Numerical computation of lightning transfer functions for layered, anisotropically conducting shielding structures by the method of moments

Abstract

A formalism for the computation of lightning transfer functions by the method of moments, which involves shielding structures that may consist of layered, anisotropically conducting composite materials, is presented in this contribution. The composite materials, being of a type that is widely used in space- and aircraft design, are electrically characterized by an equivalent conductivity. As basis for the quantitative analysis the method of moments is used where shielding surfaces can be treated by a thin layer technique which utilizes analytical solutions inside the layer. Also the effect of an extended lightning channel can be taken into account. The method is applied to geometries that resemble an actual airplane fuselage.

Adv. Radio Sci., 14, 107114, 2016 www.adv-radio-sci.net/14/107/2016/ doi:10.5194/ars-14-107-2016 Author(s) 2016. CC Attribution 3.0 License.

Fabian Happ, Heinz-D. Brns, Gazmend Mavraj, and Frank Gronwald

Institute of Electromagnetic Theory, Hamburg University of Technology (TUHH), Hamburg, Germany

Correspondence to: Fabian Happ ([email protected])

Received: 21 January 2016 Revised: 25 May 2016 Accepted: 6 June 2016 Published: 28 September 2016

Abstract. A formalism for the computation of lightning transfer functions by the method of moments, which involves shielding structures that may consist of layered, anisotropically conducting composite materials, is presented in this contribution. The composite materials, being of a type that is widely used in space- and aircraft design, are electrically characterized by an equivalent conductivity. As basis for the quantitative analysis the method of moments is used where shielding surfaces can be treated by a thin layer technique which utilizes analytical solutions inside the layer. Also the effect of an extended lightning channel can be taken into account. The method is applied to geometries that resemble an actual airplane fuselage.

1 Introduction

Despite the constantly ongoing progress in the development and application of numerical methods in electromagnetics it turns out that the computation of lightning-related effects in the framework of Electromagnetic Compatibility (EMC) still constitutes a highly challenging task. This is due to a number of difculties that can be characterized as follows (Apra et al., 2008; Anatzki and Gronwald, 2012; Prost et al., 2013): First, the modeling of an actual lightning channel as electromagnetic source requires to turn a physically complicated and geometrically extended excitation into a numerical model. Second, the rather long duration of a lightning electromagnetic pulse (LEMP), together with its associated low-frequency spectrum, requires both in time and frequency domain stable and efcient numerical algorithms. Third, in actual applications it is often necessary to calculate LEMP

Numerical computation of lightning transfer functions for layered, anisotropically conducting shielding structures by the method of moments

transfer functions of complex systems, such as aircraft, for example. This, in turn, often involves to model advanced materials and to deal with a high degree of complexity.

In this contribution it is outlined how to numerically calculate LEMP transfer functions by the method of moments (MoM), taking into account the main difculties mentioned above. To this end, in Sect. 2 it is outlined, based on early work (Brns, 1985), how to take into account the inuence of the lightning channel in a realistic way. The modeling of composite materials as an example for an advanced material is explained in Sect. 3. It is also mentioned that low-frequency shielding can be modeled by a thin layer technique (Brger et al., 1995). Then, in Sect. 4, it is summarized how to process frequency domain results obtained by the MoM in order to obtain time domain results that are useful for lightning analysis. Simulation examples that implement these methods are given in Sect. 5, followed by a short summary in Sect. 6.

2 Numerical modeling of the lightning channel

There are several engineering approaches for the modeling of the electric current of a lightning channel (Rakov and Uman, 2003). They relate the current at the channel base to the current at any position on the channel, where the channel itself can be modeled as a wire of length of several kilometers. Most models are categorized to be either of the Transmission Line (TL) or of the Traveling Current Source (TCS) type. The TL type is used here, therefore it shortly is described in the following subsection. More details on its implementation within the MoM can be found in (Brns, 1985).

Published by Copernicus Publications on behalf of the URSI Landesausschuss in der Bundesrepublik Deutschland e.V.

108 F. Happ et al.: Numerical computation of lightning transfer functions

z [prime]

100

100

i(t)[kA]

i(t)[kA]

50

50

z[prime] = vf t

0 0 200 400

0 0 1 2 3 4

Time [s]

Time [s]

Figure 2. Lightning current pulse according to Heidler.

3 Modeling of composite material

Composite materials are widely used in aircraft design for weight reduction and improvement of mechanical strength. In this publication a type of carbon ber composite is considered which consists of several layers of carbon bers with different orientations, enclosed in resin.

3.1 Equivalent conductivity

Instead of modeling single bers one may assume an anisotropic conductivity for each layer (Holloway et al., 2005; Happ et al., 2013), where the direction of the bers in the layer corresponds to the direction of the highest conductivity in the conductivity tensor.

A further simplication can be made by replacing the various anisotropically conducting layers by a single layer with the same overall thickness and an equivalent isotropic conductivity eq. An analytical investigation of the shielding effectiveness of a plane shield with several anisotropically conducting layers has shown that this simplication is valid up to several tens of MHz (Happ et al., 2015). Since lightning is a low frequency phenomenon, the equivalent conductivity can be used to model the shielding properties of the composite material with respect to lightning effects. As a rst step to calculate eq the diagonal entries of the average conductivity tensor are calculated by

xx

yy

[parenrightbigg] =1d

L

Xi=1

[bracketleftbigg][parenleftbigg]

i(z[prime],t)

z[prime] = 0

Figure 1. Illustration of the upward moving current front in the lightning channel.

2.1 The transmission line model

In the TL type model (Uman and McLain, 1969) a current wave is traveling upwards with propagation speed vf , starting at the channel base, as depicted in Fig. 1. At positions above z[prime] = vf [notdef] t the current is zero. The current at any posi

tion z[prime] along the channel and at time t is given by

i(z[prime],t) = [braceleftBigg]i

0(t z[prime]vf ) if z[prime] vf [notdef] t 0 if z[prime] > vf [notdef] t

, (1)

where i0(t) is the current pulse at the channel base and vf is the propagation speed, which typically is chosen to be one third of the vacuum speed of light vf = 13 [notdef] c.

2.2 The current pulse

For the current i0(t) the current pulse according to (Heidler, 1985), which can be found in the VG 95371-10 standard, is used. The slope of this curve is continuous, which has proven to be consistent with the results of lightning measurements. The entire pulse can be divided into three components that are multiplied by each other and determine the amplitude, the smooth rising edge, and the exponential decay, respectively. The corresponding equation is given by

i0(t) =

imax

cos2( i)sin2( i)

[parenrightbigg][bracketrightbigg] [notdef]

di, (3)

where L is the number of layers. Here it is assumed that each layer of bers has the conductivity in ber direction and in cross-ber direction. The orientation of the bers is specied by the angle i. The thickness of layer i is denoted by di and the total thickness of the multilayer material is denoted by d. Then the equivalent conductivity is chosen to be the smaller of the two conductivity values,

eq = min(xx,yy), (4)

to give a valid value for a worst case approximation. In the following, the conductivities are chosen to be =

10 kS m1 and = 100 S m1 and each layer has thickness

Adv. Radio Sci., 14, 107114, 2016 www.adv-radio-sci.net/14/107/2016/

[notdef]

t Trise

n

[notdef]

[parenleftBig]

[notdef] exp

t Thold

, (2)

where imax is the maximum value of the current, is the amplitude correction factor, which ensures that the pulse reaches the value imax, Trise is the rise time, and Thold the decay time of the pulse. For the following investigations the parameters imax = 100 kA, = 0.986, Trise = 1.82 s, and Thold =

285 s are used. This choice refers to a negative rst stroke of threat level severe. A pulse with these parameter values is shown in Fig. 2.

1 +

[parenleftBig]

t Trise

n

F. Happ et al.: Numerical computation of lightning transfer functions 109

Analytical formulation Layer

Figure 4. Two MoM regions that are separated by a thin nitely conducting layer. Its electromagnetic properties can be described by an analytical formulation.

4 Time domain simulations with a frequency domain solver

A priori, the lightning current is formulated in time domain while the MoM and the layer technique are formulated in frequency domain. In this section it is explained how to relate both formulations to calculate the impulse response of a lightning current.

The spectrum I0(!) of the current i0(t) at the channel base can be determined analytically, as described in (Andreotti et al., 2005). The spatial distribution of the current at any position z[prime] of the channel is a time shifted version of i0(t), therefore the shift property of the Fourier transformation can be used to write the corresponding spectrum as

I (z[prime],!) = I0(!) [notdef] exp[parenleftbigg]

j! z[prime]

vf

Figure 3. Local coordinate system for one layer of carbon bers, where the -direction corresponds to the orientation of the bers.

di = d/L. For a symmetric layer pattern, e.g., a multiple of

four layers with a relative rotation angle of 45 between the bers of adjacent layers, the equivalent isotropic conductivity is given by eq = 5050 S m1.

3.2 Thin layer technique

There are several methods to numerically model thin layers. The surface impedance boundary condition method (Harrington and Mautz, 1975; Chiang and Chew, 2006), for example, is applicable to scattering problems, but it is not suitable for the calculation of shielding effectiveness, since the coupling between inside and outside region of the shielding geometry has to accurately be taken into account. It might also come to mind to apply a Greens function of layered media (Michalski and Mosig, 1997; Ginste et al., 2010), but this typically requires some kind of canonical symmetry of the geometry considered. Therefore it is not immediate to apply this method to arbitrarily shaped three dimensional structures. The thin layer technique that is applied here to efciently model thin layers of nite size in conjunction with the MoM (Brger et al., 1995) is based on an analytically formulated coupling matrix. In this case, the layer has to fulll the requirement to be thin compared to the overall dimension of the body to be modeled and it has to provide a sufciently high conductivity. If the conductivity is large enough then the wave propagation inside the layer is perpendicular to its surface and can be described by an analytical solution in the form of a coupling matrix which also correctly incorporates all effects related to the wave propagation in a lossy medium. As a consequence, two regions that are separated by a two-dimensional layer, compare Fig. 4, can be treated by the MoM, where the coupling through the layer is taken into account by a coupling matrix which relates the tangential elds at both sides of the layer to each other. This hybrid-technique, which combines the MoM with an analytical solution, has proven to be stable down to frequencies in the kHz range (Happ et al., 2014).

MoM

[parenrightbigg]

, (5)

which clearly states that the magnitude of the spectrum does not depend on the position and only the phase is inuenced by z[prime]. Now the system response G(!) to a unit excitation, in this case an impressed current with a constant amplitude with respect to frequency, which ows spatially distributed on the channel, can be calculated. Then both spectra can be multiplied in frequency domain to nd the spectrum of the pulse response R(!),

R(!) = G(!) [notdef] I0(!), (6) which could, e.g., be a voltage or a eld value. The function R(!) can be transformed to time domain via an inverse Fourier transformation to yield the pulse response r(t).

The calculated spectrum consists of values that refer to discrete frequencies. This leads to a periodic extension of the pulse response in time domain. The time tex, when the periodic extension starts, is related to the frequency difference [Delta1]f between the discrete frequencies of the spectrum by

tex =

1[Delta1]f . (7)

To ensure meaningful simulation results, [Delta1]f has to be small enough such that the impulse response has decayed before

www.adv-radio-sci.net/14/107/2016/ Adv. Radio Sci., 14, 107114, 2016

110 F. Happ et al.: Numerical computation of lightning transfer functions

the periodic extension starts. This implies that a long excitation pulse requires a small frequency step width, which sets up a limit for this method, due to the fact that the MoM becomes unstable at very low frequencies. Hence, excitation signals with a duration of several milliseconds cannot accurately be modeled by this method. For the pulse in Fig. 2 a frequency step width of [Delta1]f = 600 Hz, which corresponds

to the time tex = 1.666 ms, is used for the simulations in the

next section.

Another important point is the maximum frequency that has to be considered to reproduce the steep rise of the pulse.In case of a lightning pulse with a rise time of a few microseconds a maximum frequency in the range of a few MHz is sufcient.

Finally, it should be mentioned that this is a linear formulation, hence non-linear effects, which could result from matter interaction at very high eld magnitudes, are not taken into account.

5 Simulation examples

In this section the proposed formalism is illustrated by means of several examples. In all cases the pulse introduced in Sect. 2.2 is used as excitation. In the case of a nearby lightning the channel starts at (x,y,z) = (0,0,0) on a perfectly

conducting ground and ends at (0, 0, l), where in our case the channel length l is chosen to be of length 3 km. As depicted in Fig. 1, the current pulse starts at the channel base an moves upwards with speed vf .

The rst conguration to be considered is a single lightning channel without a neighboring structure. Then a transmission line structure, which is loaded by 50[Omega1] resistors, is placed close to the lightning channel. In a next step, this transmission line structure is located inside a cylindrical cavity. The geometrical details of the resulting setup are shown in Fig. 5. The cylinder is chosen either as a closed one with conductivity = 5 kS m1, which is a realistic equivalent

conductivity for a carbon ber composite, or a perfectly electrically conducting (PEC) cylinder with 12 apertures on both sides, representing a fuselage of a passenger aircraft. As excitation a lightning channel close to the structure or a lightning channel directly attached to the structure is assumed. The resulting four different cases are summarized in Fig. 6, where only short sections of the lightning channels are shown. In the case of a direct strike the impressed current ows from the top side of the structure upwards along the lightning channel. To model the discharge along the fuselage and towards the ground a second wire connects the bottom side of the cylinder to the PEC ground. The current on this second wire is not an impressed one, as the one on the long wire that models the lightning channel, it rather results from the MoM simulation.

3 m

10 m

25 m

2.5

m

5 m

z

100 m

30 m

x

Figure 5. Dimensions and positions of the transmission line structure and its surrounding cylinder, representing a simple model of an aircraft fuselage.

I

(a)

I

(b)

I

(c)

I

(d)

Figure 6. The considered simulation models where the cylindrical cavity is positioned at a height of 100 m above perfectly conducting ground. Four cases are considered: closed nitely conducting cylinder 10 m distant from the lightning channel (a), closed nitely conducting cylinder subject to a direct lightning strike (b), PEC cylinder with apertures 10 m distant from the lightning channel (c), and PEC cylinder with apertures subject to a direct lightning strike (d).

5.1 Field of the lightning channel

As a prerequisite, in this subsection the electric and magnetic elds of the lightning channel without a neighboring structure are investigated and compared to the semi-analytical formulas derived in (Uman et al., 1975). With these formulas the elds at any position on the ground at z = 0 can be calculated

as a function of the horizontal distance to the channel.

Adv. Radio Sci., 14, 107114, 2016 www.adv-radio-sci.net/14/107/2016/

F. Happ et al.: Numerical computation of lightning transfer functions 111

8 0 20 40 60 80 100

25 m 50 m 100 m Uman 25 m Uman 50 m Uman 100 m

Figure 7. Electric eld of the channel at different distances from the channel at z = 0, calculated by the numerical method presented

in this paper and a semi-analytical formula taken from (Uman et al., 1975).

25 m 50 m 100 m Uman 25 m Uman 50 m Uman 100 m

Figure 9. Electric eld of the channel at different distances from the channel at z = 0, calculated for a longer time interval.

105

105

0

0

2

2

E z(t)[V/m]

E z(t)[V/m]

4

4

6

6

8 0 200 400 600 800 1000

Time [s]

Time [s]

4 105

25 m 50 m 100 m Uman 25 m Uman 50 m Uman 100 m

Figure 8. Magnetic eld of the channel at different distances from the channel at z = 0, calculated by the numerical method presented

in this paper and a semi-analytical formula taken from (Uman et al., 1975).

The corresponding curves for the electric and magnetic elds are shown in Figs. 7 and 8, respectively. The numerical results obtained from the MoM formalism are in excellent agreement with the semi-analytical results, as exemplied by three observation points at distances 25, 50 and 100 m. For these distances the effect of the nite length of the lightning channel in the used model is negligible.

In Fig. 9 the same curves for the electric eld as in Fig. 7 are show up to a time of t = 1 ms. It can be observed that even

for this longer time interval the results of the MoM formalism

E(t)[V/m]

2 105

Ex

Ey

Ez

600

0 100

H y(t)[A/m]

400

2 105

0 20 40 60 80 100

Time [s]

(a)

200

15

0

0 20 40 60 80 100

10

E(t)[V/m]

Time [s]

Ex

Ey

Ez

5

0

0 2 4 6 8 10

Time [s]

(b)

Figure 10. Electric eld at the center of the transmission line without (a) and with (b) nitely conducting cylinder. The different time and eld scales should be noted for comparison.

are still in very good agreement with the results of the semi-analytical method.

www.adv-radio-sci.net/14/107/2016/ Adv. Radio Sci., 14, 107114, 2016

112 F. Happ et al.: Numerical computation of lightning transfer functions

0 20 40 60 80 100

0 100

600

5 103

400

H y(t)[A/m]

u(t)[V]

1 104

Transmission

line Cond.

cylinder

200

Transmission

line Cond. cylinder

1.5 104

0

0 2 4 6 8 10

Time [s]

Time [s]

(a)

Figure 11. Magnetic eld at the center of the transmission line, 10 m distant from the lightning channel, with and without conducting cylinder.

1.5

PEC cylinder

1

100

u(t)[V]

0.5

101

102

0

103

G(!)[ ]

0 2 4 6 8 10

104

Time [s]

(b)

105

Figure 13. Voltage at the termination resistor of the transmission line with and without conducting cylinder (a) and with PEC cylinder with apertures (b) for the structures being close to the lightning channel.

an impressed current of constant amplitude of 1A per frequency, is shown in Fig. 12. This transfer function includes the following effects: the creation of the elds of the lightning channel, the coupling into the cylindrical cavity by diffusion or aperture coupling, eld distribution inside the cavity, and coupling into the transmission line structure. At low frequencies the curves for the case of the transmission line alone and the transmission line inside the nitely conducting cylinder are on top of each other. This shows that in this frequency region the presence of the conducting cylinder is hardly relevant. The magnitude of the transfer function for the case of the PEC cylinder with apertures is lower compared to the other two cases because the magnetic eld may only couple through the apertures into the fuselage. This effect is small for the considered wavelengths which are much larger than the aperture dimensions.

Finally, the time-domain voltage at the terminating resistor is shown in Fig. 13 as a response to the current pulse according to Fig. 2. A delay time of 1 s can be observed, which corresponds to the time the pulse needs to travel from the ground to the height of 100 m where the structure is located. The maximum voltage for the cases of no cylinder, nitely con-

Adv. Radio Sci., 14, 107114, 2016 www.adv-radio-sci.net/14/107/2016/

106

7

10 103 102 101 100

Frequency [MHz]

Figure 12. Frequency spectrum of the system response for a lightning channel close to the structure without cylinder, with nitely conducting cylinder, and with PEC cylinder with apertures.

5.2 Structure close to the lightning channel

The electric eld at the center of the transmission line structure, which corresponds to the point (25, 0, 100) m, with and without the nitely conducting cylinder as shield is illustrated in Fig. 10. In both cases the y component of the eld is zero. As expected, the amplitudes of the electric eld in the presence of the shield are much smaller if compared to the situation without shield.

The magnetic elds for these two cases are plotted in Fig. 11. The rise time of the magnetic eld inside the nitely conducting cylinder is much larger if compared to the rise time of the magnetic eld without the cylinder. This is due to the low pass behavior of the shield with respect to the magnetic eld, i.e., the high frequency components of the eld are attenuated and therefore the rise time is increased.

The related transfer function G(!), as dened by the ratio between the voltage U(!) at the termination resistance and

F. Happ et al.: Numerical computation of lightning transfer functions 113

10 102 101 0.372 100

101

102

G(!)[ ]

103

104

5

Frequency [MHz]

Figure 14. Frequency spectrum of the system response for the cases of a closed conductive cylinder and a PEC cylinder with apertures if a direct lightning strike is applied.

200

Cond.

cylinder PEC cylinder

100

u(t)[V]

0

100

200

0 5 10 15 20 25 30

Time [s]

Figure 15. Voltage at the termination resistor of the wire loop inside the conductive cylinder and the PEC cylinder with apertures in case of a direct lightning strike.

ducting cylinder, and PEC cylinder with apertures are 17 kV, 800 V, and 1.5 V, respectively. The main contribution to the voltage is due to the magnetic eld. Due to the low pass characteristic of the nitely conducting cylinder the voltage has a decreased rise time, similar to the magnetic eld itself.

5.3 Direct lightning strike

In this subsection cases are considered where the lightning channel is directly attached to the structure, compare Fig. 6b and d, such that the lightning current ows directly on the surface of the cylindrical structure.

In Fig. 14 the spectrum of the transfer functions for both the nitely conducting cylinder and the PEC cylinder with apertures is shown. Two maximum values can be identied in the considered frequency range. The rst maximum occurs at a frequency of 372 kHz. This frequency is too low for being a resonance frequency of the cylinder. Analysis shows that this frequency is related to the wire that connects the cylinder to

the ground. More precisely, the frequency turns out to be the antiresonance frequency of the wire (Schelkunoff and Friis, 1952), i.e., the imaginary part of the impedance of the wire as seen from the current source is zero and the real part of the impedance exhibits a maximum. The second maximum of the transfer function occurs at the frequency 1.6 MHz, where the length of the wire from the cylinder to the ground equals one half of a wavelength.

The time domain response of the voltages are shown in Fig. 15. Oscillations are clearly visible, where the time period of the oscillation corresponds to the frequency where the rst maximum of the transfer function occurs. For the case of the nitely conducting cylinder the voltage is considerably higher if compared to the case of the PEC cylinder. Therefore, in this example, the diffusion coupling is larger if compared to the aperture coupling, at least in the considered frequency range which is relevant for lightning analysis.

6 Conclusions

A formalism for the calculation of LEMP transfer functions by means of the MoM has been proposed. Composite materials are modeled by an equivalent conductivity which is applicable for the frequency range of the lightning spectrum. In the examples considered it turned out that diffusion coupling has a larger inuence on the transfer function than the aperture coupling. Therefore, the conductivity and thickness of the shell of the fuselage are very important parameters and it clearly is not sufcient to approximate composite materials by PEC material for lightning analysis.

Acknowledgements. This research was supported in part by the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG).

Edited by: F. SabathReviewed by: R. Bunger and two anonymous referees

References

Anatzki, J. and Gronwald, F.: Comparison of time and frequency domain methods for indirect lightning effect computation: A case study, Proceedings ESA Workshop on Aerospace EMC 2012, 1 5, 2012.

Andreotti, A., Falco, S., and Verolino, L.: Some integrals involving

Heidlers lightning return stroke current expression, Electr. Eng., 87, 121128, 2005.

Apra, M., DAmore, M., Gigliotti, K., Sarto, M., and Volpi, V.:

Lightning Indirect Effects Certication of a Transport Aircraft by Numerical Simulation, IEEE Transactions on Electromagnetic Compatibility, 50, 513523, 2008.

Brns, H.-D.: Einkopplung von Blitzfeldern in Stabstrukturen, International Conference on Lightning Protection, Munich, Germany, 1620 September 1985, 227233, 1985.

www.adv-radio-sci.net/14/107/2016/ Adv. Radio Sci., 14, 107114, 2016

114 F. Happ et al.: Numerical computation of lightning transfer functions

Brger, G., Brns, H.-D., and Singer, H.: Simulation of Thin Layers in the Method of Moments, International Zurich Symposium on EMC, Zurich, Switzerland, 79 March 1995, 339344, 1995.

Chiang, I.-T. and Chew, W. C.: Thin dielectric sheet simulation by surface integral equation using modied RWG and pulse bases, IEEE Transactions on Antennas and Propagation, 54, 1927 1934, 2006.

Ginste, D. V., Rogier, H., and Zutter, D. D.: Perfectly Matched Layer based modelling of layered media: Overview and perspective, 2010 URSI International Symposium on Electromagnetic Theory (EMTS), Berlin, Germany, 1619 August 2010, 803 806, 2010.

Happ, F., Schrder, A., Brns, H.-D., and Gronwald, F.: A method for the calculation of electromagnetic elds in the presence of thin anisotropic conductive layers using the method of moments, 2013 International Symposium on Electromagnetic Compatibility (EMC EUROPE), Brugge, Belgium, 26 September 2013, 579582, 2013.

Happ, F., Brns, H.-D., and Gronwald, F.: Benchmark calculations of magnetic shielding at low frequencies, 2014 International Symposium on Electromagnetic Compatibility (EMC Europe), Gothenburg, Sweden, 14 September 2014, 468473, 2014.Happ, F., Gronwald, F., and Brns, H.-D.: An extension of

Schelkunoffs shielding theory to anisotropic conducting multilayer materials, 2015 IEEE International Symposium on Electromagnetic Compatibility (EMC), Dresden, Germany, 1622 August 2015, 15421547, 2015.

Harrington, R. and Mautz, J.: An impedance sheet approximation for thin dielectric shells, IEEE T. Antenn. Propag., 23, 531534, 1975.

Heidler, F.: Analytische Blitzstromfunktion zur LEMP

Berechnung, International Conference on Lightning Protection, Munich, Germany, 1620 September 1985, 6366, 1985. Holloway, C., Sarto, M., and Johansson, M.: Analyzing carbon-ber composite materials with equivalent-Layer models, IEEET. Electromagn. C., 47, 833844, 2005.

Michalski, K. A. and Mosig, J. R.: Multilayered media Greens functions in integral equation formulations, IEEE T. Antenn. Propag., 45, 508519, 1997.

Prost, D., Issac, F., Volpert, T., Quenum, W., and Parmantier, J.-P.: Lightning-Induced Current Simulation Using RL Equivalent Circuit: Application to an Aircraft Subsystem Design, IEEE T. Electromagn. C., 55, 378384, 2013.

Rakov, V. A. and Uman, M. A.: Lightning Physics and Effects,Cambridge University Press, Cambridge, 2003.

Schelkunoff, S. A. and Friis, H. T.: Antennas Theory and Practice,John Wiley & Sons, New York, 1952.

Uman, M. A. and McLain, D. K.: Magnetic eld of lightning return stroke, J. Geophys. Res., 74, 68996910, 1969.

Uman, M. A., McLain, D. K., and Krider, E. P.: The Electromagnetic Radiation from a Finite Antenna, Am. J. Phys., 43, 3338, 1975.

VG 95371-10: Electromagnetic compatibility (EMC) including electromagnetic pulse (EMP) and lightning protection fundamentals Part 10: Threat levels for NEMP and lightning, 2011.

Adv. Radio Sci., 14, 107114, 2016 www.adv-radio-sci.net/14/107/2016/