An Experimental Study on the Luminescence of the Leader Channel During the Relaxation Process Before Restrike in a Positive 6 m Air Gap Discharge

Abstract

Restrike frequently occurs during the positive leader development of long-air-gap discharges. At present, however, its detailed physical process and mechanism remain unclear. To investigate the physical mechanism of restrike, experiments were conducted in a 6 m rod–plate air gap under positive impulses with a wavefront time of 1 ms, and the process of restrike was observed during discharge. Our experimental results showed that significant luminescence appeared at the tip of the leader channel for a relatively long time during the discharge relaxation process before restrike occurred, and the luminescence became increasingly intense as the applied voltage increased until restrike occurred. By analyzing the composition of the charged particles inside the leader channel, we inferred that, during the relaxation process, the positive ions inside the leader channel migrate toward and concentrate in the leader channel tip as the applied electrical field increases, and the concentration of positive ions at the leader channel head distorts and enhances the local field, which then induces streamer corona discharge, leading to the luminescence of the leader channel. The observations, evidence, and discussion presented herein could provide a valuable reference for more effectively understanding the physical mechanism of restrike.

1. Introduction

Positive leader discharge of a long spark generally originates from an energized anode. Before the spark bridges the cathode, four main processes occur: initial corona formation, streamer-to-leader transition, leader development, and the final jump [1,2,3]. In general, as the voltage applied to the gap increases, the leader propagates forward continuously, with minute changes in current and velocity, until the gap is broken down. However, the leader often demonstrates noticeable discontinuous development with a relatively long front time of the applied voltage or at a high ambient humidity, and the leader channel brightens and elongates abruptly with a powerful glaring corona streamer burst at its tip [4,5], associated with a sharp current pulse. This phenomenon is often referred to as restrike or re-illumination [3,6]. Discontinuous development of the positive leader occurs both in the discharge process of natural lightning and in long-air-gap discharges under laboratory conditions [7,8].

Although long-air-gap discharges have been studied for a number of years, few researchers have focused on restrike, with its physical mechanisms difficult to explain. In the field of electrical engineering, an air gap is usually used as the main external insulation method for UHV transmission and transformation systems, and the front time of switching overvoltage generated during operation can reach more than 1 ms [9]. When overvoltage accumulates, the flashover of air gaps with a gap distance of several meters is often accompanied by restrike, which, in turn, affects the flashover characteristics of air gaps in UHV transmission and transformation systems [10,11]. It is therefore of significant value in engineering research to clarify the physical process and mechanism of positive restrike in a long-air-gap discharge or to improve existing positive leader development models and refine the design of external insulation in ultra-high-voltage transmission projects.

Since the 1970s, scholars around the world have conducted numerous theoretical or experimental studies on positive leader restrike. The Les Renardières Group team in France found that the probability of restrike occurring during discharge significantly increased as the humidity exceeded 8–10 g/m³ in a laboratory environment [3]. A restrike phenomenon often occurs with sudden elongation of the leading channel, a sudden increase in the discharge current, and sudden enhancement of channel luminescence. Baldo and Rea classified the restrike phenomenon into two types, “bell shaped” and “steep rise”, based on the similarity between the channel discharge current waveform and the channel emission pulse that takes place when restrike occurs [12,13]. In 1986 [14], the Les Renardières Group used a schlieren system and conducted a “simulated” experimental study on the restrike phenomenon by applying positive and negative switching-impulse voltages of opposite polarity to a 6 m rod–plate air gap at different time points. Dissipation and activation of the leader channel during discharge relaxation were observed, and the experimental results indicated that there was no luminescence in the leader channel during the discharge relaxation stage. Furthermore, during the discharge relaxation stage, the original channel was reactivated as the electrode applied a negative voltage impulse. This phenomenon is very similar to the restrike observed in long-air-gap discharge; as a result, this experiment was also referred to as “artificial restrike” by the authors of Ref. [15]. In Ref. [15], Domens et al. observed restrike by applying a positive impulse to a 16.7 m rod–plate air gap; they found that the discharge current always disappeared for a certain period of time before the discharge activity resumed. This finding is consistent with the observation results of the “steep”-type restrike phenomenon [3].

In a similar vein, Domens et al. employed a high-speed schlieren system to observe the leader channel near the tip of the electrode with a gap distance of 2.2 m during the positive discharge process [16]. It was found that, before the restrike occurred and during the discharge relaxation process with a discharge current of I = 0, the insulation characteristics near the electrode tip were restored for a certain period of time, and the original leader channel was no longer connected to the electrode. In recent years, researchers have conducted in-depth experimental studies on the influence of humidity on the discontinuous development of leaders, the pulse characteristics of restrike current, the charge amount, the leader development speed, etc. [7,8,17,18]. Despite these advances, the restrike phenomenon and its physical process during the discontinuous development of leaders have been overlooked.

In the study presented in this paper, by constructing a synchronous observation experimental platform for long-air-gap discharge, a positive switching impulse with a wavefront time of 1 ms was applied to a 6 m rod–plate gap. The “steep”-type restrike during the discharge process was observed, and the luminescence phenomenon of the leader channel during the relaxation process before the occurrence of restrike was studied and analyzed. Lastly, based on the experimental results, the physical mechanism of the luminescence phenomenon generated by the leader channel during the relaxation process was preliminarily explored.

2. Experimental Set-Up and Conditions

The experimental set-up for the synchronous observation of long-air-gap discharge performed in this study is presented in Figure 1. In order to generate a restrike during the discharge process, it is necessary to apply a voltage impulse with a long front time to the air gap. Therefore, by precisely calculating the resistance and capacitance parameters within the synthetic circuit of a Marx generator and implementing optimized selection of wavefront resistors, we achieved reliable generation of a switching impulse voltage with a wavefront of 1 ms, which was then directly delivered to the rod–plate gap with a gap distance of 6 m. The anode terminal was a 10 cm long copper electrode with a hemispherical tip with a 1 cm radius, and the electrode itself was a cylindrical rod with a diameter of 2 cm. The plate electrode was composed of aluminum, with a size of 30 m × 30 m, and was effectively grounded. For safety reasons, during the experiment, the voltage and current waveforms, in addition to trigger signals, were all isolated and converted by electrical-to-optical converters (E/O) or optical-to-electrical converters (O/E) during transmission. Among these, the data obtained at high potential were transmitted through optical fibers, whereas those obtained at low potential were transmitted through signal cables.

It should be mentioned that the upper end of the rod electrode was a hollow cylinder, in which a transient current acquisition system was installed. The structure and principle of this acquisition system were consistent with those described in Ref. [19]. It could achieve discharge transient current acquisition with a peak current of 600 A, a bandwidth of 0–70 MHz, and a time resolution of up to 2 ns [20], and it could sensitively and accurately reflect the current characteristics during the discharge process. The output voltage impulse of the Marx generator was sent to an oscilloscope of the Tektronix DPO-4104B model (made in Beaverton, Oregon, USA) through a capacitor divider with a voltage division ratio of 1:5358. After capturing the switching-wave voltage pulse, the oscilloscope output a synchronous trigger signal [21], which was then sent to the current acquisition device and the CCD high-speed camera, separately, through an E/O, in order to achieve synchronous acquisition of the discharge voltage, current, and optical morphology of the leader channel during the discharge process.

In positive long-gap discharges, the leader channel propagates stably at approximately 1–2 cm/μs [1,3,5], with its development time ranging from hundreds of microseconds to milliseconds under varying gap configurations and applied voltages [17,18]. Capturing the optical morphology evolution of the leader channel necessitates a high-speed imaging system equipped with a μm-scale spatial resolution, a μs-level temporal resolution, and an ms-range recording capacity, coupled with high-sensitivity photosensitive characteristics. Through systematic evaluation, a Photron FASTCAM SA5 (12-bit CMOS, 20 μm pixel pitch, made by Photron Limited, Tokyo, Japan) high-speed camera was ultimately selected as the primary imaging device to capture the optical evolution of leader channels during discharge, which could operate at 1,000,000 fps with 1 μs exposure to resolve discharge transients. With an ISO sensitivity of 25,000 and 32 GB of onboard memory, this set-up enabled 1.5 s continuous capture of leader propagation dynamics, satisfying μs-scale temporal resolution requirements for leader channel characterization. Due to the short duration of the discharge development and the relatively long distance between the high-speed camera and the rod–plate gap, in order to simultaneously take into account the spatio-temporal resolution of the high-speed optical photographs, the camera’s shooting speed was set to 200,000 frames per second (fps), with an exposure time of 4 μs per frame and a spatial resolution of 152 pixels × 256 pixels. As the luminescence of the leader channel was weak, in order to avoid the interference of external light and capture the discharge development process of the leader channel more clearly, the experiment was conducted at night. The ambient temperature during the experiment was 17 °C to 23 °C, the relative humidity was 62% to 78%, and the atmospheric pressure was 1 atm.

3. Experimental Results

3.1. Gap Withstand

The typical experimental observation results when the gap withstood the voltage are shown in Figure 2. Specifically, Figure 2a illustrates the waveform of discharge voltage and discharge current, whereas Figure 2b presents partial high-speed optical photographs capturing the initiation and development process of the leader. In order to highlight the characteristics of important electrical parameters, only the first 800 µs of the voltage and current waveforms are presented in Figure 2a. When the switching voltage was applied to the gap, the positive electrode generated an initial corona discharge at t = 19.71 µs, with a pulse amplitude of 2.45 A. Subsequently, under the suppression effect of the positive space charge in the streamer region near the electrode tip, the discharge ceased. After a dark period of approximately 23 µs, the leader initiated at t = 42.35 µs and continuously propagated toward the middle of the gap. However, the continuous-development current of the leader at this time was relatively low, approximately 0.1–0.2 A. Starting from t = 136.65 µs, the leader began to enter the discontinuous development stage. Thereafter, seven typical restrikes occurred.

Among the restrikes that occurred, the current pulse amplitudes of the two restrikes at t = 373.74 µs and t = 499.33 µs reached 8.32 A and 11.6 A, respectively. Thereafter, before the front time of the voltage wave ended, the development of the leader channel had already ceased, and the gap withstood the voltage.

As can be seen from the current waveform presented in Figure 2, all of the restrikes appearing during the discharge development process were of the “steep rise” type. During the discharge relaxation stage between every two restrike current pulses, the discharge current was ultimately zero, and the duration of the relaxation process varied. Among the relaxation stages, the relaxation time between the fifth and sixth restrike pulses was the longest, at 71 µs and 100 µs, respectively, and the corresponding current pulse amplitudes were also the highest.

In order to clearly observe the current characteristics and the luminous characteristics of the leader channel before and after the occurrence of the restrike, photographs captured by the high-speed camera and current waveform corresponding to the fifth (t = 373.74 µs) and sixth (t = 499.33 µs) restrike presented in Figure 2 were extracted, as illustrated in Figure 3. The six high-speed photographs taken before and after the occurrence of these two restrikes were labeled F1–F6 and F7–F12, respectively, with the start and end moments of the exposure of each frame corresponding to the time axis of the discharge current waveform with dashed lines.

Due to the limited resolution of the high-speed photographs presented in Figure 3, in order to highlight the details of the optical characteristics of the leader channel before the occurrence of restrike, the high-speed optical photographs in Figure 3 were processed with the image pseudo-color algorithm [22]. In more detail, as the grayscale value of the image increases, its color gradually changes from blue to red, and it turns white when the grayscale value is extremely high. In addition, the area where the leader channel is located in the photograph is marked with a white dotted frame, and the area within the white dotted frame was magnified three times and is placed below the photograph.

As can be seen from the current waveform of the discharge shown in Figure 3, within 25–30 µs before the arrival of the restrike pulse, the leader discharge current was almost zero, indicating that the leader channel had stopped developing and had entered the discharge relaxation stage, and the ionization activity of the streamer region at its head had ceased. At the two moments of t = 373.90 µs and t = 499.87 µs, the restrike phenomenon occurred, and the current of the discharge channel suddenly increased. However, within 20 µs, the continuous current of the leader decayed to zero, and the discharge entered the relaxation stage again.

By analyzing the four photographs taken before the two restrike current pulses shown in Figure 3, it can be concluded that during the process of gradually approaching the restrike moment, the leader channels shown in all four photographs exhibited luminescence phenomena, which were constantly intensifying, making the shape of the original channels increasingly obvious. In particular, the leader channel on the right side of the branch in F5 before the restrike in Figure 3a, and the leader channel on the left side of the branch in F11 in Figure 3b, were particularly luminous.

However, the discharge current corresponding to the exposure time of these two photographs can be ignored. When the observation time was advanced by another 5 µs, it can be clearly seen in photographs F4 and F10 that there was obvious luminescence at the channel head, whereas the luminescence in the middle and at the upper ends of the channel was relatively weak or even non-existent. Moreover, the corresponding current during the exposure time of these two images was still zero. The current waveform and high-speed optical photographs shown in Figure 3 indicate that there may have been a discharge phenomenon at the head of the leader channel during the discharge relaxation process before restrike occurred.

3.2. Gap Breakdown

The typical experimental results of gap breakdown are shown in Figure 4. In this discharge, the initial corona started at t = 20.29 µs. After developing for a certain period of time, it stopped developing when the first restrike occurred at t = 140.67 µs. After six restrikes with larger current pulse amplitudes, the leader channel continued to steadily develop until the gap broke down. Among the seven restrikes, the current amplitudes of the fourth and seventh restrikes were significantly larger than those of the five other restrikes, with peak values reaching 6.05 A and 12.36 A, respectively.

The current pulses around the moments when the fourth (t = 290.37 µs) and the seventh (t = 519.89 µs) restrike events occurred were selected, in addition to the corresponding seven frames of high-speed optical photographs. Each photograph frame is labeled F1′–F7′ and F8′–F14′, respectively, as shown in Figure 5a,b. During the relaxation stage before the restrike pulse, as shown in Figure 5a, the luminosity at the head of the leader channel was relatively weak. It gradually became clear, starting from photograph F3′ (the unclear optical morphology in F5′ is due to the occurrence of restrike at the upper end of the electrode, and the intense luminosity caused the camera to automatically reduce the ISO sensitivity). In addition, the current formed by the discharge of the shielding cylinder on the upper part of the electrode did not enter the internal current sensor; there was therefore no change in the current on the current waveform.

However, during the discharge relaxation stage shown in Figure 5b, the luminosity at the head of the leader channel was abundantly evident. Examining the currents corresponding to the photographs, excluding a very small discharge current (I < 0.1 A) near the end of the exposure time of photograph F12′, the discharge currents corresponding to the first four frames of the photographs were all zero. The above phenomena are ultimately consistent with those shown in Figure 3a,b.

4. Discussion

4.1. Relationship Between Restrike Current Pulse Peak and Relaxation Time

In long-air-gap discharges, when the rate of rise in the electric field applied to the HV electrode is lower than the rate of rise in the space charge electric field generated by the ionization in the streamer region near the leader tip, the activity level of the streamer region will significantly weaken or prevent ionization. In addition, when the ambient humidity is relatively high, the ionization efficiency of the streamer region also decreases significantly, due to the strong affinity of water molecules for electrons. In the experiments described herein, a switching pulse with a wavefront of 1 ms was applied to the gap, and the corresponding voltage rise rate was relatively low. Coupled with the relatively high ambient humidity, the frequency of restrike during the discharge process was high. When the streamer region at the leader head gradually weakens until ionization stops, it can no longer continue to inject electrons into the leader channel. In addition, the high-temperature gas molecules in the channel continuously undergo convection with the gas molecules in the surrounding environment, causing the temperature of the channel to continuously decrease. This process additionally leads to a decrease in or the cessation of thermal dissociation within the channel, resulting in the discharge current of the leader dropping to zero and entering the relaxation stage.

Statistical analysis of the experimental observation results showed that the luminous phenomenon at the leader head could not be clearly seen on all occasions in the relaxation process. A total of 18 discharges were carried out in this experiment, and the number of restrikes in each discharge varied from one to seven times. In total, about 74 relatively obvious restrike current pulses were generated 18 times during discharge. Based on the photographs of the discharge process taken by the high-speed camera, 34 discharges exhibited luminescence at the head of the leader channel during the relaxation process before the occurrence of restrike. The relationship between the peak values of these 34 restrike current pulses and the duration of the relaxation phase before the occurrence of restrike is shown in Figure 6.

It can be seen from Figure 6 that, due to the high dispersion of long-gap discharges, when there is luminescence at the head of the leader channel during the relaxation process, there is no obviously linear relationship between the corresponding relaxation time and the peak value of the restrike current pulse that follows immediately. However, when examining the overall trend of change, it can be concluded that as the relaxation time increases, the peak value of the restrike current pulse shows a relatively obvious increasing trend. Under the experimental conditions described in this paper, when the peak value of the current pulse exceeds 5 A and the duration of the relaxation stage is greater than 40–50 µs, the frequency of the luminescence occurring at the head of the leader channel before the restrike is relatively high. Based on the corresponding high-speed optical photographs, when the relaxation time is relatively short (<40 µs), the luminescence at the head of the leader channel is not obvious. Relatively speaking, when the relaxation time is longer, the luminescence at the head of the channel is more obvious, and the luminous intensity increases significantly with the passage of the relaxation time, with the peak value of the restrike current pulse also being larger.

4.2. Luminescence of Leader Channel During Relaxation Process

Based on the results of previous long-spark studies, during the relaxation process of the leader channel before the occurrence of restrike, there is no obvious ionization activity at the head of the channel, nor does the channel show obvious luminous phenomena [3]. However, the high-speed photographs shown in Figure 3 and Figure 5 and the corresponding current waveforms indicate that during the relaxation process before the occurrence of restrike, although the discharge current measured at the high-voltage electrode is zero, the leader channel exhibits a certain degree of weak luminescence, that is, “the dark period is not actually dark”, which indicates that, during the relaxation process before the occurrence of restrike, ionization activities may take place in the leader channel, and a large number of photons are emitted simultaneously. Moreover, as the relaxation process progresses, the number of photons gradually increases toward the moment of restrike occurrence. However, due to the restrictions of the limited time resolution of the high-speed camera, within a few microseconds before the leader restrike occurred, it was impossible to conduct a more detailed observation of this discharge development process, even though there might still have been other forms of ionization in the channel.

The widely recognized critical temperature for streamer-to-leader transition is 1500–2000 K [3,23], which is the critical temperature for the dissociation of a large number of negative ions inside the streamer stem. During the continuous development of the leader channel, its axial temperature can reach more than several thousand Kelvins [3,24,25], which is much higher than the critical temperature of 1500–2000 K for negative ion dissociation. Under the effect of the high temperature in this channel, the negative ions within the leader channel are dissociated into neutral particles and free electrons, making the density of free electrons in the leader channel far greater than that of negative ions [4,24,25]. In light of the above results, we suggest that during the continuous development of the leader, the charged particle components inside the leader channel are primarily free electrons and positive ions, and the density of negative ions can ultimately be ignored; this finding differs considerably from that presented by the authors of Ref. [8], who contend that in addition to positive ions, there are also negative ions and electrons inside the channel. A schematic diagram of the charged particle components of the channel during this stage is shown in Figure 7a. Under certain external conditions, when the streamer region at the leader head stops ionizing and the discharge enters the relaxation stage, the number of electrons injected into the channel decreases sharply, causing the conductivity of the channel to drop rapidly. In addition, the channel voltage drops and the electric field inside the channel may increase rapidly. The schematic diagram of this instantaneous process is shown in Figure 7b. Subsequently, under the action of the increasing electric field within the channel, the residual electrons migrate into the anode rapidly, resulting in the charged particle components within the channel primarily being positive ions. Therefore, the leader channel at this time may not be a true plasma channel.

Under the effect of the externally applied electric field in the gap, the positive ions within the leader channel migrate toward the channel head along the direction of the electric field and collide with the high-temperature neutral molecules inside the channel, as illustrated in Figure 7c. Through the momentum exchange during collisions, the high-temperature neutral molecules in the channel gradually move away from the root of the channel along with the positive ions, resulting in a continuous reduction in the number of high-temperature gas molecules in the discharge channel originally connected to the electrode tip. With the passage of time, the gas temperature near the electrode gradually decreases to match that of the environment, and the gas density of the channel also increases to match that of the surrounding gas, resulting in the original leader channel appearing “disconnected” from the high-voltage electrode [16], as shown in Figure 7d, where the black dashed line represents the bottom edge of the leader channel and the blue dashed line represents the edge of the channel after being “disconnected” from the electrode.

Due to the high gas temperature and low density within the channel, the mobility of positive ions in the channel is greater than that at ambient temperature; therefore, these ions can migrate toward and accumulate in the channel head more quickly under the external electric field. According to the schlieren observation results and the simulation analysis of the leader channel during the relaxation process presented in Ref. [26], it can be concluded that as the positive ions migrate toward and accumulate in the leader head, the leader head additionally extends forward, and the radius range of the head region also increases further, as shown by the blue dashed line at the leader head in Figure 7d; this finding also differs from the schematic diagram shown in Figure 6 of Reference [8]. Subsequently, the Poisson field formed by the aggregated positive ions superimposes with the externally applied electric field in the gap, causing the distortion and enhancement of the local electric field in the area near the head of the original discharge channel. While this superimposed electric field reaches the electric field threshold of 26–30 kV/m in air [2,3], the streamer region near the channel tip reactivates and emits photons. The schematic diagram is shown in Figure 7e. Under the action of the electric fields both inside and outside the channel, the electrons generated in the streamer region at the channel head directly inject into the channel and then quickly recombine with the positive ions in the channel and release a large number of photons, causing the channel head to exhibit the channel luminescence phenomenon during the relaxation process before restrike, as shown in Figure 3 and Figure 5.

Because of the high density of positive ions inside the channel, the electrons entering the channel through streamer discharge neutralize directly with the positive ions or attach to neutral particles to form negative ions, and then recombine with the positive ions during their migration toward the high-voltage electrode. Ultimately, they cannot directly enter the high-voltage electrode from the inside of the channel. Consequently, as shown in Figure 3 and Figure 5, before the occurrence of restrike, although there was discharge and luminescence at the channel head, the current measured at the high-voltage electrode was ultimately zero.

As the externally applied electric field increases further, the local electric field in the head region of the leader channel increases accordingly, and the streamer discharges generated thereby become increasingly intense. It can therefore be seen from Figure 3 that compared with F4, F5 shows a larger luminous range and more obvious luminous intensity formed by the streamer discharge at the channel head, and the same finding is true for F11 compared with F10 or F9. Similarly, as shown in Figure 5, the discharge and luminescence at the channel head are also the same for F12′ compared with F11′.

Although the Poisson field formed by the positive ions in the channel have a certain inhibitory effect on the external electric field generated at the high-voltage electrode, with the continuous increase in the applied voltage and the continuous migration of the positive ions toward the channel head, the electric field at the tip of the electrode continuously increases in size. Once it reaches the breakdown electric field threshold of 26–30 kV/m, new streamer discharges restart at the electrode, as shown in Figure 7f. Thereafter, the leader discharge formed by the streamer discharge rapidly propagates forward along the thermal imprint of the original channel, and quickly extends to the channel tip at a speed of about 10⁸ cm/s [3], resulting in the occurrence of restrike and reconnection to the electrode of the channel. The corresponding schematic diagram is shown in Figure 7g. During this stage, the discharge current exhibits a sudden, drastic increase, forming a restrike current pulse with a peak value of several amperes. The leader channel reconnects to the high-voltage electrode, and the entire channel emits intense light, causing the length of the channel to increase significantly, as shown in F6 and F12 in Figure 3, and F7′ and F14′ in Figure 5.

5. Conclusions

In the present study, we experimentally observed the restrike process during discharge by applying a positive switching impulse with a wavefront of 1 ms to a 6 m rod–plate gap. The observation results of the discharge current waveform and high-speed optical photographs showed that during the relaxation process with zero current before the occurrence of restrike, when the relaxation time was relatively long, there were discharges and luminescence at the head of the leader channel. Through the discussion of the charged particle components within the leader channel during the relaxation stage, we concluded that this phenomenon is due to the continuous existence and increase in the external electric field during the relaxation process. As a result, the positive ions within the leader channel migrate toward and concentrate in the channel head along the direction of the external field, and, together with the external electric field, distort and enhance the local electric field at the channel head, exceeding the critical value of the electric field for streamer discharge in the air. Therefore, streamer discharges generate and release a large number of photons at the channel head. The observation and discussion of this phenomenon may provide a reference for further research on the physical process and mechanism of positive restrike.