IEEE Transactions on Power Apparatus and Systems, vol. PAS-94, no. 2, March/April 1975 INFLUENCE OF THE DELAY OF THE FIRST CORONA PULSE ON THE SNITCHING IMPULSE BREAKDOWN PROBABILITY C. Menemenlis H. Anis Senior Member IEEE Member IEEE Hydro-QuBbec Institute of Research Varennes, Quebec, Canada. ABSTRACT Tests with a rod-plane gap with a rounded termi- nation of therod show t h a t the delay of the first co- rana pulse affects the positive switching impulse breakdown probability. The inverted S form or local in- consistencies of the breakdown probability distribution are connected with particular statistical features of the delay of the first corona pulse. The influence of therodterminationortheimpulsefrontanthebreak- down p r o b a b i l i t y distribution is explained in terms of the delay of thefirstcoronapulse. IWl'RODUCTION In air insulation design, it is usually assumed that the breakdown probability as a function of the prospective voltage follows a Normal (Gaussian) cumula- tivedistribution. It has been observed, however, that under certain conditions the breakdown probability dis- tribution (BPD) deviates appreciably from a Normal one. In reference 1 for example, it is shown that the voltage-distance characteristics of rod-plane gaps stressed by positive switching impulses, present a transition range for gap clearances between 20 and 100 cm. Within this range of clearancesthe BPD, p l o t - ted on Normal probability paper, takes the form of an inverted S. The inverted S characteristic was attrib- uted to the existence of two different mechqnisms of breakdown, namely, leader and streamer coronas, the former mechanism prevailing at lower probabilities of breakdown, t h e latter at higher probabilities. However, the inverted S characteristic has also been observed f o r a 3-m rod-plane gap with the rod terminating in a 30-cm diametersphere2. Recent tests i n IREQ's high voltage lahoratory, using a 3-m rod-plane gap and positive switching im- pulses, reveal that the shape of t h e BPD of rod-plane gaps depends on the termination of the rod as well as on the front of the impulse. It has been observed namely that when therodterminatesin a sharp cone the BPD, for the tested 3-m gap, follows very closely a Normal. When t h er o dt e r m i n a t e si n a sphere between 20 and 50 cm in diameter, however, t h e BPD plotted on normal probability paper displays the inverted S char- acteristic. Furthermore, when the rod terminates in a sphere or a conewith a rounded tip, frequent inconsis- tencies were observed at different points of the char- acteristic; such inconsistencies were not usual for the sharp cone. For a sphere diameter over 60 CCI, the in- verted S characteristic as well as the noted inconsis- tencies disappeared, at least for the gap distance and the impulse fronts used, and the BPD again approached a Normal. Distribution Committee of the IEEE Power Engineering Society for presentation at Paper T 74 456-0, recommended and approved by the IEEE Transmission & the IEEE PES Summer Meeting & Energy Resources Conf., Anaheim, Cal., July 14- 19, 1974. Manuscript submitted January 30, 1974; made available for printing May 14,1974. The deviation of the BPD from a Normal constitutes a problem in the design of air insulation. This problem ariees more frequently for station apparatus due to the use of shielding electrodes. The inconsistency of the breakdown probability may in addition introduce a prob- lem in the design of protective gaps. In the present work it is shown t h a t , f o r a given gap and a given applied impulse, the breakdown proba- bility depends on the delay of the first corona pulse (FCP). It is thus possible to conclude that any factor affecting the FCP d e l a y should also generally affect the breakdown probability and, consequently, the re- l a t e d BPD. The observation that the front of the im- pulse and the rod termination affect the BPD does not come as a surpriseinviewofthefactthatthese same factors are affecting, the delay of the FCP. It should be noted, however, that the PCP d e l a y is also inf lu- enced by factors of a less deterministicnaturesuchas the electrode surface condition or free electron den- sity.Inconsistenciesinthe BPD could consequently be attributed to occasional fluctuations of factors such asthese. The correlation of the FCP delay to the probabil- ity of breakdown, established by the present work, offers a new perspective for the probabilistic study of thebehaviorofair-gaps.Furthermore,incertaincases it allows factors previously regarded as purely sto- chastic,tobetreatedasdeterministic. TEST ARRANGEMENT AND PROCEDURE The experimental results quoted in the present paper were collected from tests on a 3-10 rod-plane gap with the rod terminating in a rounded cone (r = 2.5 cm) or a sphere with a diameter between 20 and 75 cm. The voltage was generated by an 8-stage, 3.2 W (charging voltage), 200-kWs i m p u l s e generator and was measured with a damped capacitive voltage divider and a crest voltmeter. The FCP delay was measured by a capacitive probe which detected field variations at the plane. The time-to-breakdown was recorded using the signal of the voltagedivider and a digitalcounter. The results on which the present analysis is based were acquired during a more general study of the be- havior of a 3-m rod-plane gap under switching impulses duringwhich more than 20,000 shots were applied to the gap EXPERIMENTAL RESULTS AND PARTICULAR REMARKS Occurrence of the FCP on the Switching, Impulse The position on the switching impulse at which the FCP o c c u r s can be determined by either the delay t of t h e FCP o r its actual voltage v. If the FCP occurs on the rising part of the impulse, then v identifies the position of the FCP with good accuracy while t is a better measure when t h e FCP occurs close to the crest. In Fig. 1 the mean actual voltage V at which the FCP o c c u r s is given as a function of the prospective amplitude V of the impulse for five different rod- terminations and a 220/5000 us impulse. The different points correspond to impulse amplitudes V within the range 0 to 100% breakdown probability. The theoret- i c a l minimum voltage V a t which t h e FCP s h o u l d occur for the five different rod terminations is calculated 455 Authorized licensed use limited to: Jet Propulsion Laboratory. Downloaded on January 20, 2010 at 22:17 from IEEE Xplore. Restrictions apply.