doi: 10.1016/j.epsr.2015.11.039 persistent link using

Accepted version of paper published by Elsevier, Electric Power Systems Research.

Citation:

Y. Méndez Hernández, T. E. Tsovilis, F. Asimakopoulou, Z. Politis, W. Barton, and M.

Martínez Lozano, “A simulation approach on rotor blade electrostatic charging and its effect

on the lightning overvoltages in wind parks,” Electric Power Systems Research, vol. 139

(SI), pp. 22-31, Oct. 2016.

DOI: 10.1016/j.epsr.2015.11.039

https://doi.org/10.1016/j.epsr.2015.11.039

Please cite this article in press as: Y.M. Hernández, et al., A simulation approach on rotor blade electrostatic charging and its effect onthe lightning overvoltages in wind parks, Electr. Power Syst. Res. (2015), http://dx.doi.org/10.1016/j.epsr.2015.11.039

ARTICLE IN PRESSG ModelEPSR-4509; No. of Pages 10

Electric Power Systems Research xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Electric Power Systems Research

j o ur na l ho mepage: www.elsev ier .com/ locate /epsr

A simulation approach on rotor blade electrostatic charging and itseffect on the lightning overvoltages in wind parks

Y. Méndez Hernándeza,b,∗, T.E. Tsovilis c, F. Asimakopoulouc, Z. Politis c, W. Bartond,M. Martínez Lozanoe

a Engineering Department, Universidad Simón Bolívar, Caracas, Venezuelab R&D Department, Murayh Company, Munich, Germanyc R&D Department, Raycap SA, Athens, Greeced Electric Power Systems: Wind Power, General Electric Company, Salzbergen, Germanye Engineering Department, Universidad De La Salle Bajío, León, Mexico

a r t i c l e i n f o

Article history:Available online xxx

Keywords:Electrostatic chargingOvervoltageSurge protectionWind parksRotor bladesLightning

a b s t r a c t

This study is an attempt to present a theoretical approach of the hypothesis of electrostatic charging ofrotor blades during storms and its effects on the lightning protection and overvoltage protection systemsduring lightning strokes.

Overvoltage and overcurrent in form of surges may be the result of electromagnetic traveling wavescaused by lightning strokes, especially if the grounding (earthing) system of the wind turbines (WTs) isgalvanically connected between each other in order to ensure a common earth potential in the wind park(WP).

These effects may impose additional electrical issues to the low and medium voltage electrical instal-lation including the surge protection devices.

© 2015 Elsevier B.V. All rights reserved.

1. Introduction

The hypothesis of the triboelectric effect or electrostatic charg-ing caused by the friction between the isolating material of therotor blades (glass fiber, balsa wood, gluing material, etc.) and theair, especially during storms is the main focus of this study.

Moreover, induced or transferred charge during normal opera-tion and storms is a motivation to further explore this interaction.A simulation approach in electrostatic charging of rotor blades andits effect on the earth potential in Megawatt-class wind parks isproposed in this publication, as a first attempt to address this topic.

Before the occurrence of a lightning strike an increase in thevalue of the electric field is measured [1]. These observationsextended to wind turbines and wind parks support the hypothe-sis that, during storms, an electrostatic charging may take place onthe rotor blades and other components manufactured by compositematerials, such as, the nacelle.

∗ Corresponding author at: Universidad Simón Bolívar, High Voltage Lab,Sartenejas, Venezuela. Tel.: +58 2129014131.

E-mail addresses: [email protected], [email protected](Y.M. Hernández).

Fig. 1 shows an example of a continuous electric arc or dis-charge caused by electrostatic charging on a rotor blade root ofa wind turbine [2]; the image was taken during a stormy day andis part of a video that lasts several tens of seconds. Rotor bladesare mainly manufactured by non-conductive composite materials,with non-orthotropic dielectric properties and large dimensions(several meters of length) and weight (several tons), and thereforethe probability of electrostatic charging is not negligible.

With regards to the dimensions of modern Megawatt-class windturbines, unexpected effects in form of overvoltage and overcurrentmay be the result of electromagnetic traveling waves caused bylightning strikes; especially if the earthing (grounding) systems ofthe wind turbines (WTs) are connected to each other in order toensure a common earth potential in the wind park (WP) [3]. Theseeffects combined with electrostatic charging of rotor blades mayimpose additional requirements on the reliable operation of thewind turbine’s electrical system (low and medium voltage).

The simulation of electromagnetic transients combined with theone of electrostatic charging, such as lightning disturbances in windparks, shall include specific models to represent each componentwithin the wind turbine (WT) or wind generator (WG); this rep-resentation differs from other studies, such as stability and powerflow calculations, on the high frequency point of view.

http://dx.doi.org/10.1016/j.epsr.2015.11.0390378-7796/© 2015 Elsevier B.V. All rights reserved.

dx.doi.org/10.1016/j.epsr.2015.11.039

dx.doi.org/10.1016/j.epsr.2015.11.039

http://www.sciencedirect.com/science/journal/03787796

http://www.elsevier.com/locate/epsr

mailto:[email protected]

mailto:[email protected]

dx.doi.org/10.1016/j.epsr.2015.11.039



2 Y.M. Hernández et al. / Electric Power Systems Research xxx (2015) xxx–xxx

Fig. 1. Electric discharges on a rotor blade root [2].

hi

Down condductor: S

Down con ductor: T

i

Down con ductor: R

θ: Re f. angle

y

x

Fig. 2. Schematic diagram of the wind turbine incorporating symbols used in thecalculations.

2. Approach and modeling

The components of the wind park were modeled in form of surgeimpedances with a propagation velocity of the travelling waves,where, i is the index of the down conductor of the correspondingrotor blade and hi the height of the conductor above ground asshowed in Fig. 2 [10].

The other components of the wind turbines, such as rotor bladebearings, azimuth bearings, steel tower, earthing system, low (LV)and medium (MV) voltage distribution cables, electronic converter,

generator, distribution transformer and LV and MV MOVs were alsomodeled.

The time to front of the computed overvoltages across the windpark’s electrical installation exhibits values lower than that of theincoming lightning current (10 �s). Due to electromagnetic wavereflections in the wind park under study the significant (dominant)frequency for the calculation of electrical parameters of systemcomponents should be considerably higher than 25 kHz or at leastan order of magnitude higher. For this reason the guidelines forfast front transient calculation [21] suggest the use of a dominantfrequency of 500 kHz for calculating transmission line electricalparameters in lightning surge analysis. In light of the above andbased on several attempts of simulating the transient behaviorof the wind park’s complex system, the selection of 250 kHz as adominant frequency can be considered, in this study, as an accept-able compromise between the electrostatic charging and the surgepropagation modeling.

The topology of the wind turbine chosen for the calculations,which consists of a double-fed wounded-rotor induction machineor generator (DFIG) connected to the grid on the stator side and toan electronic converter on the rotor side, is depicted in Fig. 3.

2.1. Rotor blades, bearings and tower

The rotor blade lightning protection system (LPS) chosen forthe model was the receptor-based lightning protection systemexplained in [3,6,7]. It was modeled by means of the followingequation:

ZBlade = 30 × ln

(2

(h2

Blade + r2Blade

)r2Blade

)[Ohm] (1)

where hBlade is an equivalent mean-height and rBlade is an equiva-lent radius of the segment of the rotor blade being analyzed.

The parameters of assumed decoupled surge impedance modelZBlade for the blades as an attempt to represent their electromag-netic response to fast transients based on the theory presented in[9,10,18].

For the calculation of the surge-impedance of the rotor bladesan approximation was assumed with a reduced velocity of surgesof 65% of light speed, as an attempt to model the capacitive-effect ofthe structural and skin material of the rotor blade (Glass fiber rein-forced plastic—GFP). Carbon fiber reinforced—CFP materials, whichare widely used in the wind turbine blade manufacturing, wherenot considered during this study and will be part of a future project.

34.5Substation

To other

575

4.16 k5 k V

WG Gro

Systemr WG

25 m

25 m

120 m

120 m

5 V

kV

Tower

round ing

m

Converte r

Stator side

DFIG

Rotor side

Nacelle

Fig. 3. Schematic representation including the DFIG and the AC/AC electronic converter connected at the grid side and the DFIG rotor side [4,5].

dx.doi.org/10.1016/j.epsr.2015.11.039



Y.M. Hernández et al. / Electric Power Systems Research xxx (2015) xxx–xxx 3

In order to emulate the surge impedances of the lightningprotection system (LPS) and its corresponding connections, therotor blade was segmented into three sections. Table 1 depicts theparameters for the rotor blades.

The bearings, which consist of rotor blades, main and yaw orazimuth bearings, were modeled as a resistance RBear in series withan inductance LBear calculated at an electric frequency of 250 kHz(see Table 2). The inductance values were calculated with a FiniteElement Program-FEM (ANSYS APDL Ver. 13) [7].

The tower is usually manufactured in steel or reinforced con-crete. For simulation purposes the steel option was chosen, due tothe reason that this type of construction is widely used in this rangeof MW-class wind turbines. Table 3 shows the parameters for thefour segments of the steel tower assumed for this study. Similar tothe blade’s calculation, the tower parameters were estimated by(1).

2.2. Earthing (grounding) system

The earthing system type B described in [3], which is widely usedin the wind turbine industry, was chosen for this study. The footingresistance RT was calculated according to the guidelines proposedin [3,8]. Table 4 shows the parameters used for the calculation ofthe footing resistance with a soil with resistivity of 200 Ohm m.

2.3. Low and medium voltage cables

The non-shielded 0.575 kV LV cables connecting the powertransformer tertiary to the LV AC/AC power converter grid side,and, the cables connecting the power transformer tertiary to the LVauxiliary transformer, located in the nacelle, were modeled with aPI-equivalent model and calculated at 250 kHz; one conductor perphase and one additional copper conductor for earthing.

The 4.16 kV medium voltage (MV) cables with sheath connect-ing the secondary side of the WT transformers to the MV grid weresimulated and calculated with a PI-equivalent at 250 kHz.

Table 1Rotor blade modeling.

Segment hBblade [m] rBlade [m] V [m/�s] ZBlade [Ohm]

Root-section 28.50 1.50 195.00 197.54Mid-section 14.13 0.80 195.00 193.18Tip-section 13.88 0.50 195.00 220.25

Table 2Rotor blade, main and azimuth (Yaw) bearings.

Bearing RBear [Ohm] LBear [�H]

Blade 2.85E − 6 6.91E − 5Main 2.64E − 5 9.73E − 5Yaw 2.64E − 5 9.73E − 5

Table 3Tower modeling.

Section hTower [m] rTower [m] v [m/�s] ZTower [Ohm]

Section I (Foot) 34.50 2.40 255.00 180.87Section II 34.50 1.50 255.00 208.98Section III 34.50 1.00 255.00 233.28Section IV (Nacelle) 34.50 1.00 255.00 233.28

Table 4Earthing system.

�soil [Ohm m] Electrode [m] Radius—a [m] Buried depth [m] RT [Ohm]

200.00 9.00 0.50 3.50 2.56

Table 5LV and MV cables parameters.

Cable section Voltage [kV] Radius [mm] Length [m] εr [–] � [Ohmm]

LV 0.575 17.90 120.00 2.40 10.60E − 5LVEarth 0.575 8.95 120.00 1.00 10.60E − 5MV 4.16 20.00 120.00 2.70 10.60E − 5MVEarth 4.16 20.00 120.00 1.00 10.60E − 5HV 34.50 20.00 300.00 2.70 10.60E − 5HVEarth 34.50 20.00 300.00 1.00 10.60E − 5

The 34.5 kV high voltage (HV) cables with sheath connecting theprimary side of the WT transformers to the HV grid were simulatedand calculated with a PI-equivalent at 250 kHz. Table 5 depicts theLV, MV and HV cable parameters.

2.4. Wind turbine’s converter and generator

For the modeling of the electronic converter and the generator(DIFG) the following factors were considered:

• The capacitive coupling between DFIG stator and rotor isincluded.

• The blocked rotor inductance for a 2nd order synthetized circuiton the stator side and the capacitance between the stator andcoils are considered.

• The mutual capacitances and inductances between phases wereneglected.

• Similar effects are considered on the DFIG stator and rotor coils(inductance and capacitances).

After an extensive literature review and implementing validatedmodels for transients as reference, such as power transformers, analternative approach is proposed in order to represent the doubly-fed generator (rotor and stator windings), the electronic powerconverter (connected to the rotor side) and the capacitive couplingbetween rotor (low voltage) and stator (medium voltage) of theDFIG [8,9,11].

The LV electronic power converter was represented by a surgeimpedance model reflecting the capacitive coupling between bothAC sides and connected to the generator’s low voltage side [15,16].Fig. 4 depicts the topology adopted for the electronic power con-verter.

The model neglects the mutual capacitance between phases andconsiders mainly the Rc (ohmic losses) and Lac (coils) typical forconnections and terminals in both converter’s sides (including theharmonic filters). A concentrated impedance in form of a shuntresistance Zs and shunt capacitances to ground Cig and Cog of thecircuit represents the commercial low voltage circuit boards withpower electronics, such as IGBTs with gel based isolating materialand corresponding drive circuits.

The coupling between the converter’s AC grid side, the con-verter’s DC-link and the converter’s AC generator side was modeledusing an equivalent parallel capacitance to the power electronicscomponents Cig and Cog and the capacitance in the DC/DC converterwas modeled as a concentrated (high value) DC-Link capacitanceCdc. Table 6 shows the values of each parameter implemented inthe converter.

In the case of DFIG, the parameters Rr and Rs were obtained withtypical total winding resistances at nominal temperature. Ls is anestimation of the blocked rotor inductance, as proposed in [9]; Lr

was estimated as the 1/6 of Ls (these values range between 1/10and 1/5) [9,12,13,14]. Fig. 5 depicts the model used for the DFIG.

The constructive similarities between double fed induction gen-erators (DFIG) and synchronous generators (SG), from the modelingpoint of view, suggest similarities between both models for tran-sient calculation.

dx.doi.org/10.1016/j.epsr.2015.11.039




Grid

A

B

C

Zs

LacRc

Zs

LacRc

Zs

LacRc

Side AC

A

C

N

Cid

Cig Cdc

C

Cid

Cig

C

Cid

Cig

C

od

Cog Zs

RcLac

od

Cog Zs

RcLac

od

Cog Zs

RcLac

N

C

B

A

Rotor side AC

Fig. 4. AC/AC converter model used for transient studies (Y–Y Connection).

Table 6Converter parameters.

Rc [Ohm] LAC [mH] Zs [Ohm] Cig [nF] Cid [nF] Cog [nF] Cdc [�F]

0.10 3.00 250.00 0.10 1.00 0.10 2400.00

B

C

A

Stator side

Zs

LconRcu

Zs

LconRcu

Zs

LconRcu

Csr

Cs

Rs

Ls

Rr

Lr

Crg

Csr

Cs

Rs

Ls

Rr

Lr

Crg

Csr

Cs

Rs

Ls

Rr

Lr

Crg

Zr

RcuLcon

Zr

RcuLcon

Zr

RcuLcon

B

C

A

Rotor Side

Fig. 5. DFIG model used for transient studies (Y–Y Connection).

dx.doi.org/10.1016/j.epsr.2015.11.039




Table 7DFIG parameters (stator side and coupling).

Rcon [Ohm] Lcon [�H] Cs [nF] Cr [nF] Csr [nF] Rs [Ohm] Ls [mH]

0.05 1.00 93.00 10.00 3.00 0.10 30.00

Table 8DFIG parameters (Rotor side).

Rr [Ohm] Lr [mH] Zs [Ohm] Zr [Ohm]

0.30 5.00 500.00 1000.00

Table 9Parameters: LV, MV and HV MOVs.

Section RMOV [Ohm] LMOV [�H] CParallel [nF] Thermal limit [kJ]

LV MOV 1.00E − 4 0.50 6.50 2.00MV MOV 1.00E − 4 0.50 20.00 20.00HV MOV 1.00E − 4 0.50 20.00 20.00

Therefore the parameters Rcon and Lcon correspond to values ofresistances in contact leads and coils; Zr and Zs were modeled asconcentrated impedances in form of resistances in order to reflectthe attenuation in the oscillations and for integration stability ofthe model [9].

The surge capacitances were obtained for typical values forSGs with similar size and nominal power [9,12,16]. Fig. 5 andTables 7 and 8 depict the calculated values for the DFIG.

2.5. Low (LV), medium (MV) and high voltage (HV) MOVs

The representation chosen for the LV, MV and HV MOVs was inform of a nonlinear resistance with a capacitor connected in parallelper phase with the corresponding thermal limits as shown Table 9[16].

The characteristic curve assumed for the 0.575 kV LV MOVs isdepicted in Fig. 6. The LV MOVs were connected to the LV con-verter’s grid side (nacelle), LV converter’s generator side (same asgenerator’s rotor side in the nacelle), the tertiary side of the WT’sthree-coil transformer (tower foot) and the LV side of the auxiliarytransformer (nacelle) for a total of 4 sets of SPDs installed in theWT.

The characteristic curve assumed for the 4.16 kV MV MOVs isdepicted in Fig. 7. The MV MOVs were connected to the MV gen-erator’s stator (in the nacelle) and to the secondary MV side of theWT’s three-coil transformer (tower foot).

The characteristic curve assumed for the 34.50 kV HV MOVs,which are connected to the HV primary side of the WT’s three-coiltransformer, is depicted Fig. 8.

Fig. 6. LV MOV’s curve.

Fig. 7. MV MOV’s curve.

Fig. 8. HV MOV’s curve.

2.6. Wind turbine distribution’s transformer and wind park’sdistribution system

Each WT is equipped with a three-coil distribution transformer,whereby the primary side is connected to the HV distribution net-work (34.50 kV), the secondary side (4.16 kV) to the MV generator’sstator and the tertiary side (0.575 kV) to the WT’s LV network (Gen-erator’s rotor and auxiliary services). Table 10 and Fig. 3 showthe parameters and topology of the power transformer. The WTtransformers were modeled as a BCTRAN element and the earth-ing systems of the WTs were galvanic connected to each other andfirmly earthed [5,8,9].

2.7. Electrostatic charging (EC) and lightning stoke

The Cohen formulae, which is given by Eq. (2), establishes thatthe accumulate charge on a surface rubbing with air (effect of thewind speed and relative displacement of the blades), can be cal-culated using an approximation from Coehn and Lotz formula fortriboelectric charging [17,18,19].

As a first assumption the blades may be charged to a certainstatic voltage during its operation by friction between the air andthe composite blade material and it may be magnified when acharged cloud is close to the turbine (by the elevated electric fieldvalues up to approx. 8 kV/m).

�ac = 15 · 10−6 · (εr2 − εr1) [C/m2] (2)

Table 10WT transformer modeling.

Voltage CShunt [nF] CCoupling [nF]

34.5 kV (Delta) 0.80 0.254.16 kV (Star) 0.50 0.250.575 kV (Star) 0.10 0.25

dx.doi.org/10.1016/j.epsr.2015.11.039




SW

I

Blad e Ele ctro s

SW2

W1

R_El_Ch gV

static Cha rging bas

Vesd1

Rd

R_Environmen t

SWChg

sed on EN-6140 0

C

Ld

-4-2

Vesd2

Rch

Fig. 9. Blade electrostatic charging modeling based on EN 61000-4-2.

where, εr1 and εr2 are the relative permitivities of air (approx. 1.00)and blades material (approx. 2.40), respectively.

The following assumptions are included in the modeling of theelectrostatic charge effect in the overvoltage study:

• During normal operation, before a lightning stroke and with acharged cloud in the vicinity and above the wind turbine, a DCsource equivalent to the friction charge at 10–12 m/s wind speedis included with a value of 20–25 kV.

• Excessive charge above these critical values is expected to bediverted by the lightning protection system to the ground.

• Preceding the lightning stroke to the blade, an increassed chargeis applied to the blade (50–100 kV/m) and inmediately it is dis-charged with the equivalent circuit used in EN 61000-4-2 [20].

• The electrostatic charge (EC) of the rotor blades is modeled withthe same polarity as the lightning surge current. The aim of thisassumption is to model upward initiated flashes.

The final circuit model implemented is depicted in Fig. 9; whereVesd1 represents the electrostatic charge voltage during the electro-static charge phase and before the lightning stroke and Vesd2 reflectsthe effect when the value of the electric field collapses during andafter the lightning stroke. Concerning the charge/discharge circuit:Rch is the capacitor charging resistor (0.1 Ohm), C is the equivalentcharge capacitor (150 pF), Rd and Ld are the discharge impedance(330 Ohm and 3 �H). The Renv is an equivalent non-ionized air resis-tance (10 GOhm).

The ideal switches SW1 and SW2 represent the EC and theeffect after the lightning stroke. The ideal SWCh represents thecontrol charing capacitor in order to simulate the sudden collapseof the electric field. For simplification purposes, the ionizationof the air which is trapped in the hollow structure of the bladeand its surroundings was not taken into consideration for thesimulation. Table 11 depicts the parameters of the electrostaticcharge and lightning stroke modeling, respectively. The electro-static charge waveform with the sudden collapse of the potentialis presented in Fig. 10 assuming a lightning stroke at around50 �s.

Table 11Electrostatic charging and lightning stroke modeling parameters.

Source Peakvalue

Front duration[�s]

Stroke duration[�s]

Tstart [�s]

Vesd1 35 kV – – ≤50Vesd2 100 kV – – ≥50Heidler 200 kA 10.00 350.00 ≥40

(f ile ES

00

15

30

45

60

75

90

[kV]

SD.atp.pl4; x-v ar t) v :XX00 08

10 20 30 40 50 60[us]

Fig. 10. Waveform of the proposed model for the EC.

3. Simulation results with and without electrostaticcharging (EC)

The parameters of the wind turbines, the topology of the windpark and the simulation parameters are as follows:

• WT rated power: 3.6 MW.• WT tower height: 140 m and 4 tower segments.• Rotor blade length: 56 m, manufactured in glass reinforced fiber

composite (GRFC) and with built-in lightning protection system.• The wind park consists of five (5) WTs with a separation distance

between wind turbines of 300 m; lightning strikes the outermostwind turbine 1 (WT1).

• Electrostatic charging, which may cause wind park’s earth poten-tial rise, was modeled as depicted in Section 2.7.

• A positive lightning stroke on the end or outermost section of therotor blade (so-called “tip”) was simulated, in order to explorethe effect of the traveling waves along the wind park.

• The time parameters depicted in Table 11 were used for the sim-ulation of the electrostatic charging and the surge representingthe lightning stroke.

• Only a short period of time of approx. 50 �s for the electrostaticcharging (EC) was considered and explore the transient effect ofEC.

• Total simulation time: 100 �s and time step (delta t) is 20 ns inorder to avoid numerical instability.

• A value of 200 Ohmm is selected for the soil resistivity.

In order to establish a comparison, the simulation cases with andwithout electrostatic charging (EC) are presented in Figs. 11–18. Alldepicted voltages are presented as phase voltages and not line-linevoltages.

3.1. Simulations results: Wind turbine 1 (WT1 struck bylightning)

The WT1 MV generator stator shows similar transferred over-voltages independent of electrostatic charging EC (Fig. 11). Previousto the lightning strike initiation low amplitude disturbances in formof high frequency transient oscillations (disturbances) are observedin the case with EC (timestamp after 40 �s) at the left side of thefigure; these effects are probably caused by the capacitive couplingduring the lightning strike and the earth’s potential rise caused bythe EC. From the MOVs energy consumption point of view the MVMOVs do exceed the thermal limits of 20 kJ (see Table 9) and there-fore the risk of damage to the MV electrical installation increasesindependent of the consideration of EC. The consideration of the

dx.doi.org/10.1016/j.epsr.2015.11.039




Fig. 11. Overvoltage transferred WT1 (with/without EC): Generator stator (4.16 kV).

Fig. 12. Overvoltage transferred WT1 (with/without EC): Generator rotor (0.575 kV).

Fig. 13. Overvoltage transferred WT1 (with/without EC): Converter grid side (0.575 kV).

installation of MV surge arresters with high-energy dissipationcharacteristic could be considered.

Similar transferred overvoltages in amplitude at the WT1 LVgenerator wound-rotor side or electronic power converter’s rotorside are observed independent of EC (Fig. 12). Similar to Fig. 11and previous to the lightning strike initiation low amplitudedisturbances in form of high frequency transient oscillations areobserved in the case with EC (timestamp after 40 �s, when theEC is initiated) at the left side of the figure. During the transient

analysis the thermal limits of 2 kJ of the MOVs are not exceeded asrecommended in Table 9.

The WT1 LV electronic power converter’s grid side experiencesincreased transferred overvoltages with the consideration of EC(Fig. 13). Similar to Fig. 11 and previous to the lightning strikeinitiation low amplitude disturbances in form of high frequencytransient oscillations (disturbances) are observed in the case withEC (timestamp after 40 �s, when the EC is initiated) at the left sideof the figure. The dissipated energy does not exceed the thermal

dx.doi.org/10.1016/j.epsr.2015.11.039




Fig. 14. Surge current distribution WT1 (with/without EC): Across blade tip, tower’s grounding cables, nacelle’s metal structure and WT grounding resistance.

Fig. 15. Overvoltage transferred WT5 (with/without EC): Generator stator (4.16 kV).

Fig. 16. Overvoltage transferred WT5 (with/without EC): Generator rotor (0.575 kV).

limits of 2 kJ (as suggested on Table 9) and therefore the risk ofdamage to LV electrical installation observes less influence.

Less difference is observed after the lightning stroke concerningthe surge current across the rotor blade tip, the tower’s ground-ing cables, the nacelle’s grounding system and the WT groundingsystem (ground resistance); only during the EC (after 40 �s) andprevious to the lightning stroke a low value DC current of approx.45 A is observed across the grounding system (Fig. 14 on the left

side). A fraction of the lightning current flows across the tower’ssegments, as expected.

3.2. Simulations results: Wind turbine 5 (WT5 fahrthest windturbine)

The WT5 MV generator stator shows similar transferred over-voltages independent of electrostatic charging EC (Fig. 15). The EC

dx.doi.org/10.1016/j.epsr.2015.11.039




Fig. 17. Overvoltage transferred WT5 (with/without EC): Converter grid side (0.575 kV).

Fig. 18. Surge current distribution WT5 (with/without EC): Across blade tip, tower’s grounding cables, nacelle’s metal structure and WT grounding resistance.

effect on the WT5 observes less disturbances in regards to EC. Onlythe effect of travelling waves across the earth installation in formof overvoltages is recorded. The thermal limits of the MV MOVs, asassumed from Table 9, are not infringed and therefore the risk ofdamage to the MV electrical installation does not increase

Similar transferred overvoltages at the WT5 LV generatorwound-rotor side (or electronic power converter’s rotor side) areobserved independent of EC (Fig. 16). The EC does have less influ-ence and the predominant effect is driven by the event of thelightning stroke on the WT1. The effect of the lightning stroke isreflected on the WT5 after approx. 7 �s, which is the time requiredto overcome the distance between the WT1 and the WT5 of approx.1200 m. The thermal limits of 2 kJ of the MOVs are not exceeded asrecommended in Table 9.

The WT5 LV electronic power converter’s grid side experiencesless difference in regards to transferred overvoltages with the con-sideration of EC (Fig. 17). The dissipated energy by the MOVs doesnot exceed the thermal limits of 2 kJ (see Table 9) and therefore therisk of outages is further reduced.

Less difference is observed after the lightning stroke concern-ing the surge current across the tower’s grounding cables, thenacelle’s grounding system and the WT grounding system (Fig. 18).A remarkable portion of the lightning current flows across thegrounding system (electromagnetically coupled due to travellingwaves) and the tower’s grounding cables. In this case the earth’spotential rise causes this effect, due to the reason that not con-duction current is flowing across the tip of the WT5’s bladetip.

4. Conclusion

A modeling approach of the rotor blade electrostatic charg-ing (EC) of wind turbines is presented and its effects on lightningovervoltages in wind turbines have been investigated through ATP-EMTP simulations. These effects may open new fields of researchon the consideration of the dimensioning of the protection compo-nents in the wind park.

On the wind turbine attached by lightning with previousElectrostatic Charging, the LV converter grid side, connected tothe tertiary of the three-coil power transformer, experiences aremarkable effect in form of transferred overvoltages; thereforeelectrostatic charging should be taken into account as an additionalparameter in the design and dimensioning of electrical protectionsystems, which should be one of the main focus areas of the aca-demic and industrial research activity in this topic of electrostaticcharging.

There are facilities that offer possibilities of combined testing ofelectrostatic charging and impulse voltage (switching or lightning),especially in combined high and impulse voltage testing of rotorblades in the high voltage laboratory, which will be explored in thenear future.

The installation of MOVs on the HV, MV and LV side of the elec-trical installation is an accepted practice; however, the additionalconsideration of electrostatic charging (EC) effect may impose addi-tional requirements that have to be further analyzed, especially inthe computation of overvoltages and energy dissipation of surgeprotection devices.

dx.doi.org/10.1016/j.epsr.2015.11.039




References

[1] W. Schulz, H. Pichler, G. Diendorfer, Evaluation of 45 negative flashes basedon E-field measurements, video data and lightning location data in Austria,in: Proc. of 30th International Conference on Lightning Protection—ICLP 2010,Italy, 2010.

[2] Y. Méndez, T. Tsovilis, F. Asimakopoulou, Z. Politis, W. Barton, M. MartínezLozano, A simulation approach in electrostatic charging of rotor blades and itseffects on transferred overvoltages in wind parks, in: Proc. of 32nd InternationalConference on Lightning Protection—ICLP 2014, China, 2014.

[3] IEC, IEC 61400-24 Ed.1: Wind turbines—Part 24: Lightning protection, in: IEC61400-24, 2010.

[4] M. Reichard, D. Finney, J. Garrity, Windfarm Protection Using Peer-to-Peer Communication, 2007, 〈www.gedigitalenergy.com/smartgrid/Aug07/windfarm.pdf〉 (last access 08/2015).

[5] G.J. Wakileh, Wind Turbine Reactive Power Capability, 2005, 〈www.iso-ne.com/committees/comm wkgrps/relblty comm/relblty/mtrls/2005/jul282005/a121 reactive capability curve 0114ge are e0190 3 6mw rev0.pdf〉 (last access08/2015).

[6] Y. Méndez, W. Zischank, B. Lewke, K. Stimper, An experimental approach ofthe effects of lightning currents on rotor blade tips of wind turbines, in: Proc.of 31st International Conference on Lightning Protection—ICLP 2012, Austria,2012.

[7] Y. Méndez, A. Castro, J. Kindersberger, B. Lewke, J. DiBella, R. Hardison, R.Stolper, A first approach in the modelling of failures in rotor blades causedby lightning strikes, in: Proc. of 30st International Conference on LightningProtection—ICLP 2010, Italy, 2010.

[8] Y. Yasuda, T. Toshiaki, Equivalent equation of earth resistance for ring elec-trode of wind turbine, in: Proc. 30th International Conference on LightningProtection—ICLP 2010, Italy, 2010.

[9] CIGRE, WG 02 (Internal Overvoltages). Study committee 33, in: Guidelinesfor Representation of Network Elements When Calculating Transients, CIGRE,1990.

[10] A. Greenwood, Electrical Transient in Power Systems, second ed., John Wiley &Sons, U.S.A., 1991.

[11] B. Badrzadeh, M. Hogdahl, E. Isabegovic, Transients in wind power plants—PartI: Modeling methodology and validation, IEEE Trans. Ind. Appl. 48 (2) (2012)794–807.

[12] T. Funabashi, N. Takeuchi, T. Sugimoto, T. Ueda, L. Dube, A. Ametani, Generatormodeling for transformer transfer voltage study, IEEE Trans. Energy Convers.14 (4) (1999) 1193–1197.

[13] J. Arrillaga, S. El-Batal, Lightning-surge distribution in H.V. D.C. convertors, Proc.Inst. Electr. Eng. 120 (5) (1973) 595–600.

[14] L. Chen-long, S. Qi, Q. Lei, C. Xiang, F. Chao, W. Xiaoguang, Quantitative analy-sis of voltage distribution within ±1100 kv HVDC converter valve tower undervarious transient over-voltage conditions, in: Proc. of 32nd International Con-ference on Lightning Protection ICLP 2014, China, 2014.

[15] P. Ledesma, J. Usaola, Doubly fed induction generator model for transient sta-bility analysis, IEEE Trans. Energy Convers. 20 (2) (2005) 388–397.

[16] M. Martinez, L. Díaz, Consideration of TVSS Energy Consumption in Wind-mills During Lightning Strokes, International Colloquium Lightning and PowerSystems, CIGRE, France, 2014.

[17] E. Kuffel, W. Zaengl, J. Kuffel, High Voltage Engineering Fundamentals, seconded., Butterworth-Heinemann, U.K, 2000.

[18] H.W. Dommel, EMTP—Theory Book, Branch of System Engineering, BonnevillePower Administration, U.S.A, 1986.

[19] B. Hermoso, E. Eizaguirre, I. Eizaguirre, Risk assesment on a windmill higherthan 90 meters, in: Proc. of 26th International Conference on LightningProtection—ICLP 2002, Poland, 2002.

[20] EN, Electromagnetic compatibility. Part 4-2: Testing and measurementtechniques—electrostatic discharge immunity test, in: EN 61000-4-2:2009,2009.

[21] IEEE Fast Front Transients Task Force, Modeling guidelines for fast front trans-ients, IEEE Trans. Power Delivery 11 (1) (1996) 493–506.

dx.doi.org/10.1016/j.epsr.2015.11.039

http://refhub.elsevier.com/S0378-7796(15)00372-7/sbref0110















































































































































































































































































































































































































































































































































































doi: 10.1016/j.epsr.2015.11.039 persistent link using

Documents