On the Mechanism of the “Barrier Effect” in Pulsed Breakdown of Liquids

V. Ya. Ushakov

(Presented by the scientific seminar of the Research Institute of High Voltages at TPI)

Introduction

The “barrier effect,” as is known, is understood as an increase in the breakdown voltage of a discharge gap when a barrier made of insulating material is installed in it. Despite the many years of practical use of barriers to increase the reliability of operation and reduce the dimensions of apparatus and electrical installations, the mechanism of the “barrier effect” has been insufficiently studied, especially for liquid insulation. The authors of most works devoted to the study of the “barrier effect” in liquids \[1, 2, etc.\] limited themselves to obtaining quantitative data for certain conditions and did not provide sufficiently substantiated explanations of the observed effects.

Of great interest for analyzing the mechanism of the “barrier effect” are attempts to explain the mechanism of barrier action under pulsed breakdown of long air gaps, undertaken in the works of V. S. Komelkov and A. M. Lifshits \[3\], N. N. Nikolaevskaya \[4\], and I. S. Stekolnikov \[5\].

In work \[3\], the main role in the “barrier effect” is assigned to space charges deposited on the barrier as a result of the development, in the rod–barrier gap, of “streamer” processes (the primary pulsed corona and the leader corona). These charges retard the development of the leader channel in the rod–barrier gap and lengthen its trajectory, thereby causing an increase in the breakdown voltage of the discharge gap. A sufficient condition for completion of the discharge in a gap with a barrier is considered to be breakdown of the barrier by the leader.

According to the data of \[4\], the “barrier effect” is caused by the action of the barrier as a mechanical obstacle to streamers. It is noted that breakdown of the gap is possible only if the voltage across the gap provides, beyond the barrier, the gradients necessary for the development there of impact ionization and the formation of streamers. A sufficient condition for completion of the discharge is considered to be complete bridging of the discharge gap by a streamer, leading to breakdown of the barrier and of the discharge gap.

According to \[5\], breakdown of the barrier occurs at the stage of the pulsed corona under the action of a strong electric field localized at the end of a corona branch. The gradients of the external electrostatic field beyond the barrier in this case need only be no less than \(1\ \text{kV/cm}\),

in which a pulsed corona can develop. Breakdown of the barrier at the pulsed-corona stage is necessary in order to release in the gap a certain amount of space charge participating in the formation of the leader channel.

Earlier studies of the spatiotemporal development of discharge in liquids in large discharge gaps with a nonuniform field distribution [6, 7] showed that, in discharges in liquids, the primary pulsed corona and the leader corona characteristic of discharge in long air gaps are absent. In liquids the ionization zone developing from the head of the leader channel or from the electrode is small and constitutes a spherical diffuse glow about 1–3 mm in diameter. These results give grounds to suppose that the deposition of charges on the surface of the barrier as a result of the development of ionization processes cannot substantially change the conditions for leader development in the rod—barrier gap and, consequently, cannot play the determining role in the mechanism of the “barrier effect” in pulsed breakdown of liquids. Experimental studies confirmed this supposition.

Experimental results and their analysis

The development of discharge in transformer oil and distilled water was investigated in rod—plane gaps for positive (+ R—P) and negative (−R + P) polarities of a voltage wave of 1.5/70 μsec. Most of the investigations were carried out in gaps 50 mm long. As the barrier, sheets of celluloid (395 mm × 295 mm), 0.13 mm thick, were used; to impart the required rigidity to the barrier they were fastened to a textolite frame. The frame was suspended on four threads and could be moved relative to the electrodes. In some experiments barriers of organic glass 3 mm thick were used. The use of barriers made of porous materials in these experiments is inadvisable, since their impregnation in one case with transformer oil and in another with distilled water will affect the electric strength of the barrier itself, which will rule out the possibility of comparing the results obtained. To investigate the prebreakdown period of the discharge and to record the instant of barrier breakdown, simultaneous recording was made of the voltage across the gap, of optical spatiotemporal patterns, and of static discharge patterns. An electron-optical converter with light amplification was used to record the spatiotemporal discharge patterns. The circuit, the experimental procedure, and a description of the construction of the electrodes and discharge chambers are given in [8].

When a voltage pulse is applied to a discharge gap with a barrier, the discharge, as in the case without a barrier, begins near the rod electrode in the form of a leader developing into the depth of the gap. In structure and development velocity the leader process is then similar to the leader process in a rod—plane gap without a barrier [6]. At a voltage below the discharge voltage, the leader channel approaches the barrier to a distance of 3–5 mm, after which it changes the direction of its motion almost at a right angle. Over a length of 5–10 mm the leader channel develops above the barrier, then is pressed against it and subsequently develops as a sliding discharge.

At a certain value of the voltage, some of the applied pulses cause breakdown of the barrier at points located at some distance from the projection onto the barrier of the initial trajectory of the leader motion. A further increase in the amplitude of the voltage pulse leads to a decrease in the distortion of the trajectory of leader motion by the barrier.

and the barrier has practically no effect on the development of the leader process. Breakdown of the barrier by the leader is always accompanied by complete breakdown of the discharge gap. The described discharge pattern is observed irrespective of the position of the barrier in the discharge gap. Figure 1 shows the dependence of the percentage increase in the 50% breakdown voltage of R–P gaps in transformer oil on the position of the celluloid barrier.

It is seen from Fig. 1 that the maximum increase in the strength of the gap (30%) due to the barrier is observed in the electrode system \(+\,\mathrm{R}-\mathrm{P}\) when the barrier is located practically close to the rod electrode. As the barrier is moved away from the rod, the increase in the strength of the gap due to the barrier decreases to 15%. In the gap \(-\,\mathrm{R}+\mathrm{P}\), for all barrier positions, the barrier effect is manifested

Fig. 1. Dependence of the percentage increase in the breakdown voltage of \(+\mathrm{R}-\mathrm{P}\) (1) and \(-\mathrm{R}+\mathrm{P}\) (2) gaps in transformer oil on the position of the celluloid barrier \((d = 0.13\ \mathrm{mm})\) at \(S = 50\ \mathrm{mm}\).

to a lesser extent. The breakdown voltage of the celluloid barrier in a nonuniform field under \(1.5/70\ \mu\mathrm{s}\) impulses is about \(6\text{–}7\ \mathrm{kV}\).

The increase in the strength of the discharge gap in distilled water due to the barrier is small and, for positive polarity of the rod, varies from 13% (barrier at the rod) to 7% (barrier at the plane), Fig. 2. For negative polarity of the rod, the strength of the gap increases only by 8.5% when the barrier is located at a distance of \(10\ \mathrm{mm}\) from the rod. When the barrier is moved away from the rod to a distance of more than \(30\ \mathrm{mm}\), the increase in breakdown voltage lies within the limits of experimental error.

The discharge pattern in gaps with a barrier and the qualitative mechanism of the “barrier effect” may be represented as follows. In the rod–barrier gap, as in the rod–plane gap, the leader develops due to ionization processes near the head of its channel at a distance from it of the order of \(1\text{–}3\ \mathrm{mm}\). The barrier begins to exert a substantial influence on the development of the leader

only when the leader approaches the barrier to a distance commensurate with the dimensions of the ionization zone. From this moment on, the space charge begins to slide along the surface of the barrier and to retard the motion of the leader toward the barrier.

At a voltage insufficient to complete the discharge in a gap with a barrier, ionization processes develop in the liquid above the barrier, i.e., a creeping discharge develops. If in the material of the barrier there are defects (microcracks, gas inclusions, etc.) lying in the path of the creeping discharge, breakdown of the barrier may occur at the weakened spot. The presence in the barrier of a small number of through holes up to 2 mm in diameter has practically no effect on the “barrier effect.” The leader passes through the barrier via a hole only if it is located in the path of the leader’s motion.

Fig. 2. Dependence of the percentage increase in the breakdown voltage of the gaps $-\mathrm{P}+\mathrm{S}$ (1) and $+\mathrm{P}-\mathrm{S}$ (2) in distilled water on the position of the celluloid barrier $(d = 0.13\ \mathrm{mm})$ at $S = 50\ \mathrm{mm}$.

At a certain field strength at the head of the leader channel, breakdown of the barrier occurs. The critical value of the potential of the rod electrode that causes breakdown of the barrier depends on the electric strength of the barrier and on the ratio of the dielectric constants of the liquid and the barrier material. It is probable that breakdown of the barrier occurs as a result of the development in it of ionization processes in a field of strength $10^6$–$10^7\ \mathrm{V/cm}$, localized near the head of the leader channel. Melting of the barrier under the action of the high temperature of the leader channel is unlikely, since during development of the discharge along the surface of the barrier no substantial melting of the barrier surface is observed.

When the leader approaches a barrier made of celluloid $(\varepsilon = 3.5)$ in transformer oil $(\varepsilon = 2.3)$, the field strength in the barrier and in the oil layers adjacent to it is approximately the same. For distilled water, the field strength in the barrier is 23 times higher than in the adjacent layers of water, and breakdown of the barrier in this case occurs at lower values of the voltage on the electrodes. In distilled water, the use of a barrier in gaps with sharply non-

a uniform field is accompanied only by a slight increase in the breakdown voltage (Fig. 2), while in gaps with a uniform and weakly nonuniform field, according to E. Marx [9], it even causes a substantial decrease in the breakdown voltage.

Thus, the most probable determining factor in the “barrier effect” during breakdown of liquid dielectrics is the action of the barrier as an obstacle hindering the development of the leader process. At the same time, owing to the small dimensions of the ionization zone propagating ahead of the leader channel, the charges deposited on the surface of the barrier can play a definite role only in the development of intense creeping discharges. Under the action of creeping discharges the surface of the barrier becomes charged, which leads to a redistribution of the field in the gap. As the barrier is brought closer to the electrode of greater curvature, the intensity and number of creeping discharges increase, since almost every one of the numerous leader channels developing from the rod reaches the barrier. The latter may intensify the “barrier effect.”

Some role in the “barrier effect” is probably also played by the distribution of the field among the capacitances formed by the electrode–barrier and barrier–second-electrode systems, as was suggested by A. A. Shilvan [2].

REFERENCES

1. A. V. Panov, L. A. Mravyan. Problems of Transformer Engineering. Tr. VEI, issue 31, 129–149, 1938.
2. A. A. Shilvan. TVN, Tr. LPI, No. 1, 1954.
3. V. S. Komelkov, A. M. Lifshits. Izv. AN SSSR, OTN, No. 10, 1950.
4. N. N. Nikolaevskaya. TVN, Tr. LPI, No. 1, 1954.
5. I. S. Stekolnikov. The Nature of the Long Spark. Publishing House of the USSR Academy of Sciences, 1960.
6. I. S. Stekolnikov, V. Ya. Ushakov. ZhTF, Vol. XXXV, issue 9, 1965.
7. V. Ya. Ushakov. ZhTF, Vol. XXXV, issue 10, 1965.
8. V. Ya. Ushakov. Breakdown of dielectrics and semiconductors. In: Proceedings of the IV Interuniversity Conference on Breakdown of Dielectrics and Semiconductors, 207–211. Energiya, Moscow–Leningrad, 1964.
9. E. Marx, ETZ, Bd. 49, S. 50–53, 1928.