Winter lightning in Japan is known for such characteristics as frequent occurrence of upward lightning and of positive ground flashes. On the engineering side, higher frequencies of troubles at transmission lines or wind turbines in winter due to lightning than those in summer have been experienced in the winter thunderstorm area of Japan, despite the much smaller number of lightning strokes in winter observed by lightning location systems (LLS). Such frequent troubles by lightning in the cold season are unique in Japan, which have promoted intensive research on winter lightning in Japan since 1980s.
The demand for both data quality and the range of Cloud-to-Ground (CG) lightning parameters is highest for forensic applications within the electric utility industry. For years, the research and operational communities within this industry in Japan have pointed out a limitation of these LLS networks in the detection and location of damaging (high-current and/or large charge transfer) lightning flashes during the winter months (so-called “Winter Lightning”). Most of these flashes appear to be upward-connecting discharges, frequently referred to as “Ground-to-Cloud” (GC) flashes. The basic architecture and design of Vaisala’s new LS700x lightning sensor was developed in-part to improve detection of these unusual and complex flashes. This paper presents our progress-to-date on this effort. We include a review of the winter lightning detection problem, an overview of the LS700x architecture, a discussion of how this architecture was exploited to evaluate and improve performance for winter lightning, and a presentation of results-to-date on performance improvement. A comparison of GC detection performance between Tohoku’s operational 9-sensor IMPACT (ALDF 141-T) LLS and its 6-sensor LS700x research network indicates roughly a factor-of-two improvement for this class of discharges, with an overall detection of 23/24 (96%) of GC flashes.
Seasonal variations of number of high current lightning discharges exceeding 100kA observed by JLDN (Japanese Lightning Detection Network) were analyzed. The months with averaged altitudes of -10°C level higher than 5.7km are classified as ordinary summer from the viewpoint of lightning activity. Meanwhile, on the coast of the Sea of Japan, more than 90% of negative high current lightning discharges were -GC (Ground to Cloud) strokes in the months when monthly averaged altitudes of -10°C level are lower than 2.7km. These months are classified as the winter lightning season when upward lightning flashes frequently occur. Months other than winter or summer are classified as spring or autumn. In these seasons, the proportions of positive high current lightning discharges are higher than those of negative discharges like winter. Thus, the charge structure in the thunderclouds of spring and autumn may be similar to that in winter, and high current lightning strokes tend to occur. Since this variation of seasons is different in each area of Japan, relation of seasons, areas, and densities of high current lightning discharges were analyzed.
In order to further understand the initiating mechanism of positive cloud-to-ground (PCG) flashes, the characteristics of initial breakdown pulse trains are investigated by using electric field data of Guangdong province. Considering the difference in initial polarity between the initial breakdown pulses and the first return stoke (RS), three types of bipolar pulse trains, including the same polarity (type I), the opposite polarity (type II) and the composite polarity (type III), are identified. The percentages of type I, type II and type III are about 84%, 10% and 6%, respectively. The statistical results show that PCG flashes including initial breakdown pulse trains, which occur frequently when PCG flashes are not active, are about 12% of all the PCG flashes. The width of individual pulse (Tpw) and the interval of adjacent pulses (Tpi) are about 25μs and 271μs, respectively. The mean values of pulse train duration (Td) and the interval (Tp-ri) between initial breakdown pulse train and RS are 8.88ms and 98.48ms, respectively. The average value of the ratio (Rp-rr) between the maximum amplitude of pulse train and the amplitude of the first stroke pulse is 0.16. The ratio is smaller in Guangzhou province than in Japan and in Brazil. Comparing of the parameters of type I and type II, it can be found that Tpw, Td and Tpi of type II are less than those of type I, however, the values of Tp-ri and Rp-rr of type II are larger than those of type I. The possible causes of the observating several types of pulse trains and the significantly diversified pulse characteristics are discussed.
In this paper, the characteristics of the positive lightning flashes in the coastal area of the Sea of Japan and the lightning stroke model for the estimation of an upward lightning current waveform were described based on observation results. The positive lightning flashes initiated upward from a radio tower were observed in winter 2001. The current peak value and the electric charge of the lightning flashes were much larger than those of summer lightning flashes. Slow front and fast transition, which are common characteristics of the electric field waveform associated with all the return strokes, were not recognized clearly at the rising portion of the lightning waveform. The calculated waveform using the lightning stroke model for the upward lightning, in which the current wave propagates downward from the cloud and reflects at the ground, agreed well with the observed waveform.
Surge arresters and distribution equipments with zinc-oxide elements are used for lightning protection of overhead power distribution lines in Japan. However, these surge arresters are sometimes damaged by direct lightning strokes, especially in winter. Increasing of surge arrester failures in winter is attributed to a very large electric charge of winter lightning than that of summer lightning. For improvement of surge arresters, we have measured the energy absorption capability of surge arresters using a half cycle of alternating current with a frequency of 50Hz for simulating a winter lightning current. The mean values of arrester failure energy increased in proportion to the volume of zinc-oxide element, however the values of arrester failure energy were quite uneven. We also have observed the aspects of damaged zinc-oxide elements, and have investigated the relationship between the arrester failure energy and the failure types of zinc-oxide elements. From these results, we suggest the improvement of the energy absorption capability of distribution surge arresters, especially for the uniform energy absorption capability.
The characteristics of upward leaders were observed at a 200m chimney in the coastal area of the Sea of Japan from 1989 to 2008 for clarification of the initiation processes of upward lightning using a high-speed video camera and the ALPS(Automatic Lightning Progressing feature observation System). In the observation, we obtained images showing the inception processes of both positive and negative upward leaders with the corresponding current waveforms. Positive upward leaders propagate continuously and negative upward leaders propagate intermittently. We evaluated the line charge densities of the upward leaders, which were found to be much larger than those used in the lightning shielding model for the lightning protection design.
We have observed lightning that struck a wind turbine and its neighboring lightning-protection tower during the past six winter seasons (2005 to 2010) using various lightning observation instruments. Our results show that the upward lightning from high structures can be classified into self-initiated and other-triggered types according to whether there is a discharge activity prior to the upward lightning. Furthermore, we found that although other-triggered upward lightning can start at a relatively low wind speed, self-initiated upward lightning always started either from the stationary tower under a larger wind speeds or from a rotating wind turbine blade. It appears that the wind does have considerable effect in assisting the initiation of an upward leader. In addition, we found that the self-initiated upward positive leaders from structures with different effective heights exhibited remarkably different initial speeds. Higher structures tend to initiate faster upward leaders. Finally, we discussed the pulse discharges observed in the very initial stages of positive upward leaders and how to protect structures from upward lightning as well.
In this paper, an overview of recent theoretical investigations and experimental characterization of the electromagnetic fields in the vicinity of a tower struck by lightning is presented. The presence of an elevated strike object has been included in two classes of return-stroke models, namely the engineering models and the electromagnetic or Antenna-Theory (AT) models. Extensions of engineering models based on distributed-source and lumped-source representations are described. The adequacy of the representation of the tall structure as an ideal uniform transmission line is discussed. Different approaches and assumptions for the computation of electromagnetic fields are summarized. Finally, recent progress in the characterization of electromagnetic fields radiated by lightning to tall structures is presented.
The performance of the IMPACT sensor network in the Tohoku region for winter lightning detection was evaluated referring to another study of winter lightning in the coastal area of the Sea of Japan. The detection efficiency for lightning flashes causing transmission-line faults was 62%, but decreased to 56% for the first lightning strokes. Classifying these first lightning strokes, the detection efficiencies were 23% for -GC strokes, and 67% for +GC strokes. GC strokes accounted for 73% of the entire group of investigated strokes, and rejected GC strokes also accounted for 95% of all rejected strokes. In analyzing the LEMP waveforms, it was found that the rejection was caused by an incompatibility of the characteristics of the LEMP waveform with the waveform discrimination criteria, and the influence of the field pulses prior to the main pulse on the signal processing in the IMPACT sensors, which could be reduced through the propagation effect.
In this paper, the authors propose a new identification method of the magnetizing inrush current phenomena. In general, the identification is done using with current waveform. However, the saturation of current transformer can't give waveform. Therefore, the authors introduce the identification method using with voltage waveform, in which the saturation of voltage transformer doesn't happen. And then, applying the Aitken's Δ2-process, it is showed that the new identification method gives the exact saturation on/off time.
This paper proposes a new estimation method to determine any circuit parameters for the harmonic analysis of distribution network. The harmonic analysis is important to maintain quality of electric power. An interesting phenomenon, which is the cut-down of harmonic voltage level temporarily at dawn even if demand is a little bit increasing at this time, is often observed in the actual power system. This paper proposes utilizing the observation results of this phenomenon to estimate circuit parameters. In addition, this paper claim that detail equivalent circuit is more preferable for explaining that phenomenon than conventional simplified equivalent circuit. This paper also discusses the usefulness of the Harmonic Characteristics Curve, which is the proposed by this paper as new concept, at secondary winding of distribution transformer and the correlation of PQ, which are monitored values of active and reactive power at secondary winding of distribution transformer. This correlation should determine the equivalent fundamental frequency admittance of among of the shunt capacitors installed concerning distribution lines.
In this paper, representations of a perfectly conducting thin wire and an imperfectly conducting medium in the transmission-line-modeling (TLM) calculation are briefly explained. Then, the method is applied to analyzing surge responses of a vertical parallelepiped grounding electrode and a square-loop grounding electrode. Surge responses calculated using the TLM method agree reasonably well with the corresponding responses measured and calculated using the finite-difference time-domain method. It is probably the first time that surge responses of grounding electrodes have been analyzed reasonably accurately using the TLM method.