NECESSARY SEPARATION DISTANCES FOR LIGHTNING PROTECTION SYSTEMS - IEC REVISITED

Transcription

1 X International Symposium on Lightning Protection 9 th -13 th November, 29 Curitiba, Brazil NECESSARY SEPARATION DISTANCES FOR LIGHTNING PROTECTION SYSTEMS - IEC REVISITED Fridolin H. Heidler 1, Wolfgang J. Zischank 2 1 University of Federal Armed Forces, Munich, Germany fridolin.heidler@unibw.de 2 University of Federal Armed Forces, Munich, Germany wolfgang.zischank@unibw.de Abstract - In case of direct lightning strike to a building dangerous sparking may occur between the external lightning protection system and conductive installations inside the building. To avoid such side flashes a minimum separation distance between conductive parts inside the building and the air termination or down system is required. The standard IEC provides formulae to determine the necessary separation distance. Objective of the paper is to re-visit the determination of separation distances. The international standard IEC recommends using electrically conductive parts of a building or structure as so-called natural components of the lightning protection system. The use of such extended metal parts can lead to significant reduction of the necessary separation distances. Objective of the paper is also to study the necessary separation distances for buildings using metal roofs, walls or attics as integral part of the lightning protection system. Such configurations are not yet covered by the IEC-standards. 1 INTRODUCTION According to the standard IEC 6235 part 1 [1] and part 2 [2] the lightning protection system (LPS) consists of the external and the internal LPS. Components of the external LPS are the air termination system, the down system and the earth termination system. The functions of the external LPS are to intercept the lightning flash, to conduct the lightning current safely to earth and to disperse it into the ground. In case of direct lightning strike, dangerous sparking may occur between the external LPS and metal installations inside the building. The function of the internal lightning protection is to prevent such dangerous sparking either by lightning equipotential bonding or by keeping the separation distance. The separation distance is the minimum clearance required at the proximity of conductive parts inside the building and the external LPS to avoid side flashes. According to the standard IEC [2] the necessary separation distance is given by the following equation kc s = ki l (1) k m The induction coefficient k i takes into account the highest voltage expected at the proximity considering the behaviour of the dielectric strength of air for sub-microsecond impulse voltages [3, 4]. With the steepness (di/dt) of the lightning current, the induced voltage is given by u = M ( di / dt) (2) The mutual inductance M is given by the structure resulting from the wiring of the external LPS and of installations inside the building. Because the subsequent strokes have much higher current steepness than the first strokes, in IEC [2] only the subsequent strokes are considered as short stroke current with the waveform.25/1 μs (front time T 1 =.25 μs, decay time to half value T 2 =.25 μs) [1]. Four lightning protection levels (LPL) are defined, the highest level being LPL I and the lowest level LPL IV. According to that classification, the current peak (5 ka) is highest for LPL I and reduced to 37,5 ka for LPL II and to 25 ka for LPL III and LPL IV. With the reduction of the current amplitude, the average front steepness i max /T 1 is lowered from 2 ka/μs for LPL I to 15 ka/μs for LPL II and to 1 ka/μs for LPL III/IV, seen in Table 1. 91

2 Corresponding to the four lightning protection levels, in IEC [2] four classes of LPS (I, II, III, IV) are defined. In the highest class of LPS I the induction coefficient is fixed to k i =,8 and reduced to k i =,6 and k i =,4 for LPS II and LPS III/IV considering the reduction of the current steepness (see Table 1). The configuration coefficient k c takes into account the percental current share to the individual down s being the same for all classes of LPS. Fig. 1 visualizes the configuration coefficient k c for three basic arrangements. The arrangements apply for simple structures with earth termination systems consisting of horizontal or vertical earth electrodes [2]. For a direct strike to a Franklin rod, the total current flows through that rod and thus the configuration coefficient results in k c = 1 (Fig. 1a). For the 2-dimensional arrangement in Fig. 1b, approximately 66 % of the lightning current flows through the outer down. That current share gives the value k c =,66. For the 3- dimensional arrangement in Fig. 1c, the fraction of current flowing through the corner down becomes about 44 % resulting in k c =,44. Table 1: Peak current i max, average front steepness i max /T 1 of the subsequent short stroke [1] and the configuration coefficient k i [2] LPL/LPS i max i max /T 1 k i I 5 ka 2 ka/μs,8 II 37,5 ka 15 ka/μs,6 III and IV 25 ka 1 ka/μs,4 a) Franklin rod i o (t) s metallic/electric installation down i o (t) l k c = 1 b) i o (t) air termination wire outer down s w,66 i o (t) l down w k c =,66 c) air termination mesh corner down,44 i o (t) s i o (t) w l down down w k c =,44 Fig. 1 Separation distance s and coefficient k c (a) Franklin rod (1-dimensional) (b) Air termination wire with down s (2-dimensional) (c) Meshed air termination and down system (3-dimensional) 92

3 For 3-dimensional grid-like structures using ring earth electrodes or foundation earth electrodes, in IEC annex C [2] the following formula is given k c 1 2 n c h = +,1 +,2 3 (3) with the spacing between the down-s c, the total number of down-s n and the height of structure h. The coefficient k m finally considers the dielectric strength of materials other than air present at the location of the proximity. Table 2 contains the values of the coefficient k m. For air k m = 1 applies. For construction materials (e.g. brick) this coefficient is reduced to the half. Table 2: Values of the coefficient k m. Material k m Air 1 Concrete, bricks,5 2 MOTIVATION TO REVISIT IEC Originally, Eq. (1) was developed using the vertical distance between the point, where the separation distance is to be considered, to the nearest equipotential bonding point for the length l. Meanwhile the length has been re-defined in the IEC standard as the total length along the air termination and the down s from the point, where the separation distance is to be considered to the nearest equipotential bonding point. The formula for k c later on was refined taking into account the height of the structure and the distance between the down s [2]. Furthermore, the original values of k i were reduced by 2 %. The k i -values (see Table 1) were originally fixed to,1 for LPS I, to,75 for LPS II and to,5 for LPS III and LPS IV [5]. Eq. (1) and the values for the parameters are based on calculations published in the mid 198s [3]. Due to the limited computer capacity available at that time, the modeling was limited to simple one-, two-, and three dimensional (cubic) lightning protection systems consisting of only stretched wires. Objective of the paper is to test the IEC equation for separation distances with state of the art computer codes solving the complete Maxwell equations. The international standard IEC [2] further recommends using electrically conductive parts of a building or structure as so-called natural components of the lightning protection system. Suited as natural components of the air termination and down system are e.g. the reinforcement of concrete, metal roofs, metal facades or attics as long as they are electrically continuously interconnected and provide a sufficient cross sectional area to conduct the lightning current without undue heating. The use of such extended metal parts as natural components of the lightning protection system can result in a significant reduction of the necessary separation distances between conductive parts inside the building and the air termination or down system. However, formulae to determine the necessary separation distance for such natural components are not yet contained in the IEC 6235 standard series. In this paper the influence of metal roofs, metal facades and attics on the separation distance is studied. The calculations were performed with the software CONCEPT II, which solves the Maxwell equations in the frequency domain using the method of moments (MoM). Exemplarily, the structures studied were all equipped with a lightning protection system according to class II of IEC [2]. The dimensions of the structures were varied in a wide range. In each case the induced voltages were calculated for three different points of strike on the roof and for two different induction s inside the structure. From the induced voltages the necessary separation distances were determined using the constant-area-criterion. 93

4 3 CONSTANT-AREA-CRITERION Fig. 2 shows the spark-over behavior of an air gap when exposed to impulse voltages of different steepness. The static onset voltage U o is the threshold below which no spark-over occurs. The voltage-time characteristic shows that the dielectric strength is a function of the voltage waveshape. For fast rising voltages, the air gap sparks over at higher voltage levels than U o. That behaviour is addressed to the fact that a certain period of time is needed to built up the arc between the electrodes of the air gap. The time characteristic of the dielectric strength is taken into account by the well established constant-area-criterion [6]. For unipolar impulse voltages of arbitrary wave shape the following equation applies: t 2 [ u(t) U ] dt = A (4) t1 The definitions used in Eq. (4) are illustrated in Fig. 3. The voltage-time area A is a constant value for a particular air gap. If the air gap is altered by varying the distance s between the electrodes, both the static onset voltage U o and the voltage-time area A are changed. For rod-rod gaps exposed to negative impulse voltages the following values apply [7]: U o s =,63 (5) [ MV ] [ m] A s =,59 (6) [ Vs] [ m] Eqs. (4 6) are used to evaluate the separation distance s between the wiring of the electrical installations inside the building and the external LPS. u(t) u Voltage-time characteristic of an air gap A U o U o t Fig. 2 Spark-over behavior of an air gap stressed by impulse voltages with different steepness t t 1 t2 Fig. 3 Illustration of the constant-area-criterion 4 COMPUTATIONAL APPROACH The fast rising current of the subsequent stroke induces very short voltage impulses in the considered installation s. Fig. 4 gives two examples of the induced voltage waveshapes: Almost any waveshape may occur, from a dominant peak at the beginning followed by only minor oscillations up to only slightly damped oscillations. The oscillations originate from the resonant effects of the considered. The resonances are due to the fact, that the dimensions are not small enough compared to the corresponding wave lengths of the fast rising current. These resonance effects are not covered by the simple formula of Eq. (2). 94

5 More precise computation of the voltage requires the solution of the complete Maxwell s equations. Therefore, the electromagnetic computations are carried out using the computer code CONCEPT, which has been developed during the last two decades by the Technical University Hamburg-Harburg [8]. This computer code is based on the so-called Method of Moments (MOM) [9] and is written in FORTRAN 77. It is a well-known computer code in the area of electromagnetic computations, and has been validated by several tests [8,1]. This computer code solves the complete Maxwell s equations in the frequency domain. Therefore, the time-domain solutions of currents and voltages are obtained from the inverse Fourier transformation. The fundamental assumptions of the computer code are given in [8,11] and the handling of the program package is described in [11]. 2 1,5 (MV) 1,5 u -,5 (a) 1,5 (MV) 1,5 u -,5,5 1 1,5 2 t (µs) -1 (b),5 1 1,5 2 t (µs) Fig. 4 Examples of voltage shapes induced by the subsequent return stroke current (waveform.25/1μs) (a) LPS type a (Fig. 6), Air termination: mesh, Point of strike: corner, Induction: Corner (b) LPS type b (Fig. 7 ), Air termination: metal roof, Point of strike: center, Induction: Center wire In the computer code CONCEPT, the return stroke process is simulated with the transmission-line (TL)-model introduced by Uman [12]. Using this model, the return stroke channel is assumed to be straight and perpendicular to the earth surface. The return stroke channel is considered to increase along the z-coordinate with the constant return stroke velocity chosen to v = 1 m/s. The TL-model only describes the interferences due to the return stroke process. This involves that the coupling of the lightning channel to the structure under study is taken into account, while the coupling from the structure back to the lightning channel is ignored. The TL model uses a pre-defined current source i B (t) at the channel-base, from where the time-varying current waveform propagates upwards in z-direction (Fig. 5). The current along the lightning channel as a function of time t and coordinate z is given by i( z, t) = ib( t z / v) (7) i(z,t) v z Lightning channel i B (t) striking point Fig. 5 Propagation of the lightning current along the return stroke channel according to the TL-model 95

6 This behavior is transferred to the frequency domain using the time shifting theorem of the Fourier analysis. The propagation of the current wave is simulated by means of the following equation z jω v I( z) = Î B e (8) Î(z) denotes the current on the lightning channel at the coordinate z and Î B is the channel-base current [13]. According to the IEC standard [1], the channel-base current i B (t) of subsequent stroke is simulated with a front time of T 1 = 25 ns. The following channel-base current it considered in the paper: ib / ib ( t) = T1 ib / max max t,, for for t T1 t T 1 (9) Eq. (9) defines a lightning current with a constant steepness during the current rise. After the current rise the current is kept constant at the peak value i B/max. The so-called thin wire approach is used to simulate the cylindrical s. The cylindrical s of the air termination system and of the down system are taken into account with the radius of 4 mm and with the conductivity of 56,2 1 6 S/m. These values are typical for an external lightning protection system consisting of copper. Metal sheets are simulated by rectangular and triangular patches assumed as ideal s. The segmentation of the patches can be seen in the Figs The ground is considered as plane also with ideal conductivity. Isolating materials other than air are disregarded (k m = 1). Three different frequency regimes are chosen in order to minimize the number of frequencies. Starting with a lowest frequency of 1 khz, the frequency is increased in steps of Δf = 2 khz up to 99 khz. Then in the second frequency regime, the frequency step is increased to Δf = 3 khz up to 2 MHz. In the highest frequency regime between 2 MHz and 2 MHz, the frequency step is further increased to Δf = 4 khz. As a general rule, the dimensions of the wires and of the patches should not exceed about λ/8, where λ is the wavelength of the highest frequency considered. In the paper, the highest considered frequency of 2 MHz corresponds to the wavelength of 15 m. Consequently, the wires and patches were subdivided into segments with maximum dimensions of 2 m. 5 EXAMINED STRUCTURES Various structures simulating the dimensions of typical buildings were selected for this study. Four structures had a base area of 2 m x 2 m with different heights of 1 m, 2 m, 4 m and 6 m, seen in Figs Further, a fifth structure with a large base area of 6 m x 6 m and a height of 1 m was included, seen in Figs. 11,12 [14,15]. The mesh size of the wire air termination system was 1 m x 1 m and the interspacing between the down s was 1 m, corresponding to a LPS of class II in IEC [2]. Whenever applicable, ring s, horizontally interconnecting the down s, were installed every 1 m of structure height. Four basic types of lightning protection systems (LPS) were considered for these five structures, as shown in Table 3. The LPS type a comprises stretched wires used for the air termination and the down s, as shown in Fig. 6. Fig. 7 and Fig. 8 show the LPS of the types b and c. Compared to the LPS of the type a, in the LPS of the types b and c the stretched wires are substituted by flat metal plates in order to simulate the metal roof and the metal walls, respectively. LPS type d shown in Fig. 9 considers the typical case of a circumferential metal attic at the roof. Three different lightning attachment points are considered, as shown in the Figs. 6-9: to the corner of the roof, to the middle of the roof side and to the center of the roof. In the following, they are denoted as corner strike, side strike and center strike. For calculation purposes, at these locations short lightning rods of 1 m length are placed and connected to the air termination system. The vertical lightning channel is attached to the tops of the rods. The channel-base current is injected at these attachment points corresponding to the TL-model (see Fig. 5). According to LPL II of IEC [1] the peak value of a subsequent stroke is i B/max = 37,5 ka, the front time being T 1 = 25 ns (see Table 1). 96

7 For the evaluation of the necessary separation distances two induction s were installed inside each structure. They are denoted as corner and center wire and shown as dashed lines in the Figs The corner starts from the roof corner of the structure with a 1 m long horizontal section pointing diagonally to the center. Following, a vertical section leads to ground. Of course, in case of the metal roof the horizontal section is missing. The center wire connects the air termination system and ground vertically in the center of the structure. Each is loaded by a high resistance of 1 MΩ in order to simulate open conditions at the proximity between the lightning protection system and internal conductive parts. Type of LPS Table 3: Types of lightning protection systems Air termination Down s a meshed wires wires b metal roof wires c meshed wires metal walls d meshed wires with metal attic Wires corner strike side strike corner strike center strike side strike corner center wire corner center wire Fig. 6 LPS type a Fig. 7 LPS type b corner strike corner center strike side strike corner strike side strike center strike corner center wire Fig. 8 LPS type c Fig. 9 LPS type d 97

8 6 CURRENTS 6.1 Current distribution to the down s The share of the injected current to the down s is determined for the asymmetric case of a lightning strike to the corner of the roof. Of special interest is the current through the corner down located directly beneath the point of strike: The ratio of the peak current through this down, i 1/max, to the peak of the incident lightning current, i B/max, equals to the parameter k c of Eq. (1). Table 4 gives the ratios of the corner down peak current i 1/max to the incident lightning current peak i B/max for the meshed air termination system (LPS type a ) and the flat metal roof (LPS type b ). In comparison to the meshed wire air termination also the values of k c according to Eq. (3) are listed in Table 4. In case of the structures with meshed air termination systems the values for k c according to IEC are in good agreement to the values calculated, the maximum deviation being 18 % in case of the large 6 m x 6 m structure. Fig. 1 shows the percentage current share p = i n/max / i B/max to the down s for the large 6 m x 6 m base structure. The numbering n of the down s can be seen from Figs. 11 and 12. Table 4: Ratio of the corner down current to the incident lightning current for strikes to the corner Structure size i 1/max / i B/max Meshed wire k c acc. to Eq. (3) flat metal roof x 1m,4,36,25 x 2m,33,32,22 6m x 6m x 1m,39,32,22 p [%] b a n Fig. 1 Percentage current distribution to the down s for the 6 m x 6 m x 1 m structure in case of corner strike. a) LPS type a with meshed air termination b) LPS typ b with flat metal roof Corner strike Side strike Center strike Meshed air termination system Corner strike Side strike Center strike Flat metal roof Corner 4 5 Center wire Down Corner Down s Fig. 11 LPS type a (6 m x 6 m x 1 m) with meshed air termination system of 1 m x 1 m mesh size and down s with 1 m interspacing Fig. 12 LPS typ b (6 m x 6 m x 1 m) with metal roof and down s with 1 m interspacing 98

9 Obviously, the down at the corner and its immediate neighbours carry the bulk of the current, while the rest of the down diverts only 5 % or less of the incident lightning current to ground. It should be noted that also in the case of the metal roof a remarkable part of the incident current flows through the corner down (about 22 %), although it is clearly less compared to the case of a meshed wire air termination (about 4 %). 6.2 Influence of the lightning current waveform For comparison, calculations were also performed using lightning current waveforms other than the linear rise according to Eq. (9). In these cases, the injected negative subsequent stroke (i B/max = 37,5 ka, T 1 = 25 ns) was simulated using the standardized lightning waveform of IEC annex B [1] given with Eq. (1) as well as a double exponential current waveform given with (11): i ( t) B B i ( t) i ( t / τ ) 1 = B / max 1 t / τ 2 e 1 η 1 + ( t / τ1), with τ 1 =.454 μs, τ 2 = 143 μs and η =.993 (1) i t / τ1 t / τ 2 ( e e ) B / max =, with τ 1 =.92 μs, τ 2 = 143 μs and η =.995 (11) η The comparison was performed for the x 1m LPS typ a structure and with lightning current injection to the roof corner. The maximum voltages induced to the corner are quite similar for the linear rising current waveform of Eq. (9) and the IEC current given in Eq. (1). Differences here are less than 25 %. Compared to the linear rising current waveform, the double exponential current waveform given by Eq. (11), however, produces maximum voltages about twice as high. This is due to the significantly higher maximum current steepness at t = inherent to a double exponential current waveform. 7.1 LPS type a consisting only of stretched wires 7 INDUCED VOLTAGE AND SEPARATION DISTANCE Table 5 contains the peak values of the induced voltages for the five examined structures of LPS type a consisting of only stretched wires. The corresponding separation distances are summarized in Table 6 for air (k m = 1). For all structures the corner strike represents the worst case, because it produces the highest peak values of the induced voltage in the corner. On the other hand, the induced voltages in the center wire are also very high in case of the center strike. Table 5: Induced voltages for LPS type a (k m = 1) Structure Peak voltage [kv] Induction Base area s Lightning attachment point Height Corner Side Center 2 m x 2 m Corner m Center wire m x 2 m Corner m Center wire m x 2 m Corner m Center wire m x 2 m Corner m Center wire m x 6 m Corner m Center wire Table 6: Separation distance for LPS type a (k m = 1) Structure Separation distance [cm] Induction Base area Loop Lightning attachment point Height Corner Side Center 2 m x 2 m Corner 29 9,6 7,3 1 m Center wire 6,1 8, m x 2 m Corner m Center wire m x 2 m Corner m Center wire m x 2 m Corner m Center wire m x 6 m Corner 28 2,1 3,2 1 m Center wire 2,5 3,

10 In the following, the CONCEPT calculation results are compared to the calculation results according to IEC , clause 6.3 and annex C (Eq. (1,3) in section 1). Fig. 13 illustrates the separation distances for the cases of corner strike and center strike. For the corner strike, the calculations with CONCEPT reveal separation distances being about 15 % to 3 % higher compared to IEC The deviations are due to the fact that in IEC the k i - values were reduced by 2 % compared to the original calculation results [3-5]. s [cm] Center strike Concept VDE 185 CONCEPT IEC m x 6m 1m 2m 4m 6m 1m s [cm] Corner strike Concept VDE 185 CONCEPT IEC m x 6m 1m 2m 4m 6m 1m Fig. 13 Comparison of the separation distances resulting from the CONCEPT computer code and from IEC (eq. (3)) for the cases of corner strike and center strike to LPS type a For the center strike, the deviations are much higher between the CONCEPT computation and IEC : The values determined according to the formula of IEC are higher by a factor of about 1,5 2,5. These high separation distances are due to the modified calculation method of IEC Whereas originally [3,4] only the height between the equipotential plane bonding point and the location of the proximity is considered, according to IEC now the total length of the path along the air termination and the down s is to consider for the length l in Eq. (1). 7.2 LPS type b with flat metal roof Table 7 summarizes the separation distances for the LPS type b using flat metal roofs (k m = 1). For the structure with 2 m x 2 m of base area, the separation distances are fairly independent on the location of the striking points and of the induction. On contrast, the structure with the large base area of 6 m x 6 m has a somewhat different trend. Anyhow, the highest value of the separation distance (6,1 cm) is very low and 3 % less compared to the 2 m x 2 m structure of the same height of 1 m. Table 7: Separation distance for LPS type b (k m = 1) Structure Separation distance [cm] Induction Base area Lightning attachment point Height Corner Side Center 2 m x 2 m Corner 8,4 8,1 7,8 1 m Center wire 7,6 7,8 7,8 2 m x 2 m Corner m Center wire m x 2 m Corner m Center wire m x 2 m Corner m Center wire m x 6 m Corner 6,1 2,8 2,2 1 m Center wire 2,4 2,6 2,6 1

11 Compared to LPS type a, the use of a metal roof reduces the separation distances significantly, especially for low structures of 1 m or 2 m of height. For these low structures the reduction factors vary from about 2 to 4. For higher structures the reduction is less pronounced. Fig. 14 shows the separation distances for the structures with 2 m x 2 m base area as function of the structure height. The following linear relation is found for the separation distance s and the structure height h (for a class II LPS): s = k h, 83 h (12) An approach for structures with a metal roof and classical down s with rounded up figures, also allowing for some safety margin, could be: h s = k (13) k m The constants k could be,12 for a class I LPS,,9 for a class II LPS and,6 for a class III/IV LPS. The coefficient k m denotes the material factor as in Eq. (1). s [cm] h [m] Fig. 14 Separation distance s for the 2 m x 2 m base area structures of LPS type b (metal roof) as function of the structure height h. 7.3 LPS type c with metal walls The LPS type c uses metal walls as down system and meshed metal wires for the air termination system. The calculations revealed induced voltages typically strongly oscillating. Because the constant-area-criterion given by Eq. (4-6) applies only for unipolar impulse voltages, the separation distance cannot be specified in detail. Therefore, only the peak values of the induced voltages are summarized in Table 8. For the induced voltages the following two cases are found distinguishing between the locations of the lightning attachment point: 1. In case of the corner and side strike, the attachment point is directly at the top of the metal walls (see Fig. 8). In these cases the induced voltages are always less than 5 kv, irrespective of the structure height. Due to these low voltage level, a separation distance of s 15 cm seems to be adequate concerning the class II lightning protection system according to IEC In the case of center strike, the peak values of the induced voltages (in the center wire) are close to the values found for LPS type a, seen in Table 5. Obviously, in case of the center strike use of metal walls does not reduce the separation distances significantly. The values given in Table 8 are based on ideal metal walls directly connected to the equipotential plane. In reality, however, short metal wires are used to connect the metal facades to the earth termination system. The influence of 1 m long connecting wires is examined for the structure with the base area of 2 m x 2 m and the height of 1 m. The interspacing between the connecting wires is taken into account with 5 m and 1 m, respectively. Table 9 summarizes the results in comparison to the ideal case of the walls directly connected to the equipotential plane. Only the values of the worst-case combinations corner strike/corner and center strike/center wire are given. The results reveal that the use of such connecting wires increases the induced voltages only marginally by less than 1 %. 11

12 Structure Table 8: Induced voltage for LPS type c (k m = 1) Peak voltage [kv] Base area Induction Lightning attachment point Height Corner Side Center 2 m x 2 m Corner m Center wire m x 2 m Corner m Center wire m x 6 m Corner m Center wire Table 9: Influence of the wires connecting the metal walls to the equipotential plane on the induced voltages for LPS type c (k m = 1). The connecting wires are 1 m long and the 1 m high structure has the base area of 2 m x 2 m. Interspacing between the connecting wires Induction Peak voltage [kv] Lightning attachment point Corner Center Walls directly connected to Corner ground plane Center wire m - Interspacing 1 m - Interspacing Corner Center wire - 17 Corner Center wire LPS type d with circumferential metal attic at the roof The influence of the circumferential metal attic (see Fig. 9) is examined for the structure with the dimensions of 2 m x 2 m x 2 m and for two flat attic sheets having a width of,5 m and 1 m. Table 1 summarizes the resulting separation distances in comparison to the LPS type a configuration without attic. For the cases of side strike and center strike no significant reduction of the separation distance is found, irrespective of the sheet width. Only for the worst-case combination of corner strike/corner the separation distance is reduced by about 3 % for the sheets width of,5 m and by about 4 % for the sheets width of 1 m. Table 1: Separation distance for LPS type d (k m = 1) with the base area of 2 m x 2 m and the height of 2 m circumferential metal attic at the roof Without attic,5 m 1, m Separation distance [cm] Induction Lightning attachment point Corner Side Center Corner Center wire Corner Center wire Corner Center wire

13 8 CONCLUSION Extensive sheet metal plates used as natural components of lightning protection systems may reduce the separation distances considerably. In the present paper the influence of metal roofs, of metal facades and of metal circumferential attics is analyzed. The calculations are based on the CONCEPT II computer code which solves the Maxwell s Equations with the Method of Moments in the frequency domain. The use of metal roofs can reduce the separation distances significantly by the factor of about Here, the separation distance is proportional to the height of the proximity over the equipotential plane. For metal walls two different cases are found depending on the location of the point of strike: If the point strike is at the top of the metal wall as is the case of the corner strike and the side strike, the separation distances are relatively small. For instance, about 15 cm are sufficient to keep the required separation distances for class II lightning protection system according to IEC In case of center strike, however, the use of a metal facade gives only marginal reduction of the separation distances. In this case the induction resulting from the air termination wires on the roof dominate. The use of a circumferential metal attic is of minor influence, whereat the separation distance is reduced by some tens of percents, at best. 9 REFERENCES [1] IEC :26-1, Protection against lightning - Part 1: General principles, Jan. 26. [2] IEC :26-1, Protection against lightning-part 3: Physical damage to structures and life hazard, Jan. 26. [3] O. Beierl, H. Steinbigler, "Induced over-voltages at lightning protection systems with meshed air termination s, Proc. of the 18 th International Conference on Lightning Protection ICLP, Munich, paper 4.1, Sep (in German). [4] W. Zischank, Isolated lightning protection systems for buildings with flammable content, Proc. of the 19 th International Conference on Lightning Protection ICLP, paper 6.8, Graz, April 1988 (in German). [5] IEC , Protection of structures against lightning, Ed. 2., [6] D. Kind, The constant-area-criterion for impuls voltages at electrodes in air, Ph. D. Thesis, Technical University Munich, 1957 (in German) [7] L. Thione, The Dielectric Strength of Large Air Insulation in K. Ragaller: Surges in High-Voltage Networks, Plenum Press, New York, 198. [8] H. Bruens, Pulse generated electromagnetic response in three-dimensional wire structures, Ph. D. Thesis, University of the Federal Armed Forces Hamburg, Germany, 1985 (in German). [9] R.F. Harrington, Field Calculations by Moment Methods, New York, The MacMillan Company, [1]H. Bruens, D. Koenigstein, Calculation and measurements of transient electromagnetic fields in EMP simulators, Proc. of the 6th Symposium on Electromagnetic Compati-bility, Zurich, paper 66L2, pp , March [11]H. Singer, H. Brüns, T. Mader, A. Freiberg, CONCEPT II Manual of the program system, University Hamburg-Harburg, Germany, 23. [12]M.A. Uman, R.D. Brantley, Y.T. Lin, J.A. Tiller, E.P. Krider, D.K. McLain, Correlated electric and magnetic fields from lightning return strokes, J. Geophysical. Res., vol. 8, pp , Jan [13]H. Bruens, H. Singer, F. Demmel, Calculation of transient processes at direct lightning stroke into thin wire structures, Proc. of the 7th Symposium on Electromagnetic Compatibility, Zurich, paper 17D5, pp. 85-9, March [14]W. Zischank, F. Heidler, Reduction of separation distances by using extensive metal parts as natural components of the external lightning protection system, Proc. of the 29 th International Conference on Lightning Protection ICLP, Uppsala, Sweden, paper 1-1, June 28. [15]F. Heidler, W. Zischank, A. Kern, Analysis of necessary separation distances for lightning protection systems including natural components, Proc. of the 28th Intern. Conf. on Lightning Protection ICLP, Kanazawa, Japan, vol. II, report X-1, pp , Sep