TRANSIENT ANALYSIS ON WIND FARM SUFFERED FROM LIGHTNING YASUDA Yoh(1) and FUNABASHI Toshihisa(2) (1) Kansai University, (2)Meidensha Corporation ABSTRACT Wind power generations are often struck by lightning because of its special shape of open-air and very-high construction. Although some reports, e.g. IEC TR61400-24 and NREL SR-500-31115, indicate those accidents and the problem, there seems to be room for more work to be done in this area. The authors, therefore, focus on and have investigated surge analysis on wind farm connected to a power system. Under the conditions of several values of earth impedance (considering its inductivity as well as resistivity) around thunderstruck wind turbine, sensitivity analysis was performed using digital transient simulator ARENE. The result of the analysis shows that the surge propagating from the thunderstruck wind turbine to the next one may become large depending on the conditions of earthing. The result suggests possibilities of serious accidents to electric and electronics devices in wind farms due to surge propagation, as well as direct lightning strikes. Keywords: wind power generation, wind farm, lightning, earthing(grounding) 1. INTRODUCTION Wind power generation is one of the fastest-growing categories of the electric power industry. As its installation has been explosively introduced worldwide, the problems in the interconnection to the grid have been brought up. Recently lightning protection of wind power generation is surfacing into the public as an important matter [1]. Especially Japan is suffering from frequent and heavy lightning strikes, one of which is notorious as “winter lightning” in costal areas of The Sea of Japan. It is pointed out that wind power generation facilities tend to be exposed to lightning damage due to their configurations and it is considered that different measures from the conventional electrical equipments are necessary. However, the surge propagation during lightning strikes at wind power generation facilities is still far from being clearly understood. It seems that each manufacturer of wind power generators has accumulated on-site know-how for analysis of lightning accidents and the measures against them. However accident reports and analysis studies are not much published in academic and industrial societies, excepting in some foresighted studies [2]. Their needs to further disclose information and make discussions regarding these problems open. According to the report IEC61400-24 [1], the most frequent accident is dielectric breakdown on low-voltage circuits including electric and telecommunication equipments. The events on low-voltage circuits are not triggered by only direct lightning strikes but also induced lightning and back-flow surges propagating around wind farms just after lightning strikes on other wind power generators. In general, converter circuits and boost transformers are set up in the very vicinity of or inside windmill. Surge arresters may often take inverse function from earth to the line if earth potential near transformers is raised at the moment of lightning strikes. Therefore, there is a possibility that the lightning current will flow into distribution lines at the wind farms surged. The real cases of lightning accidents at wind farms often see dielectric breakdowns at wind turbines near or far from thunderstruck turbines, not only at the thunderstruck ones. In this study, we analyse and discuss the general tendency in lightning accidents at wind farms by utilizing an ideal wind farm model including two wind turbine for the analysis of lightning surges, The main objective of this study is to grasp the cause of dielectric breakdowns at non-thunderstruck wind turbine regarding of the wind turbine’s earthing and location of lightning strikes. 2. WIND FARM MODEL CONSISTING OF TWO WIND TURBINES Figures 1 and 2 show a wind farm model composed of two wind power generators. This configuration is assumed as an installation of wind farm on linear and narrow land along seashore or on the ridge of mountains. It was supposed in this model that a grid-interactive (6.6 kV / 66 kV) and two wind power generators, which are identical with each other in performance and condition, of the 1 MW class with 1 km intervals. The considerable WT#2 WT#1 Tr WT#2 WT#1 Case A1 Tr WT#2 WT#1 Tr Case A2 Case B Fig.1 Considerable combination of wind farm installation and lightning strike. grid-interactive transformer 400 Ω 66 kV 50 km wind farm Infinite bus grid system lightning 660 V WT #1 1 km 1 km 1 km 6.6 kV 6.6 kV boost transformer ~ ~ ~ 1 km surge arresters boost transformer 660 V ~ ~ ~ ~ ~ ~ WT #1 WT #2 (a) configuration of Case A detail model (see Fig.3) ~ ~ ~ ~ ~ ~ WT #1 WT #2 grounding resistance wind tower #1 (b) configuration of Case B Fig.2 Model for surge analysis on wind farm with two wind turbines installation combination of those is twofold as shown as Fig.2 (a) and (b), and the considerable cases of lightning strike to wind turbine exist three sets, i.e. Case A1, A2, and B, as illustrated in Fig.1. Although a wind power generator consists of a gear box, a synchronous of induction generator, a rectifier, a 3-phase inverter, and so on, we adopted a synchronous machine model for simplicity. Boost transformer (660 V / 6.6 kV) for each wind turbine is equipped in the vicinity or inside the wind tower and surge arresters are attached to the primary and secondary terminals and common-earthed, as shown in the detailed diagram in Fig.3. We conducted sensitivity analysis with varying parameters in an assumption that resistance is from 1 to 10 Ω and inductance is from 0 to 10 µH in lumped constant. It was assumed that the lightning strike is a standard summer lightning of pulse height: 30 kA, crest width: 2 µs, and wave-tail length: 70 µs and it hits a blade of a windmill and reaches into an earth electrode through earthing wire and then boosts earth potential. In addition, because what we want to see in this simulation is a dielectric breakdown at not-thunderstruck wind turbines, we assumed that blade burnout or explosive destruction Fig.3 Detail model for each wind turbine. and dielectric breakdown at the thunderstruck wind turbine had been avoided by certain measures. More detail information for the models is shown in Table 1. In the present investigation, we used ARENE digital power system simulator [3], which has the same analysis algorism as EMTP (Electro-Magnetic Transients Program), for analysis work. Table 1 Analysis Conditions. Wind turbine (Synchronous Generator) model rating power [MVA] 1.00 direct-axis reactance Xd, Xd’, Xd” [p.u] 2.00, 0.25, 0.20 quadrature-axis reactance Xq, Xq’, Xq” [p.u] 1.90, 0.50, 0.20 time constants Tdo’, Tdo”, Tqo’, Tqo” [s] 6.0, 0.03, 0.50, 0.06 Boost transformer model connection method Y (neutral earthing) /Δ rating power [MVA] 1.0 rating voltage [V] 600 / 6,600 frequency [Hz] 60 % impedance [%] 2.48 mutual leakage inductance [mH] 18.2 Line Model positive- / zero- phase resistance [Ω/km] 0.00105 / 0.0210 positive- / zero- phase inductance [mH/km] 0.83556 / 2.50067 positive- / zero- phase capacitance [nF/km] 12.9445 / 6.4723 frequency-dependent characteristics available 3. RESULTS OF ANALYSES 3.1 Cases with resistive earth Figure 4 displays the analysis results of Case A and B under the assumption of earth resistance as 10 Ω. In the figures, graph (a), (b) and (c) in the Case A1 show a lightning current waveform, three-phase voltage waveforms at the output terminal of the boost transformer for the thunderstruck wind turbine, and propagated surge waveforms to the boost transformer for the next non-thunderstruck turbine, respectively. As shown in the waveforms, it is clear that the surge tends to flow out through the surge arresters at thunderstruck tower, which arresters reversely pass through back-flow surge from common-earth to the power line, toward the next non-thunderstruck wind tower. current [kA] 0 (a) Summerising above discussions, Figs.5 and 6 are given to clarify relationship between earth resistance and tendency of the surge propagation. Firstly, it is evident that less resistance of earthing would cause to less surge damage both in thunderstruck and non-thunderstruck wind turbine. Moreover, the results indicates that the surge altitude in the thunderstruck wind turbine itself is not much different among three cases, while the surge reached to the non-thunderstruck turbine clearly depends upon the configuration of wind farm. From the viewpoint of the surge protection of direct lightning strike, Case A2 and Case B are worse than Case A1, which means that effective countermeasure for the wind turbine at the end of branch line in wind farm are needed. Another point of view on the back-flow surge accident, Case B is rather safer than the other cases, that is, parallel-line configuration in wind farm may avoid surge’s concentration to the certain equipment. In any cases, it is indisputable that the surge tends to concentrate the turbine at end of branch lines. It is therefore suggested 20 earthing resistance [Ω] 10 (b) 200 100 0 50 10 5 2 Case A1 (Lightning on WT#1) Case A2 (Lightning on WT#2) Case B 1 0 (c) 50 100 150 200 250 surge altitude [kV] 25 Fig.5 Surge arising in thunderstruck wind turbine. (cases with resistive earth) 0 Case A1 -25 50 phase a 25 earthing resistance [Ω] voltage @WT2# [kV] voltage @WT2# [kV] 30 voltage @WT1# [kV] Figure 4 also shows the result by another case, Case B as shown in (d). Comparing two waveforms, the surge altitude reached to the next tower in Case A is contained almost half rower. Case B has the parallel configuration in which the transformer is installed in the centre of two wind turbines. In the parallel configuration of wind farm, the lightning surge may firstly reach the low-voltage terminal of the grid-interactive transformer and flow into another branch of line toward the next wind turbine. This tendency is shown in the results under other consumption of line length, e.g. 500 m, 200m. Consequently, the reason is considered to the configuration of wind turbines and the grid-interactive inverter. (d) phase b 0 phase c Case B -25 0.00 0.05 0.10 time [ms] 0.15 Fig.4 Surge waveforms in wind farm. (cases with resistive earth Rg = 10 Ω) 0.20 10 5 2 Case A1 (Lightning on WT#1) Case A2 (Lightning on WT#2) Case B 1 0 10 20 30 40 50 surge altitude [kV] Fig.6 Surge flowing into non-thunderstruck wind turbine. (cases with resistive earth) that the grade and cost for lightning protection for each wind turbine should be varied according to its geometrical and electrical location. 3.2 Cases with inductive earth Not only resistance, inductance also gives strong influence to lightning protection of electrical equipment. Existence of inductive factor in earth impedance may cause unexpected surge and result to breakdown accident even if the earth resistance has enough measured to be lower than several ohms. Inductive earthing can have a characteristic of typical impedance curve, where the value rapidly crests under lightning invasion, and then it draws slow decent curve and finally convergences to the study value of resistance. It is known than inductive earthing tends to be represented when mesh electrode or long lateral electrode like a ling-shaped earthing are used. The results of the sensitivity analysis varying inductance factor of the thunderstruck tower shown in Figs.7, 8 and 400 Surge Altitude [kV] Case A1 Case A2 Case B Fig.10 300 Surge when Rg0 = 10 Ω Surge when Rg0 = 10 Ω Surge when Rg0 = 10 Ω 200 Rg0 = 5 Ω 100 Rg0 = 2 Ω Rg0 = 1 Ω 0 0 10-1 100 inductance [µH] 101 0 10-1 100 inductance [µH] 101 0 10-1 100 inductance [µH] 101 Fig.7 Surge arising in thunderstruck wind turbine. (cases with inductive earth) left: (a) Case A1(Lightning on WT#1), centre: (b) Case A2(Lightning on WT#2), right: (c) Case B 60 Surge Altitude [kV] Case A1 Case A2 50 Surge when Rg0 = 10 Ω Surge when Rg0 = 10 Ω 40 30 Rg0 = 5 Ω 20 Rg0 = 2 Ω 10 Rg0 = 1 Ω 0 0 10-1 100 inductance [µH] Case B Surge when Rg0 = 10 Ω 101 0 10-1 100 inductance [µH] 101 0 10-1 100 inductance [µH] 101 Fig.8 Surge flowing into non-thunderstruck wind turbine. (cases with inductive earth) left: (a) Case A1(Lightning on WT#1), centre: (b) Case A2(Lightning on WT#2), right: (c) Case B 40 Surge Altitude [kV] Case A1 Case A2 Case B Surge when Rg0 = 10 Ω 30 Surge when Rg0 = 10 Ω Surge when Rg0 = 10 Ω Rg0 = 5 Ω 20 Rg0 = 2 Ω 10 Rg0 = 1 Ω 0 0 10-1 100 inductance [µH] 101 0 10-1 100 inductance [µH] 101 0 10-1 100 inductance [µH] Fig.9 Surge flowing into grid-interactive transformer. (cases with inductive earth) left: (a) Case A1(Lightning on WT#1), centre: (b) Case A2(Lightning on WT#2), right: (c) Case B 101 In the viewpoint of comparison of wind farm’s configuration, Case B (parallel arrangement of two wind turbines centring around grid-interactive transformer) is the safest to avoid back-flow surge propagating, while Case A1 (cascade arrangement of wind turbines and transformer) is relatively better as far as direct surge protection. Concerning to the effect to grid-interactive transformer and the outage grid, Case A2 seems to be best. Thus, there is considerable difference varying electrical arrangement of wind turbines and lightning-strike location. This suggests detail and careful design and analysis when a new wind farm installed or adoptive surge protection devices are connected to an existing wind farm. 4. CONCLUDING DISCUSSION In this paper lightning surge analysis on an ideal wind farm model consist of two wind turbines and a grid-interactive transformer is performed for the purpose of understanding the general tendency of propagating surge in the wind farm. The sensitivity analyses with varying parameters of wind farm’s configuration, lightning-strike locations, resistance and inductance of earth are calculated. From the result of analysis, the following tendencies are confirmed; (i) Less resistance of earth provides safer protection for both direct thunderstruck and back-flow surge propagating around the wind farm. (ii) However, only a few inductance of no less than 10 µH gives considerable effect to rise surge propagation as same as with higher resistance. (iii) Parallel arrangement is the safest from the viewpoint of back-flow surge protection. (iv) The wind turbine at the end of cascade branch line tends to be suffered from larger surge in case of both direct thunderstruck and invading back-flow surge. As concluding these discussions, it is necessary to conduct more detailed analysis considering wind farm’s arrangement and electrical location of wind turbines as lightning current [kA] 9 delineate the above feature. The lateral thin line in each graphs indicates the surge value when earth resistance is assumed to be 10 Ω and non-inductive. Comparing this 10-Ω line, it is easily understood that existence of inductive factor of only several µH affects large surge voltage as same as with higher resistance, even if it has low steady resistance Rg0 of 1 Ω. The reason is thought that the impedance waveform of inductive earth responds rapidly in very high frequency domain like a lightning impulse. This phenomenon can be expressed in the calculated impedance curve shown in Fig.10, which is obtained as the quotient of simultaneous earth potential and simultaneous current passing through the earth electrode. More than 10-Ω crest can be recognised in the earth impedance curve in spite of only 1-Ω steady resistance. 30 20 10 earth potential [kV] 0 300 200 100 earth impedance [Ω] 0 12 8 4 0 0.00 0.02 0.04 0.06 time [ms] 0.08 0.10 Fig.10 An example of waveforms in case with inductive earth. (Rg = 1 Ω, Lg = 10 µH) well as earth impedance. For example, if a new wind farm is designed, electrically parallel arrangement of wind turbines is recommended as possible as it can. Also, if adoptive surge protection devices are being connected to an existing wind farm, the wind turbine at the end of cascade branch line should be measured preferentially. We hope that an effective methodology will be established for insulation coordination and lightning protection for wind power generation and wind farm in the not-so-distant future. REFERENCES 1 “Wind turbine generation systems – 24: Lightning protection”, IEC Technical Report, TR61400-24, 2000. 2 B. McNiff: “Wind Turbine Lightning Protection Project 1999-2001”, NREL Subcontractor Report, SR-500-31115, 2002. 3 http://rdsoft.edf.fr/ AUTHOR’S ADDRESS The first author can be contacted at; Department of Electrical Engineering and Computer Science, Kansai University Yamate-cho 3-3-35, Suita City, Osaka 564-8680, JAPAN email: yasuda@kansai-u.ac.jp