progress in protecting power systems against impulse ... - CiteSeerX

0 downloads 0 Views 2MB Size Report
Nov 30, 2007 - The I2t duty from severe lightning is considerably smaller than the power ...... [18] B.A. Zajac and S.A. Rutledge, “Cloud-to-Ground Lightning ...
IX International Symposium on Lightning Protection 26th-30th November 2007 – Foz do Iguaçu, Brazil

PROGRESS IN PROTECTING POWER SYSTEMS AGAINST IMPULSE CHARGE AND CONTINUING CURRENT EFFECTS OF LIGHTNING FLASHES William A. Chisholm Jody P. Levine Craig J. Pon Marco A.R. Jusevicius Kinectrics / University of Hydro-One Kinectrics Insituto Tecnológico Quebec at Chicoutimi SIMEPAR [email protected] [email protected] [email protected] [email protected] Université de Québec à Chicoutimi, 555 boul. de l’Université, Chicoutimi, Québec, Canada G7H 2B1 Abstract - This document summarizes the international developments of appropriate test methods, standards and risk models for the damage of lightning charge to the metallic components and metal-oxide surge arresters of high voltage electric power systems and focuses some attention on how these parameters may be estimated in southeast South America.

1 INTRODUCTION The normal function of an overhead groundwire (OHGW) is to intercept lightning flashes that would ordinarily terminate on a phase conductor. By providing a local ground electrode, typically the metal tower base or vertical rod in the soil, the local potential rise at the top of the tower is limited to the product of the peak current times the footing resistance. In the majority of cases when OHGW are installed, they perform their interception and protection functions with high efficiency. Selection of OHGW size to meet the expected peak lightning impulse current duty has been traditional. The highamplitude impulse currents of 200 kA do not cause much damage to the aluminum or coated steel wires normally used in OPGW. OHGW are also specified to carry power system fault current, of perhaps 10 kA with up to thirty fault cycles, and tested to ensure that the resistive temperature rise associated with I2t duty does not lead to fusing or “birdcaging” damage from thermal expansion. The I2t duty from severe lightning is considerably smaller than the power system fault current duty, and coordination has usually been automatic. Designs that meet typical fault current requirements for transmission lines also satisfy severe I2t requirements from lightning flashes. Disappointing performance of some optical-fibre ground wire (OPGW) systems in several countries, notably Japan, China and Brazil, and problems in the USA and New Zealand led to re-evaluation of the testing requirements and specification process for these new power system components. Improved and more representative test methods have been proposed, developed and adopted. Most OPGW damage from lightning is now recognized to occur from charge transfer, especially associated with the continuing-current component of up to 0.5 s duration, at relatively low currents of 50 to 400 A. Fundamental work by [1] and others established the relation between holes in metal sheets used in aircraft construction and stroke charge in the range of 10 to 200 Coulomb. Several researchers have carried out tests and reported service experience with OPGW damage in this same range of charge. Application criteria for wire materials and sizes have evolved. Parameters for the statistical distribution of positive and negative charge have been recommended. There has also been convergence towards international standardization of the lightning test for OPGW, considering activity within Brazil (ABNT) [2] and at IEC [3-5] and IEEE [6]. There remain some areas of uncertainty and interest in the regional variation of charge parameters. For example, areas of Japan are subject to winter lightning having a high fraction of positive flashes with large peak amplitude and charge. OPGW damage is more common in these areas. Conversely, in spite of its high ground flash density, there are only limited reports of OPGW damage in Florida, USA. Reports of lightning location systems, tracking the density of positive flashes with high peak currents, suggest an explanation. A region of high density of large positive flashes

occurs in an unexpected location, centered in the central US and Canada. Data for Brazil show a high density of large positive flashes on its western border with Argentina and Paraguay. This suggests that testing standards for adverse effects from the total flash charge associated with strokes of large peak current need to consider both positive and negative polarities, taking into account the local fraction of positive events, which varies widely by region and season. The issue of direct damage to conductors takes more prominence as compact transmission line designs use surge arresters at every tower (TLSA) rather than conventional overhead groundwires. In addition, the lightning charge rather than the peak current establishes the energy duty of the TLSA. Issues relating to the calculation of energy duty stress for repetitive surges are described. The multi-pulse duty has a particular influence on the externally gapped applications where fewer TLSA appear in parallel. The effect of charge duty on the false operation of line surge arrester accessories such as explosive disconnect devices is also a potential problem. The possibility of adapting OPGW test standards to apply as well to conductors used in unshielded transmission line applications is explored. Some conductors have already been tested with charge levels up to 200 Coulomb. Specific recommendations for integration and for future application are then developed. 2 CHARGE PARAMETERS IN THE LIGHTNING FLASH 2.1 The lightning flash Lightning is a consequence of the separation of positive and negative charge regions in clouds, above and below an area where the temperature is about -15°C. The charging mechanisms themselves are of considerable scientific interest, with hypotheses related to friction and collision among water and ice carrying the most weight. Lightning occurs when the electric field intensity between charged areas reaches a critical breakdown level. The most common form of lightning is a cloud flash. In cases where one of the lower charge pockets is close to the ground, the earth can become the termination point of a ground flash. Three types of weather cause most thunderstorms: •

Large Frontal Storms: these are formed by collision between an incoming cold air mass an existing warmair mass. These thunderstorms are dynamic and often associated with other severe weather, such as intense rain and high winds. In North America, these tend to form in the center of the continent and move to the northeast. Local conditions affect the direction of frontal storm movement: for example, in Italy, the large frontal thunderstorms come from the north-west and move in a south-east direction. While they get stronger and weaker during the day and night respectively, frontal storms move hundreds of km over their lifetime and can generally pass a specific region at any time. Winter lightning typically occurs during frontal storms.



Local Convective Cells develop during afternoon or evening, often when the dew point temperature exceeds about 22-25°C and other conditions also favor vertical instability. This type of thunderstorm is more typical in tropical regions or in summer in temperate climates. Vertical instability is needed to drive the charge separation process. This instability is achieved over relatively small land areas, leading to a strong land-water contrast in lightning flash density for Cuba, Java, Jamaica, Hawaii and other islands.



Orographic Thunderstorms are caused by the uplift of an air mass as wind blows uphill. For example, prevailing winds from the west are lifted by the Rocky Mountains. This depresses lightning activity west of the Cascades mountain range and increasing it to the east in Alberta, Canada and the Central Plains of the USA. Areas of high lightning density found in Columbia, Saudi Arabia, the Alps and the Pyrenees are other examples of strong orographic thunderstorm effects.

Variations in flash polarity and intensity are usually estimated with remote measurements of radiated electromagnetic fields, with ground-based networks of lightning location systems. Co-ordination of multiple local electric field measurements near lightning with radiated signals [7][8] can allow researchers to build up a comprehensive hypothesis of the discharge process and, at the same time, provide important warning and alarm functions to sensitive installations.

2.2 The negative lightning discharge process The total lightning flash is a complex process, normally involving: •

Development of a vertical plasma channel, called a downward leader, from the cloud towards ground in steps, with potential in the range of 20-100 MV and reaching typical capacitance to ground of 50-100 nF at the late stages of development



Development a counter-leader from grounded object towards the downward leader, in a process that can be described using the physics of switching-surge flashover for voltage surges of about 2 ms



A powerful first return stroke, sweeping the typical 1 to 10-Coulomb charge from the downward leader into ground with a wave of zero potential with an upward velocity of about c/3, where c is the speed of light.



Formation of dart leaders and subsequent return strokes with considerably reduced charge (1 Coulomb) and peak current, but higher rate of current rise, as shown schematically in Fig. 1



Periods shown in Fig. 1 where the plasma channel remains electrically conductive for the entire 30-500 ms period between subsequent return strokes, leading to a continuing current that can transfer considerable charge



Periods shown in Fig. 1 where M-components, pulses with typically 2-ms duration and about 0.1 Coulomb charge transfer, flow up the plasma channel.

Fig. 1 – Schematic of charge transfer from subsequent lightning return strokes, continuing current and M-components [9,10]

There can be several subsequent return strokes and many M components in the negative lightning flash. The most detailed measurements of charge transfer have been obtained from studies that have inserted a conducting object into the air space between the charged regions, such as a grounded instrumented tower [11][12] or wire from rockettriggered lightning [9][10]. Examples of how complex the return stroke can be are given in Fig. 2 [12] for a tower instrumented with a 0.25 mΩ / 3.2 MHz series resistor and suitable record duration to observe slowly-changing continuing currents.

Continuing current with 166 Coulomb of charge

Sequence of 39 negative and positive impulses

Fig. 2 – Two examples of measured current using 0.25-mΩ shunt [12].

2.3 The positive lightning discharge process The cloud charging tends to force positive charge aloft, with negative charge at the bottom of the cloud. This means that most lightning flashes to earth have negative polarity. However, small pockets of positive charge near the base of the cloud have been observed to increase the fraction of ground flashes. A small fraction (2-5%) of positive flashes is typical in summer thunderstorms. In winter, this fraction rises, and the reasons are related to the fact that the upper, positive charge region is closer to the ground. A marked shift from “winter” to “summer” lightning, described by the monthly fraction of positive events, occurs in April or May in North America, with reversion in October and November. Fig. 3 shows a similar pattern in Spain and Portugal, measured with a lightning location system [13].

Fig. 3 – Percentage of positive flashes recorded in the Iberian penninsula [13]

Within the network coverage of land area in Fig. 3, the fraction of positive flashes is more-or-less constant. Part of tendency towards measuring a higher ratio of positive flashes outside the land-based network area (in the ocean in Fig. 3) is a function of the slower rise time of the positive strokes, which suffer less signal attenuation over a long propagation distance than first or subsequent negative strokes. Median amplitudes of positive flashes are also slightly higher than negative flashes. These biases should be constant in summer and winter. The development of positive and negative leaders in laboratory tests at the 10 to 100-m scale indicate significant differences in the speed of development and in the electric fields necessary to sustain leader growth [14][15] . There is thus no reason to believe that positive downward leader development from lightning would be similar to the wellstudied negative downward leader process. However, preliminary studies do suggest that the return stroke velocities of natural positive and negative flashes are equal [16], and for large gaps, the difference between positive and negative polarity flashover tends to diminish [15]. This has provided some motivation to use equal scaling for radiated signals of positive and negative observations of peak radiated magnetic field when inferring the source current.

2.4 Direct observations of the charge in negative and positive lightning Most studies on natural, tall-structure or rocket-triggered lightning have focused on capturing the high-frequency components and peak current with good fidelity using appropriate transducers and bandwidth. It is relatively difficult to measure the total charge in a lightning stroke or flash directly, since this involves the use of transducers with good low-frequency response and long recording times [12]. The low-frequency energy does not have a strong electromagnetic radiation, so it is not possible to infer total stroke or flash charge from wideband receivers, unless they are very close to the source and can respond to both quasistatic and radiated electric field components. In 2005, a review of the lightning parameters by an IEEE Task Force [17] concluded that the measured distributions of downward-lightning impulse charge [11] were still most appropriate for evaluation of overhead line reliability. This paper made the data easier to use by providing invertible approximations for the log-normal probability distributions, Pc(Q) of the form:

1 PC (Q ) = n 1 + (Q QMedian )

1n

⎛ 1 − PC (Q ) ⎞ ⎟⎟ Q = QMedian ⎜⎜ ⎝ PC (Q ) ⎠

(1).

Table 1: Recommended parameters for median charge QMedian and exponent n in Eqn. 1. [17] Parameter Impulse Charge, first negative return stroke Impulse Charge, any subsequent stroke Impulse Charge, positive stroke Total Flash Charge, negative flash Total Flash Charge, positive flash

For natural downward flashes n QMedian, Coulomb QMedian, Coulomb [IEEE] [Berger 1975] 5 2 5.2 1 2.2 1.4 19 1.34 16 7 1.7 7.5 85 2 80

For triggered lightning QMedian, Coulomb [Fischer 1993] 2.1

Fig. 4 shows the original data from [11] for total flash charge.

Fig. 4 – Total lightning flash charge for negative (curve 1) and positive (curve 3) polarity [11]

For the total positive flash charge, substitution of a probability Pc(Q)=0.05 (5%) in Eqn. (1) along with the corresponding values from Table 1 gives a positive charge level of 370 Coulomb. The corresponding value for probability Pc(Q)=0.05 and negative flash charge is 40 Coulomb.

2.4 Correlation of flash charge with other lightning discharge parameters Most researchers making direct measurements of lightning parameters have noted a significant correlation between the peak current of the first downward stroke and the impulse charge in that stroke. A lower, but still significant, correlation usually exists between the peak current of the first return stroke and the total charge in the flash. Table 2 shows that the correlation for positive flash peak current and impulse charge was fairly strong at 0.62. This is partly because positive flashes tend to have a single stroke. For negative lighting, considering that each of the subsequent strokes has an independent statistical distribution and also that there may or may not be a continuing current, the correlation between first peak current and total charge in the flash should be weaker. Simultaneous measurements of flash charge and peak current by [11] indicated that significant correlation of 0.77 exists between the negative first stroke impulse charge and peak amplitude. Equations in Table 2 give the median charge level, given the peak current, with for example the median 31-kA peak first return stroke current suggesting a median negative flash total charge of 7.5 Coulomb. The uncertainty in the estimate of charge, given the peak current, is relatively large. If the first stroke peak current is 31 kA, there is a 68% (±σ) chance that the charge is between e-0.86 and e0.86 times its derived median value, covering a range from 3 to 18 Coulomb. Table 2: Correlations of Impulse and Flash Charge to Peak Stroke Current Amplitude [17] Parameter Negative First Stroke Impulse Charge* Negative Subsequent Stroke Impulse Charge Negative Flash Total Charge Positive Flash Total Charge

Correlation Coefficient ρ

Median as function of Peak Stroke Current I in kA

Standard deviation σln(Q|I)

0.77

Q=0.032 Coulomb (I/1 kA)1.48

0.59

0.43

Q=0.11 Coulomb (I/1 kA)1.01

1.13

0.54

Q=0.19 Coulomb (I/1 kA)1.14

0.86

0.62

Q=15.5 Coulomb (I/1 kA)0.46

0.71

*Correcting misprinted value of 3.2 in [17] using I|QImpulse expression Considering this high level of uncertainty in the results, it is fair to say that using large-amplitude flashes as an indicator of flashes with high levels of charge is more appropriate for positive flashes than for negative flashes. Fig. 4 suggests that this is helpful in treating the flashes with the highest levels of charge transfer. With the improved quality of measurements of subsequent-stroke continuing current in tower or rocket triggered lightning, improved statistics on the distribution of total negative flash charge associated with continuing currents and M components are being developed. The preliminary indications such as Fig. 2 suggest that high levels of negative charge transfer, closer to those obtained for positive flashes, may be appropriate when selecting threat levels in testing standards. 3 DENSITY OF LARGE-AMPLITUDE LIGHTNING FLASHES 3.1 Remote observations of the lightning discharge process Lightning location systems have been deployed in many countries. They have delivered reliable long-term estimates of ground flash density, making use of the strong electromagnetic radiation from the return stroke. For ground flash density results, what matters mainly is that the system has sufficient dynamic range to capture the signals with adequate signal-to-noise level. The dynamic range requirements are fairly severe: • • • •

The lightning peak current can vary by two orders of magnitude, from 2 kA to 200 kA (40 dB) The return stroke velocity can vary by a factor of ten, from 0.05c to 0.5c (20 dB) where c is the speed of light For the ground wave of the radiated signal, lightning may be anywhere between 20 and 600 km from the receiver, and the signal attenuates inversely with distance (50 dB) High-frequency components of the ground wave of the signal may attenuate by a factor of three (10 dB) over lossy ground [20] or may reflect at impedance discontinuities such as the sea-land interface

Considered as independent requirements, the receivers need about 120 dB of dynamic range in a sparse network. Configurations with short baseline distances and high sensor density can relax the most difficult requirements, related to distance from the flash to the sensor. Wide-area networks such as the NLDN [18] and NALDN [19] have given uniform coverage over extensive interior areas for many years, with degrading detection efficiency only over the oceans. 3.2 Scaling of remote radiated fields to estimate peak current and charge Charge and peak stroke current are moderately correlated, as discussed in Section 2. There are only limited possibilities to measure electrostatic fields to refine estimates of charge, since these signals attenuate as the cube of distance from the lightning flash. This leaves us with the need to rely on remote radiated field strength as the best available indicator of the amount of charge in a flash. While discussing scaling of radiated signals to estimate large-amplitude peak currents, it is important to note ongoing discussion about the possible relations of return stroke velocity with leader potential, charge or peak amplitude. Lightning location systems measure a signal that is the product of surge current and return stroke velocity v. This means that, for example, a 12 kA current rising to crest in 1 μs in a 300-m vertical tower should appear to the system as a 36 kA current in remote radiated field measurements. The entire rising edge of the signal is propagating within the tower (at the speed of light c) rather than up the return stroke channel (at the assumed velocity of v=c/3). This is an oversimplification, and for example, Rachidi [22] nicely summarized the main issues to resolve in the remote observation process when a tall tower forms part of the lightning return stroke path. However, the main point is that, if return stroke velocity increases with leader potential, charge or peak stroke current, the radiation from the signal will not vary linearly with return stroke current. Theoretical models for the negative leader channel such as [23] derive the return stroke velocity from charge and energy balance criteria, leading to a nonlinear but monotonic relation between this velocity and peak current. If perfect rank correlation is assumed, along with the typical range of velocity values and currents as in [17][23], then the lightning location systems would be reporting a variable which is empirically varying with the current raised to the 1.4 or 1.6 power. However, the introduction of such a strong relation between return stroke velocity and peak current also introduces some important contradictions. • • •

Parameters of the lossy transmission line model for the return stroke path, including impedance and propagation velocity, are established during the leader development phase, in advance of the first return-stroke wave front, and are not sensitive to ionization [16]. A similar physical process with reduced energy should govern dart leaders and subsequent strokes – yet comparison of return stroke velocity and peak subsequent stroke currents do not show any correlation [24]. If the peak first-stroke current I and first-peak remote radiated field H are individually related by a power (such H ∝ I1.6) rather than a linear relation, they cannot both be log-normal [25], and the standard deviation of the radiated-field distribution σln H should be greater than σln I. In fact, with an especially high degree of confidence in statistics with millions of radiated field observations, both distributions are log-normal and the values of σln I and σln H are equal.

For the present, it is sufficient to acknowledge that the measurements of radiated field provide estimates of negative peak stroke currents that are modulated by the distribution of first negative downward flash return stroke velocity. Modeling and observation of positive downward leader development and return stroke has not reached anywhere near the level of attention given to negative flashes, since the incidence of positive flashes is relatively low. 3.3 Observations of Positive-Flash Events The common assumption that the return stroke velocity is independent of peak stroke current, and also that the return stroke velocities of first negative and positive downward flashes are similar, based on the experimental evidence available, can form a limited basis for risk analysis for areas with severe charge. Positive flashes tend to have somewhat slower rise times than negative flashes. This means that they are relatively immune to problems with signal attenuation over lossy ground. Most lightning location networks introduced a deliberate bias against the detection of positive flashes, by using a trigger threshold that was higher than for negative

flashes at each receiver. The technical purpose of this bias was to reduce the number of false detections from cloudflash signals. This bias generally does not affect the detection efficiency for large-amplitude flashes. Network bias towards more efficient detection of large positive flashes is also seen as an increasing ratio of positive to negative flashes at the fringe of the network, for example in the Atlantic Ocean in Fig. 3 above. A typical distribution of lightning amplitudes, estimated from radiated fields, is given in Fig. 5. For this study in Italy [26], there are nearly twice as many positive-polarity large peak-current events with inferred amplitude over 75 kA.

Fig. 5 – Measured frequency of large-amplitude lightning flashes (>75 kA) in Italy [26]

3.4 Observations of Large-Amplitude Positive-Flash Events Large positive-flash events are expected to have more charge than large negative flashes. The fact that they are positive means that the median charge level should be 80 Coulomb rather than 7.5 Coulomb for the total charge in the negative flash. The fact that they have large peak amplitude suggests, with a correlation coefficeint of 0.62, that they also have large charge. While exploring an anomaly in the ratio of optical to ground flashes, Boccippio et al [27] noted that the density of large positive flashes in the USA, shown in Figure 6, is considerably different from the total ground flash density, which has its maximum values around the Gulf of Mexico and Florida.

Fig. 6 – Measured frequency of large-amplitude positive lightning flashes (>75 kA) in USA [27]

The level of +75 kA would suggest a median charge of 113 Coulomb using Table 2.

The anomaly of Fig. 6 in the USA state of South Dakota actually continues north into the Canadian provinces of Manitoba and Saskatchewan [28]. In this more recent study, flashes above 100 kA were considered to be large. The median charge associated with 100 kA is 129 Coulomb. The distribution of large negative flashes was relatively uniform and had a closer match to the overall ground flash density. The daily peak in the diurnal cycle of largeamplitude events appeared at midnight local time [28], compared to about 6 PM for all events.

Fig. 7 – Measured frequency of large-amplitude positive lightning flashes (>100 kA) in Canada [28]

In Brazil, the Rede Integrada Nacional de Descargas Atmosféricas (RINDAT) network in 2003 became a nationally integrated set of lightning location sensors, initially within the states of Parana, Sao Paulo, Mato Grosso del Sul, Goias and Minas Gerais. The initial 24 sensors provided uniform and efficient detection of flashes within these states. Detection efficiency of large-current positive flashes in neighboring Bolivia, Paraguay and Argentina is also relatively high. Generally, the ratio of positive to total lightning events increases outside lightning location neworks as a result of the increase in the ratio of positive-stroke to negative-stroke detection efficiency. This is a network deficiency, but it does not contribute much bias to the basic observation of large-amplitude positive-flash density that is shown in Fig. 8. The density of positive flashes with I>75kA has been tabulated in areas of 4 km by 4 km. The RINDAT observations for a warm season show extensive areas with density of large-amplitude positive flashes, reaching a peak density of 0.75 large flashes per km2 per year. Within the boundary of the RINDAT states, elevated levels are noted in the north of the Paraná, the west of Sao Paulo and Minas Gerais, and south Goias. It is interesting that the density of high-amplitude positive flashes on both continents increases somewhat from equator to mid-latitudes, and that the high-density areas are located about 1000 to 1500 km east of the major mountain range. There are other, more specific meteorological similarities. The significant large-positive flash density values areas observed in southeastern South America are located mainly in the area of genesis and intensification of mesoscale convective systems (MCS), which are very frequent at the warm season of the year [29]. The South American lowlevel jet is also observed over this area. This jet advects heat and moisture from the Amazon basin southward into central plains over that area, increasing the instability of the atmosphere. This means that the high positive-flash density areas in Fig. 8 have many similarities with the Great Plains of the United States, where the maximum of atmospheric discharges with I>75kA density also is verified [27].

Fig. 8 – Political map (left) and measured density (flashes/km2) of warm-season large positive lightning flashes (>75 kA) in Brazil

For reference, the peak level of large positive-flash (>75 kA) density reported by [27] was 0.128 flashes per km2 per year. It is clear that extensive areas in Brazil, with 0.25 or more positive flashes greater than 75 kA per km2 per year, exceed the peak density observed in the USA by a factor of two. With positive-flash peak current being a reasonablycorrelated predictor of positive flash charge, this means that there are extensive areas of Brazil and its neighbors where flashes with a median of 118 Coulomb of charge could be expected at this same density of 0.25 per km2 per year. 3.5 Esimating the Incidence of Positive Flashes with Large Charge The attractive radius of a transmission line to a large, positive lightning flash has not been established by observation. In the high voltage laboratory, breakdown of gaps using negative polarity requires higher voltage than positive polarity. However, as the gap length increases, [15] explains that this difference diminishes. Gaps on the order of 100 m have the same impulse flashover voltage under positive and negative polarity. This suggests that the use of expressions derived for the incidence of negative flashes to transmission lines using switching-surge leader inception criteria [14][30] can also be applied to positive flashes. For a large flash of 60-kA amplitude, and transmission line height of 30 to 60 m, the attractive width derived by Rizk [30] is between 120 and 150 m. Considering the observed density of large positive-flash events (0.25 per km2 per year) and a transmission line attractive distance of 120 m, the estimated rate of positive flashes delivering a median of 118 Coulomb of charge would be six per 100 km of transmission line length per year. Statistically, one of these six will be more than a full standard deviation greater than the median value. From [17], σln(Q+) is about 0.90 for the positive flash, so this average “extreme” event each year would have more than 290 Coulomb of charge. In cases where charge durability requirements for reliable OPGW lead to excessive costs, the use of lightweight OPGW, strung below the phase conductors using inter-phase spacers, may be competitive, especially if a significant improvement in lightning backflashover rates is desirable. While not used only for the improved lightning performance, an under-built shield wire was fitted to a recent compact urban 230-kV transmission line design [31].

4 TEST METHODS FOR EFFECTS OF CHARGE ON OVERHEAD GROUNDWIRES AND OPGW 4.1 Traditional Aerospace Test Methods While instrumented aircraft have been used for these measurements [32], aerospace protection standards were based more on laboratory reproduction of observed in-flight damage such as [1]. After a pair of problems in the 1960s, the SAE, responsible for these requirements, revised its evaluation and their recommendations were accepted in the aeropsace industry in the early 1980s [33]. Four different components of current were identified, each posing a different risk. For example, the steep rate of change of current in subsequent strokes was represented in component “D” of Fig. 9. Specialized facilities such as [34] were constructed to carry out these tests using the combination wave for various nuclear, aerospace and military requirements.

Fig. 9 – Schematic, four components of lightning threat currents to aerospace vehicles [33]

For transfer of charge, the component “C” highlighted in Fig. 9 now forms the main basis of testing methods to establish the most severe damage to overhead condcutors. 4.2 IEC/IEEE Test Method The present IEEE Standard [6] for optical fibre ground wires, dating from 1994, was re-affirmed without revision in 2002 while a series of major changes were being debated with regard to lightning arc testing. The 1994 standard held that OPGW with adequate ampacity for typical transmission-line short-circuit requirements would have ample I2t margin for any conceivable lightning stroke. However, this approach did not consider either the fast-rising nature of lightning or the high plasma temperature associated with lightning arc roots. Generally, damage from simulation of the first stroke, such as component “A” in Fig. 9, is minimal. Many possible standardized test waves for lightning surge curents can also be criticized for use for testing lightning damage to conductors, OHGW and OPGW. The IEEE 4μs x 10μs and 8μs x 20μs current waves are specific for silicon carbide surge arrester testing, and not appropriate for simulation of conductor damage, even at high test levels of up to 200 kA. The major effect of any of these impulse waves is to melt the fuse wire and spray it onto the tops of the OPGW strands. At present, a revised 2007 version of the IEEE Standard 1138 is in the ballot process, under supervision of the IEEE Standards Association. The lightning arc test in this standard has been harmonized with the IEC test specification [2], and calls for application of a test current for 500 ms using a geometry shown in Fig. 10.

Fig. 10 – Recommended arrangement of lightning arc test for optical fibre groundwires in IEC 60794-1-2 [2]

While the idea of having test levels that were negotiated between supplier and purchaser was fully considered, in the end, the IEC philosophy of having four threat levels, Classes (0,1,2,3), was endorsed. The test levels for 0.5-s pulse duration specify (100, 200, 300, 400A) for (50, 100, 150, 200 Coulomb). The electrode polarity is specified as positive rather than negative, simulating the effects of positive flashes which have higher median charge than negative flashes. These continuing current test levels do progressive damage to conductors as charge level increases. However, test-totest consistency cannot be achieved using the IEC specifications alone. The proposed geometry includes a rod electrode item (5) in Fig. 10, with its tip 6 cm from the conductor and connected initially by a metal fuse wire. The IEC specification for the electrode is “iron or copper” with wolfram copper being preferred. The rod is mounted with its axis aligned with the radius of the cable, with the tip 6 cm from the cable surface. The rod is connected to the cable with a fuse wire that will melt early in the test wave. One reason for selecting positive polarity on the electrode in Fig. 10 relates to the relative stability of the arc, compared to negative polarity. This effect was also noted on metal surfaces [42]. The choices of electrode material, radius and end treatments also affect shot-to-shot consistency. The IEC spacing and polarity is biased towards a stable arc with relatively high currents, and the arc tends to wander for lower levels of charge exposure. This can be corrected with a symmetrical cage for the return wiring. This geometry needs to be magnetically balanced with a minimum of four conductors forming a cage around the testing arc, item (8) in Fig. 10. While the IEC standard specifies only a minimum distance of 1 m between anchor clamps in Fig. 10, tests performed on OPGW samples that leave a minimum of 2 m on either side of the test point will allow better distribution of tension among the strands. This is especially important if the failure criteria will be established with a residual strength test rather than by counting the number of broken strands. The cable should be terminated at each end and tensioned to 20±5 % of rated tensile strength (RTS) during the lightning arc test. This normal operating tension will allow realistic strand movement. The surface of the cable in the test is maintained at room temperature, 23 ± 5 °C with no need to maintain 40°C as for short-circuit tests. This parameter is less important for the lightning arc test because the arc temperature is high. The conductor damage from shot to shot can vary, so five samples are to be tested with positive electrode polarity. Standard welding handbooks suggest that a positive arc root is three times more damaging than negative, meaning that positive lightning can often be less damaging than negative lightning in spite of higher charge levels, but testing experience suggests that results vary depending on the material and wire strand diameter, as well as polarity. Damage criteria related to changes in optical attenuation and number of broken strands will normally be monitored during the electrical tests. Disputes about the remaining strength of partially melted strands with this approach can be resolved by monitoring tension while pulling the test sample to destruction.

4.3 Lightning Arc Test Results on OPGW, OHGW and Conductors There have now been many published photos [35-39] of the typical damage to optical fiber groundwires from real lightning and from simulated lightning arc tests using a low-current pulse of about 500 ms duration. Overall, there is a good agreement, with some broken strands being observed in severe cases but few examples of complete conductor failure. Recently, with the possibiltiy of using transmission line surge arresters instead of overhead groundwires being studied in areas of heavy icing, more attention has been given to the effcts of charge damage on typical phase conductors. As an example, Fig. 11 shows obsevations of lightning arc test damage to the common 795-kcmil Drake conductor. The susceptibility of ACSR to the same type of charge damage as overhead groundwires means that, for example, the use of transmission line surge arresters to replace overhead groundwires will call for more frequent inspection and repair (using armor rods) of charge damage. Given the relatively small size of the damaged area, visual inspection at high speed from a helicopter may not have enough resolution to find problems efficiently and other detection tools (thermal, corona noise or corona spectrum) may be needed.

Damage to 795-kcmil ACSR Conductor, -50 Coulomb

Damage to 795-kcmil ACSR Conductor, -189 Coulomb

Fig. 11 – Photos of progressive damage to standard phase conductor from lightning arc tests [39]

4.4 Advanced Modeling of Charge Damage to Metal Components Using simulated component-C lightning arc test waveforms, an excellent relation between arc spot area in sheet metal and charge was shown by [42]. Their thermal model of the arc, including effects of conduction and radiation heat loss, was also found to be satisfactory as shown in Fig. 12. This work serves as a guide to researchers who may take a similar interest in calculating the damageability of overhead conductor designs, using a model that incorporates several domains of physics, including mass transfer, heat transfer, chemical energy released in oxidization as well as heat input from electrical energy in the arc root and stem.

Fig. 12 – Measured arc spot radius and calculated plasma arc root radius in aluminum sheets [42]

4.5 Advanced Testing Methods: Current Waves In 1999, NASA specifications for lightning compatibility of aerospace vehicles [40] included descriptions of: • Multi-pulse stress with twenty-three subsequent flashes • A series of 24 bursts of 20 pulses of a new component “H” with peak current derivative of 2x1011 A/s • Tests involving standard high-voltage waves [41] and a rising voltage of 1000 kV/μs to flashover. Figs. 13 and 14 show the recommended waveforms in [40] for multi-pulse current tests.

Fig. 13 – Recommended sequency of 24 flashes for testing aerospace vehicles [40]

Fig. 14 – Details of lightning pulse burst test using Component H waveforms [40]

Multi-pulse testing of surge arresters in the electric power industry has also been evaluated, as first described by Darveniza et al [43]. Recently, Lee and Kang [44] carried out experiments with this approach. They reported that surface flashover was a dominant failure mechanism with this stress rather than accumulated energy from the impulse charge. Fig. 15 shows such a failure on the fifth and final impulse of one of their tests.

Fig. 15: Visual discharges around metal oxide surge arrester during multipulse test [44] 4.6 Advanced Testing Methods: Voltage Waves Many aerospace components such as radomes and propeller blades are constructed of non-conducting materials such as fiberglass or carbon fibre. For radomes, covering the radar equipment in the nose of aircraft, electrically conducting paths to divert lightning interfere with the radar signal. For blades, including wind turbine blades, the inclusion of a continuous metal path or a series of conducting areas increases cost, reduces strength and degrades environmental reliability. It is especially important that local electric fields around these conductors under high electric fields do not de-laminate or damage the nonconducting substrate. For qualification of semiconducting or inslating components such as radomes for airplanes or wind turbine blades, it has been noted [45] that voltage impulse tests using switching-surge waveforms [41] with 250-μs rise time and 2500-μs time to half value have been more successful at reproducing obseved damage than tests with the standard 1.2/50 μs wave. In [40], the voltage waveform “D” is recommended as illustrated in Fig. 16. Actual tests of switching surge breakdown tend to have a minimum flashover voltage for a front time of 150 to 300 μs, as shown for example for rodto-plane gaps in Fig. 17 [15].

Fig. 16 – High-voltage waves for testing aerospace components [40]

Fig. 17 – Breakdown voltage of rod-to-plane gaps as a function of switching-surge rise time tf [15]

The physical buffers needed to manage risks of broken-blade trajectory away from wind turbines may constrain the application of wind power, especially in populated areas. Appropriate high-voltage flashover tests on the blades, using switching surges [46] to simulate the lightning effects in non-conducting materials, may provide wind-turbine manufacturers with an important new way to reduce the risks of blade damage or failure. 5 CONCLUSIONS Measurements of subsequent return strokes from tower- or rocket-triggered lightning contain continuing currents and other components that should be representative of those in natural lightning. The slow rate of rise and modest contribution of M components to the overall level of charge transfer can be considered as a part of the overall charge transfer when evaluating the parameters for total flash charge. Large positive-flash events can be classified as those that have radiated field strengths that convert to source currents of more than +75kA using a transmission-line return stroke model and a fixed return-stroke velocity of c/3. Detection efficiency of large positive-flash events in lightning location networks tends to be higher than for negative flashes, as shown by an increasing ratio of positive to negative events at the fringe of most networks. Large positive flashes of 75 kA should have a median 118 Coulomb of charge, compared to 85 Coulomb for all positive flashes. Lightning location system measurements show areas where the incidence of large positive flashes is elevated. Areas of high density of flashes exceeding +75 kA are found in the Great Plains of North America and in a similar region in southeastern South America, far from areas of highest negative ground flash density. The spatial extent of the area of elevated large-flash density can be on the order of 500,000 km2, with increasing density from equator to pole, with the highest density areas located 1000-1500 km east of major mountain ranges. Lightning arc test levels of 50 to 200 Coulomb, using currents of 100 to 400 A for 500-ms pulse duration, with positive polarity of electrode relative to conductor, have received wide acceptance as a suitable method for establishing the damaging effects of lightning on optical fiber groundwires (OPGW). This type of test is also appropriate for overhead groundwires and for phase conductors that will be used on unshielded transmission lines. Most recently, the IEEE 1138 Standard has been harmonized with the requirements in the 2003 version of IEC 60794-1-2 [3] in this respect. Areas with a high density of large positive flashes may be more appropriate for IEC Class-3 test level of charge (+200 Coulomb) in the lightning arc test. High-voltage tests on non-conducting aerospace and wind turbine blades use standard switching surge waveforms to simulate damage. The ability of these tests to reproduce field damage has become an interesting feature of the updated test specifications by SAE/EUROCAE [46].

6 REFERENCES [1] J.H. Hagenguth, “Lightning Stroke Damage to Aircraft”, AIEE Transactions, Vol. 68, pp. 1036-1046, 1949. [2] ABNT, "Cabos pára-raios com fibras ópticas para linhas aéreas de transmissão (OPGW) - Determinação dos efeitos da descarga atmosférica - Método de ensaio", NBR14586, ABNT/CB-03 – ELETRICIDADE, August 2000. [3] IEC Standard 60794-1-2, “Optical fibre cables –Part 1-2: Generic specification –Basic optical cable test procedures”, Lausanne, International Technical Commission, May 2003. [4] IEC Standard 60794-4, “Optical fibre cables – Part 4: Sectional specification - Aerial optical cables along electric power lines”, Lausanne, International Technical Commission, June 2003. [5] IEC Standard 60794-4-10, “Optical fibre cables – Part 4-10: Aerial optical cables along electric power lines – Family specification for OPGW (Optical Ground Wires), August 2006. [6] IEEE Standard 1138-2002, “Construction of Composite Fibre Optic Overhead Groundwire (OPGW) for Use on Electric Utility Power Lines”, Piscataway, IEEE Press, 2002. [7] NASA, “Launch Pad Lightning Warning System (LPLWS)”, Meteorological Report CCAFS\81900\, January 2002. [8] V. Mazur and L. Rhunke, “Evaluation of the Lightning Protection System at the WSR-88D Radar Sites”, National Oceanic and Atmospheric Administration Final Report, May 2001. [9] V.A. Rakov, D. E. Crawford, K. J. Rambo, G. H. Schnetzer, M. A. Uman, and R. Thottappillil, M-component mode of charge transfer to ground in lightning discharges, Journal of Geophysical Research, 106(D19), 22,817–22,832, 2001. [10] V. Jayakuma, “Estimating Power, Energy, and Action Integral in Rocket-Triggered Lightning”, MS Thesis, University of Florida, 2004. [11] K. Berger, R.B. Anderson and H. Kröniger, “Parameters of Lightning Flashes”, Electra, No.41, pp. 23-37, 1975. [12] G. Diendorfer, M. Mair, W. Schutz, W. Hadrian, “Lightning Current Measurements in Austria – Experimental Setup and First Results”, Proceedings of 25th International Conference on Lightning Protection, 1.14, Rhodes, Sept. 2000. [13] L.R. Soriano, F. de Pablo and C. Tomas, “Ten-year study of cloud-to-ground lightning activity in the Iberian Peninsula” Journal of Atmospheric and Solar-Terrestrial Physics, Vol. 67, pp. 1632–1639, 2005. [14] F.A.M. Rizk, “Switching Impulse Strength of Air Insulation: Leader Inception Criterion”, IEEE Transactions on Power Delivery, Vol. 4 No. 4, pp. 2187-2195, October 1989. [15] E.M. Bazelyan and Y.P. Raizer, Spark Discharge, Boca Raton, FL: CRC Press, 1998. [16] V. A. Rakov and M. A. Uman, Lightning - Physics and Effects, Cambridge, Cambridge University Press, 2003. [17] P. Chowdhuri, Task Force Chair, Lightning and Insulator Subcommittee of the T&D Committee, “Parameters of Lightning Strokes: A Review”, IEEE Transactions on Power Delivery, Vol. 20, No.1, January 2005. [18] B.A. Zajac and S.A. Rutledge, “Cloud-to-Ground Lightning Activity in the Contiguous United States from 1995 to 1999”, Monthly Weather Review, Vol. 129, pp. 999-1019, May 2001. [19] W. Burrows, “Lightning in Canada: Detection, Forecasting and Climatology”, Proceedings of CEATI Seminar on Transmission Line Surge Arrester Application, Montreal, September 2006. [20] V. Cooray, ed., The Lightning Flash, Cornwall: IEE Press, 2003. [21] K.L. Cummins, E.P. Krider and M.D. Malone, The U.S. national lightning detection network and applications of cloud-toground lightning data by electric power utilities, IEEE Transactions on Electromagnetic Compatibility, Vol. 40 no. 4, pp. 465–480, April 1998. [22] F. Rachidi, “The Quandary of Direct Measurement and Indirect Estimation of Lightning Current Parameters”, 27th International Conference on Lightning Protection, Avignon, France, September 2004. [23] C.F. Wagner, “The Relation between Stroke Current and the Velocity of the Return Stroke”, AIEE Transactions, pp. 609617, October 1963. [24] D.M. Mach and W.D.Rust, “Photoelectric return-stroke velocity estimates and peak current estimates in natural and triggered lightning”, Journal of Geophysical Research, Vol. 94, pp. 13237-47, 1989. [25] W.A. Chisholm and K.L. Cummins, “Lightning Parameters: A Review, Applications and Extensions”, Proceedings of IEEE T&D Expo, Panel PN-11, Dallas, TX, May 2006. [26] L. Telesca, M. Bernardi, C. Rovelliet, “Time-scaling analysis of lightning in Italy”, Communications in Nonlinear Science and Numerical Simulation, doi:10.1016/j.cnsns.2007.02.001, February 2007. [27] D.J. Boccippio, K.L. Cummins, H.J. Christian and S.J. Goodman, “Combined Satellite- and Surface-Based Estimation of the Intracloud–Cloud-to-Ground Lightning Ratio over the Continental United States”, Monthly Weather Review, Vol. 129, pp. 108-122, January 2001. [28] B. Kochtubajda, W. R. Burrows , B.E. Power, “Large Current Flashes in Canada”, American Meteorology Society Bulletin P2.11, 2007.. [29] P. Salio, M. Nicolini and E.J. Zipser, “Mesoscale Convective Systems over Southeastern South America and Their Relationship with the South American Low-Level Jet”, Monthly Weather Review Vol. 135 No. 4, April 2007, pp.12901309 [30] F.A.M. Rizk, “Modeling of Transmission Line Exposure to Direct Lightning Strokes”, IEEE Transactions on Power Delivery, Vol. 5 No. 4, pp 1983-1997, November 1990. [31] Hoffmann, J. N., R. L. de Souza, N. Prosdócimo, I. Moreira, F. V. Swinka. “Linha de Transmissäo Urbana Compacta Experimental em 230 kV.” XVII SNPTEE. , Minas Gerais, Brazil October 2003. [32] Nanevicz and Adamo, “Airborne Measurement of Electromagnetic Environment near Thunderstorm Cells (TRIP-76), NASA Report AFFDL-TR-77-62, August 1977.

[33] USA Military Standard MIL-STD-1757 Revision A, “Lightning Qualification Test Techniques for Aerospace Vehicles and Hardware”, 1983. [34] M. Caldwell and L.E. Martinez, “The Sandia Lightning Simulator: Recommissioning and Upgrades”, 2005 International Symposium on Electromagnetic Compatibility, Vol. 2, pp. 368-371, August 2005. [35] W.A. Chisholm, J.P. Levine and P. Chowdhuri, “Lightning arc damage to optical fibre ground wires (OPGW): parameters and test methods”, IEEE 2001 Power Engineering Society Summer Meeting, pp. 88– 93, Vancouver, July 2001. [36] T. Du, Y. Zhang and W. Xia, “Study on the problem of lightning strike OPGW”, Proceedings of 2006 Intl. Conf. on Power Systems Technology, PowerCon, Chongqing, October 2006. [37] Y. Goda, S. Yokoyama, S. Watanabe, T. Kawano and S. Kanda, “Melting and breaking characteristics of OPGW Strands by Lightning”, IEEE Transactions on Power Delivery, Vol. 19, No. 4, pp 1734-1739, October 2004. [38] M. Yokoya, Y. Katsuragi, Y. Goda, Y. Naga and Y. Asano, “Development of lightning-resistant overhead ground wire”, IEEE Transactions on Power Delivery, Vol. 9, No.3, July 1994. [39] H. Deve and T. Anderson, "3M Aluminum Conductor Composite Reinforced Technical Notebook (795 kcmil family), Conductor and Accessory Testing", April 2003. [40] C.C. Goodloe, “Lightning Protection Guidelines for Aerospace Vehicles”, NASA/TM-199-209734, May 1999. [41] IEEE, Standard Techniques for High-Voltage Testing, IEEE Standard 4-1995, October 1995. [42] R. Brocke, F. Noack, F. Reichert, L. Ruales and J.Schönau, “The effects of long duration lightning currents and their simulation”, Proceedings of 25th International Conference on Lightning Protection, 5.7, Rhodes, Sept. 2000. [43] M. Darveniza, D. Roby and L.R. Tumma, “Laboratory and Analytical Studies of the Effects of Multipulse Lightning Current on Metal Oxide Arresters”, IEEE Transactions on Power Delivery, Vol. 9 No. 2, April 1994. [44] B-H Lee and S-M Kang, “Properties of ZnO varistor blocks under multiple lightning impulse voltages”, Current Applied Physics Vol. 6 pp. 844-851, May 2005. [45] A. Hall, “Thunderstorm Protection of Aircraft Radomes”, Proceedings of the International Conference on Lightning and Static Electricity, 19 September 2005. [46] SAE International, Ae-2 Lightning Committee, “Aircraft Lightning Test Methods, Section 5, 3”, SAE ARP5416 / EUROCAE ED105, March 2005.