vlllbrass covered with chromium (Although the rod electrode used during the experiments is hemispherically capped with a radius of 1 mm). The lower electrode in both arrangements is a plate electrode with an overall diameter of 75 mm and its edges are rounded with a radius of curvature of 3 mm. For the 50 Hz AC test, with voltages up to 100 kV r.m.s. a MWB 0,22/100 kV, 5 kVA high voltage transformer is employed. To limit the discharge current at breakdown and reduce the erosion at electrode surfaces, a 10 Mfi protecting resistor is connected between the high voltage supply and the test object. Furthermore, the high voltage transformer is disconnected from the mains immediately after breakdown. The peak value of AC voltage is measured by using the method of Chubb and Fortescue. The waveform of the 50 Hz voltage applied to the test object is observed by means of a capacitive potential divider and a Textronix 7623A, dual beam oscilloscope. The corona inception voltage is determined by observing the voltage drop due to the corona impulse current across a 75 fi resistor inserted in the earth lead. A two stage voltage doubler circuit is used to obtain up to 280 kV DC voltages via a 0,22/100 kV, 5 kVA MWB transformer. Similar to the AC test, an automatic switch and a 50 kfi current limiting resistor are connec ted to the DC supply and test object. The DC voltages are measured with twot 140 MÎ2 resistors connected in series and a mA-meter. DC corona inception voltage is determined in the same way as for AC measurements, but the voltage across the test object is not observed. In all experiments, the applied voltage has been raised at a rate of 5 kV/s until breakdown occurred. Before starting any experiments, electrodes are treated with a metal polish and clean washed carefully with ethyl alcohol. After mounting the electrodes inthe vessel, it is evacuated to a pressure less than 10 Torr for at least 20 minutes, and then the lower constituent is admitted to a partial pressure corresponding to the desired mixture ratio. The second constituent is then added to the predetermined total pressure. Unless other wise mentioned, the percentage of mixture refores to the percentage of SF, in N~ by pressure. Althought the SF, gas used in the experiments is supplied by the Turkish Electricity Authority with a commerical grade and also N` gas is obtained from HABAŞ A.Ş., both gases are stored in a high pressure tanks. To make sure of the stability of gas and gas mixture, the measurements are started 6 hours after filling. During the test, room temperature varies between 11 and 17 C. At least ten different voltage values are taken to estimate the mean value of each measuring point. axThe standard deviations and %95 confidence inter vals are also calculated. Especially in the decreasing part of U,=f(p) curves a relatively higher scattering and a coefficient of dispertion up to %5 are observed. It can be stated without any doubt that the presence of sustained corona discharges, in both positive as well as negative rod-plane and sphere-plane gaps, considerably enhances the breakdown strength of systems insulated with SF, and SF,+N` gas mixtures. The pressure over which this enhancement occurs is often referred to as the `corona stabilized breakdown region`. The high pressure limit of this region at which the breakdown and tha corona onset voltages coincide is known as the criti cal pressure p, The present results clearly indicate that the breakdown of both the negative and positive rod-plane, sphere-plane gaps is corona stabilized in the low pressure region. At high pressures, however, the break down occurs in the absence of any corona discharges. The width of the corona stabilized breakdown region is much smaller for the positive rod-plane and sphere-plane gaps as compared to the negative ones. Consequently, the critical pressure for the negative gaps is higher than that observed for the positive ones as shown in Figures 4.8 and 4.9. Furthermore, the breakdown voltages for negative gaps are higher than those of positive gaps at the low pressure range of the corona stabilized region. However, the reverse behavior is observed at higher pressures where the breakdown occurs without corona. In the corona stabilized region at low pressure, the break down voltages increase linearly with pressure for both polarities of applied voltage. The absolute width of the pressure region, where the breakdown voltage-pressure curve is linear is much wider for the negative gaps, photographic records and visual observations revealed that in this region, the sparks follow straight paths for both polarities of the applied voltages. As the pressure in increased, the breakdown voltages attain maximum values followed by reduction in magnitudes with increasing gas pressures. It is quite likely that extremely non uniform field gaps of negative polarity with smaller diameter cathods have maximum-minimum in the breakdown voltage-pressure curves especially at high values of the gas pressures not investigated in our studies. The corona inception voltages is higher for the positive gaps as shown in figures of Chapter 4. Furthermore, the magnitudes of the average prebreakdown currents for a given applied voltage and gas pressure are larger in the negative gaps than those observed for the positive gaps. This is probably due to the fact that the secondary ionization process such as photo ionization ou the gas, photoemission from the cathode, and electron emission from the cathode due to positive ion bombardment are more effective when the rod or sphere cathode is in the highfield region such as in the negative rod-plane and sphere-plane gaps. Furthermore, when the highly stressed electrode is the cathode, the chances of field emitted electrons, initiating corona discharges are present espe cially at high gas pressure. These reasons and the fact that it is easier for the positive than the negative streamers to propagate and bridge the gap make the corona stabilization processes more effective especially at low pressures, for the negative gaps. Consequently, the corona stabilized breakdown region is wider in the case of the negative rod-plane and sphere-plane gaps than that of the positive gaps. At low pressures, the breakdown voltage of SF, and SF,+N` gas mixtures is higher for the negative sphere- plane gaps. This is due to the fact that it is more difficult for the negative streamers to advance in a divergent field as compared to the positive streamers. In the case of negative sphere-plane gaps, an avalanche starts from the highly stressed electrode and the elec trons diverge outward from a high field to a lower field region. Thus any electrons created by photoionization will move ahead of the streamer to create new avalanches and thus move into a rapidly declining field region. Furthermore, when the streamer propagates in the presence of corona discharges, the low field region is filled with slowly drifting negative ions. This makes the streamer propagation forther into the gap extremely difficult unless the external fields are high enough to ensure the, effective fields at the streamer tip â 0,8775 kV (cm.kpa) in the case of pure SF,. On the other hand, for a posi tive sphere-plane gap the electrons created ahead of the avalanche by photoionization travel towards the high field regions there by creating more positive ion space charge at the avalanche head. In this fashion, the posi tive ion space charge is advanced towards the cathode maintaining the field at its tip and creating more elec trons a head of itself to continue more advance. This not only results in a lower value of the breakdown voltage for a positive gap, but, also, results in a much lower value of the critical pressure in such gaps. Since rod-plane configuration with rod radius equal to 1 mm is an extremely non-uniform gap, therefore, the corona stabilized breakdown voltages for such gaps are higher than sphere-plane electrode configuration with larger sphere radius. Ultimately, negative breakdown voltages of rod-plane electrode system are higher than positive breakdown voltages in the pressure range between 0,5 and 4 bar. Under 50 Hz AC voltages, corona pulses appeared at first on the negative half-cycle, perhaps conforming an earlier observation under dc voltage. Under higher voltage amplitudes corona pulses appeared over the xipositive half-cycle, and at a yet higher voltage, they are split into two populations on both sides of the peak of the AC wave. mcep diffe rough SF,+N have For e witho has h betwe In tion rent nesse 9 gas diff xampl ut su igher en 1 Chap volt elec s. mix eren e, i rf ac val and ter ages trod In g ture t va n he e ro ue t 1,57 5, w of e ge ener s wi lues misp ughn han bar e co SF6f omet al, th a due here ess %100, wh mpar and ries brea nd w to -pla unde ich e breakd SF6+N g, with a kdown vo ithout s specif ie ne type r DC (+) for the is shown own and as mixtu nd witho ltages o urface r d pressu electrod voltage pressur in Figu corona res for ut surface f SF6 and oughnesses re region, e system, %80 SF6 e region re 5.2. attri meter distr of pu mean as we compa photo addit tive the p pure This buted s. It ibutio re SF6 energy 11 as red to n abso ion of ioniza ressur beh to t is ns i cas of diff pur rbti > tion e re ].[6 avio he p very n th e f o the usio e SF on c [5]. coe duce ur of t ossible probab e mixtu r certa electro n coeff,. AİS oef f ıcı Furth f f icent d field he SF, chang le tha re wil in mix n-swar icient o, it ent of ermore of th is le +N2 es i t th 1 be ture m, d s wi is p.'& e mi ss c mixt n it e el dif rat rift 11 b ossi wil e va xtur ompa ure s tr ectr fere ios. vel e di ble 1 ch riat e wi red may ansp on e nt t H ocit ffer that ange ion th r to t be ort para llergy han those ence, the ies and ent the with of effec- espect to hat of It is observed that SF^+N^ gas mixtures are not so sensetive to electrode surface roughness in the case of macroscopic size protrutions used during the tests. All of these are explained briefly in Chapter 5. Shifting of critical pressure in gas mixtures to higher values than pure SF, particularly, in extremely non-uniform fields, is an interesting result observed in breakdown versus pressure characteristics of SF,+N« gas mixtures. The ritical one shown in fi voltages fo using the e to corona i guration, a quasi-unif o the radius in non-unif with respec equations a voltages fo gurations w Experimenta predicted v compar s for gures r all quatio ncepti nd als rm fie of sph orm f i t to p re onl r sphe ith ra 1 resu alues. ison the of C elec n (2 on v o it Id a ere elds osit y us re-p dius Its Th of m three hapte trode.34). oltag is v rrang elect » neg ive o efull lane of 2 are s is ma easu ele r 6. arr Gi es i alid emen rode ativ nes, to and mm ligh y be red ctro Th ange ven n no to t, t is e co the pred rod- and tiy due val de c eori ment equa n-un brea hat equa rona ref o ict plan 1 mm high to ues w onf ig tical s are tion if orm kdown is ob 1 to have re, t DC (- e ele resp er or the f ith t urati thre comp is pr fiel volt taine 4 mm. lowe heori ) cor ctrod ectiv lowe act t he t ons shol uted opos ds c age d wh Si r va tica ona con ely. r th hat heo- are d by ed onf i- in en nee lues 1 fi- an xixtheoritical values were based on the assumption that the plane electrode radius was infinite. In this investiga tion, the plane electrode had a diameter of 75 mm only. Therefore, the calculated field utilization factors may be smaller than the actual values. Thus, measured incep tion voltages can be higher or lower than predicted values. Dispersion between calculated and measured values are observed specially for higher non-uniform field geometries. This is due to the fact that corona inception voltages are not so stable in non-uniform fields particularly at lower pressure. The minimum breakdown voltages and the pressure at which this minimum occurs are important factors in the design of SF, and SF,+N` gas mixtures. That is used in high-voltage apparatus. Test results show that, for a given gap spacing (U,). and n.p. values are approximately constant irrespective of the electrode configuration and gas pressure. Sangkasaad in 1975, from optical observation has found out that in the pressure range p<p there is a uniform corona discharge at rod tip with a few millimeters length. The spark followed the shortest distance between electrodes. By increasing the pressure to p the streamer discharges are developed into different direc tions from the rod tip. This means that space charge increases the radial field strength and introduces a spark in that direction. In the region around p, the spark discharge doesn't follows the shortest patn between electrodes, but, advances along a curved path to the cathode. Around p the spark discharge will again follow the shortest way. According to the above fact, we observed that in non-uniform field configurations, around p, the effect of surface roughness is not so significant on the break down voltages in SF, and SF,+N~ gas mixtures. A semi-emprical criterion is obtained for minimum breakdown voltages in SFg and SF,.+N2 gas mixtures which is shown by Equation (2.38) in Chapter 2. Xlll