Effective Lightning Protection For Wind Turbine Generators

Bruce Glushakow, Member, IEEE

AbstractA wind turbine generator is the most exposed of all types of generators connected to electric utility systems. Wind turbines are most often erected in hostile lightning environments. Lightning damage to wind turbines is costly in terms of repair and replacement of equipment. Lightning damage is the single largest cause of unplanned downtime in wind turbines, and that downtime is responsible for the loss of countless megawatts of power generation. There is currently no international standard governing wind turbine lightning protection. This paper offers guidelines for effective protection. Index TermsLightning, risk analysis, surge protection, wind energy, wind power generation, wind turbine generators.
Fig. 1. First process in the generation of lightning.

I. INTRODUCTION INDMILLS have been in use for over 2000 years, the earliest uses of which were grain grinding and water pumping. In todays energy-hungry world, the skyrocketing prices of fossil fuel, its negative impact on the environment, and reluctance to depend on an uncertain source of supply have prompted many nations to invest in alternative, renewable sources of energy. Wind energy tops the list of such alternative energy sources. Areas of favorable locations for wind turbines coincide with areas of thunderstorm activity. Maps prepared by NASA show that in most areas where wind density is high, there are 30 or more thunderstorm days per year [1]. In a study completed in 2002, the National Renewable Energy Association statistics showed that up to 8 out of 100 wind turbines could be expected to receive one direct lightning strike every year [2, p. 18]. Between 1992 and 1995, Germany alone reported 393 incidents of lightning damage to wind turbines (124 direct strikes to the turbine, and the remainder through the electrical distribution network) [3]. From an electrical point of view the earths surface is negatively charged and the upper atmosphere is positively charged. Most of the time, the atmosphere is a good enough insulator to keep these two charges apart. However, when storms build up close to the earths surface, electrical charges in those clouds (either positive or negative) can be transferred to earth. This is called lightning. The most common lightning and the one we are interested in with regard to wind turbine lightning protection, is cloud-to-ground (CG) lightning and currents as great as 300 000 A have been recorded from CG ashes. (Turbines over 100 m high can also be subject to upward lightning, but this is relatively rare.)

Manuscript received July 12, 2006; revised October 19, 2006. Paper no. TEC00336-2006. The author is with MCG Electronics Inc., NY 11729 USA (e-mail: bglushakow@ieee.org). Digital Object Identier 10.1109/TEC.2006.889622

Fig. 1 shows the rst process in the generation of lightning: the separation of positive and negative charges in the thunder cloud over a wind turbine. Positive charge accumulates at the top of the cloud; negative charge at the bottom. Large quantities of accumulated negative charge will have two effects on the ground directly below the cloud: it will repel negative charges and attract positive charges. When the atmospheres insulative effect breaks down, charge will be transferred to the ground in the form of a lightning stroke. The earth is normally negatively charged with respect to the atmosphere. However, as the thunderstorm passes over the ground, the negative charges at the bottom of the cloud cause the positive charges on the ground to gather along the surface for several miles around the storm. This positive charge becomes concentrated in vertical objects such as trees and tall buildings and these objects emit streams of positive energy that shoot up to try and meet the down-rushing negative charge from the cloud. Lightning occurs when the negative charge from the cloud joins the positive charge from an object on the earth. Whenever a lightning charge propagates to earth, it must have an attachment point, which is the origin of the upward stream of charge. Wind turbines are excellent attachment points because they are often in exposed locations and signicantly taller than adjacent objects. It is the blades that are by far the wind turbines most likely attachment points. However, almost any spot on the turbine is susceptible to direct lightning strikes including: air terminals (lightning rods), the nacelle (cab housing the generator and controller on the top of the tower), protuberances near the top of the structure, and the tower itself. When any part of the turbine is hit by lightning, it becomes a part of the lightning discharge path. From the attachment point to the ground, the lightning current will ow predominantly through the lowest-impedance paths available to it. Lightning protection mitigates lightning damage by providing a low-impedance path to the ground, which will shunt the lightning current away from the components susceptible to lightning damage.

0885-8969/$25.00 2007 IEEE

Authorized licensed use limited to: King Mongkuts Institute of Technology Thornburi. Downloaded on October 16, 2008 at 00:24 from IEEE Xplore. Restrictions apply.

A lightning protection system consists of lightning receptor points, conductors, grounding system, and surge protection for every subsystem. A low-impedance grounding system is a prerequisite for any lightning protection plan. This includes bonding all metal elements, systems, and potential attachment points together. If this is not done, a voltage potential of millions of volts can build up between the discrete turbine elements. Attention must also be paid to the wiring conguration of conductors adjacent to the lightning down-conductor(s), to avoid inducing large voltages (and the resulting surge currents) in the loops formed by these power and control circuits within the tower and nacelle. Generators and control system elements are susceptible from both direct strikes to the turbine and from currents induced on the connected power distribution lines beyond the tower itself. Therefore, in addition to good grounding, surge protection devices (SPDs) are required to protect these elements. II. EXISTING STANDARDS FOR LIGHTNING PROTECTION Several international lightning protection standards exist, which have been applied to wind turbines. These include: r IEC 61024 Protection of Structures Against Lightning; r IEC 61662 The Assessment of Risk Due to Lightning; r IEC 61312 1-5 Protection Against Lightning Electromagnetic Impulses. Wind turbines that have been protected using the principles from these three standards, have still experienced signicant lightning damage. Documentation of this damage has been produced by researchers at the University of Manchesters Institute of Science and Technology [4], Danish Electric Utilities Research Institute (DEFU) [5], International Energy Agency (IEA) experts [6], and the National Renewable Energy Laboratory (NREL) [2, p. 20]. There is no standard specically written anywhere in the world for protecting wind turbines against lightning damage. The closest document is IEC/TR 61400-24 Ed.1.0 en:2002 entitled Wind turbine generator systemsPart 24: Lightning protection. The TR in the title stands for Technical Report and in its own words on page 25: This document, which is purely informative, is not to be regarded as an international standard. IEC/TR 61400-24 should be credited with opening a very important door. It addresses the major issues relating to lightning protection of wind turbines. Its strengths include: r the presentation of wind turbine lightning damage statistics (Section 4); r the comprehensive analysis of lightning protection of the rotor blades, bearings, and gearbox (Sections 6 and 7); r agreement with other international standards and studies that a lightning protection system should be capable of diverting a peak current of 200 kA without damage (Paragraph 5.3); r the emphasis placed on the need for proper grounding and bonding procedures. Its weaknesses are covered in Section VII of this paper and it is hoped that these issues will be rectied in the 2007 revision of the document.

Fig. 2. Relative incidence of lightning strikes. (Chart reproduced from the Danish database consulted by IEC/TR 611400-24.)

In the meantime, the current paper will propose some workable guidelines for protecting wind turbines from lightning damage. III. LIGHTNING DAMAGE STATISTICS The statistics quoted in IEC/TR 61400-24 are highly informative. The document notes that in northern Europe (Germany, Sweden, and Denmark) where lightning is comparatively infrequent, 4%8% of all wind turbines will suffer lightning-caused damage every year. However, in areas of greater lightning density, this gure is reported to be considerably higher. One European manufacturer puts it at 14% in southern Germany.1 In Japan, CRIEPI reports the gure as 36% [7]. The document further reports that in its northern Europe study 7%10% of the wind turbine damage was to the rotor blades. This is important because the rotor accounts for 15%20% of the cost of a wind turbine and thus, damage to rotor blades is the most expensive type of lightning damage. It should be noted that most of the rotor damage reported was to older style blades manufactured out of nonconducting ber composite materials. Almost all modern turbine blades are constructed with built-in lightning protection in the form of conducting elements. This improved blade design has signicantly reduced the amount of blade damage [2]. Another observation from the reports northern European data is that only one-third of the lightning faults were caused by direct strikes to the turbine. The other two-thirds were attributed to lightning strikes to the power and telecommunication networks connected to the turbines. Data from a US source suggests that this ratio is closer to 2% damage from direct lighting strikes and 98% from strikes to the nearby ac power and signal lines [8]. One of the most signicant statistics in the IEC/TR 61400-24 document is that 50%70% of lightning faults are to the control and electrical systems and this damage accounts for at least twice as many days of downtime as damage to rotors. Fig. 2 illustrates the relative incidence [9].
1 LM

Glasber A/S Denmark www.lm.dk

Authorized licensed use limited to: King Mongkuts Institute of Technology Thornburi. Downloaded on October 16, 2008 at 00:24 from IEEE Xplore. Restrictions apply.

Fig. 3. Two types of structures that share all the problems implicit in protecting a wind turbine from lightning except for lightning attachment to a rotating blade.

IV. LIGHTNING RISK FACTORS TO BE ADDRESSED The above statistics give credence to the main conclusion of a study commissioned by the European Union and conducted by the University of Manchester: the protection of wind turbine electronic systems from indirect effects is of equal importance to, if not greater than, the protection against direct effects [10]. The three risk factors to be mitigated then are as follows. 1) Damage to blades caused by direct strikes: This damage can be caused by strikes to the tips of the blades and also to strikes along the length of the blades. Almost all direct strikes to a wind turbine will hit the rotor blades. 2) Damage caused by surge currents: This damage can be caused by surge currents originating from either direct strikes to the blades or coming from (indirect) strikes to connected power and data lines. This would include the ac power lines as well as the telephone or supervisory control and data acquisition (SCADA) lines used to remotely control the turbines. 3) Damage caused by voltages: This damage can be caused by voltages induced in circuits (power as well as control) adjacent to the necessary down-conductors that carry the lightning current to earth. V. RECOMMENDED LIGHTNING PROTECTION GUIDELINES A. Successful Protection Models of Kindred Applications IEC/TR 61400-24 Section 3.5 suggests that lightning protection of wind turbines presents problems that are not normally seen with other structures. In fact, there are at least two types of structures that share all the problems implicit in protecting a wind turbine from lightning, except for lightning attachment to a rotating blade (see Fig. 3). Both radome antenna towers and lighthouses are geometrically similar to wind turbines, are most often sited in highlightning remote and exposed locations, and have sensitive electronics embedded in their towers. Radomes can be constructed over 150 m off the ground. After years of studying the problem, the Federal Aviation Administration (FAA) analyzed successful principles of protecting its radome antenna towers from lightning and made them mandatory in all FAAs hundreds of installations [11, p. 11, Tab. 1]. Similarly, after years of studying successful and unsuccessful lightning protection applications at hundreds of its lighthouses, D/GPS stations, and maritime coastal installations, the Chinese Ministry of Transportation, at the end of 2005, issued guidelines for protecting hundreds of its maritime installations [12].

Both these guidelines promote the grounding and bonding practices outlined in IEC/TR 61400-24. These two guidelines have both found it necessary to mandate minimum surge protector current handling ratings for protecting their facilities. Surge protectors specied by the FAA are required to protect against a minimum of 1500 surges (10 kA 8/20 s) and against one surge (between 180 and 240 kA). The Chinese maritime guidelines concur with that assessment and added the requirement of individually-fused back-up protection paths on every phase to assure continuity of protection. This approach is called out in IEC 60364-5-53 (see Section V-H-2). Such successful eld experience can and should be consulted in protecting wind turbines. B. Grounding Practices 1) Wind turbine grounding systems should follow the recommendations of IEC/TR 61400-24 Section 9. 2) Minimum dimensions of down-conductors are given in IEC 61024-1 and in Table II of IEC/TR 61400-24. 3) A ring ground around the periphery of the wind turbine should be constructed to which the tower is connected. The steel reinforcement in the turbine tower should be exothermally integrated into the turbine grounding system. 4) The ground system should interconnect with all driven electrodes, any underground metal objects (storage tanks, pipes, etc.), and the grounding system of an operations building, if existing. In a wind farm, the ground systems of all turbines will interconnect. 5) The grounding system must be compact. Any part of the grounding system further than 30 m from the point where lightning current is injected will not help decrease the magnitude of the peak lightning overvoltage [13]. 6) The design goal is set for the grounding system of 10 or less [1], [13]. 7) The ground system must be kept in good condition at all times. Yearly inspections should be scheduled where the system can be examined for any broken or loose connections, corrosion, and/or changes to the ground impedances. C. Bonding and Shielding All systems and metal components must be bonded together into one low-impedance path to the ground as per IEC/TR 61400-24, Section 8.5. D. Rotor Blades 1) Conducting elements: Rotor blades should be constructed with conducting elements capable of conducting lightning current to the hub in such a way so as to avoid a lightning arc inside the blade. This conducting material can be either the frame of the blade or on the surface of the blade. An extensive NREL study in USA showed that blades with built-in blade conductors were far less likely to suffer catastrophic damage than blades without these conductors [2].

Authorized licensed use limited to: King Mongkuts Institute of Technology Thornburi. Downloaded on October 16, 2008 at 00:24 from IEEE Xplore. Restrictions apply.

2) Multiple Receptors: Studies have now shown that long wind turbine blades tted with an internal type lightning protection system are vulnerable to lightning strike attachment away from the receptor at the tip. The use of multiple receptors is required to provide better protection for a blade than would otherwise be provided by a single receptor located at the blade tip [14]. Rotor blades shorter than 30 m shall have at least one receptor at each shell. For blades greater than 30 m, multiple receptors are required. E. Bearings Next to blade damage, bearings are most subjected to damage from direct lightning strikes to the turbine. Statistics suggest that pitch bearings are not likely to suffer any lightning damage while in service. The bearings in the main shaft and those in the drive train are subject to damage because they are smaller and rotate rapidly. And these are more likely to sustain damage if lightning strikes when the rotor is turning [13]. To avoid this, one could stop the turbines briey during times of lightning risk. Just as turbines are taken ofine during high winds to protect them from damage, so they could also be stopped briey when lightning threatens in order to protect the bearings. F. Control System Recorded incidents of lightning damage from 3000 wind turbines of over 14 000 turbine years worth of information suggest that the wind turbine control system is the most vulnerable component contained in a wind turbine [13]. It is clear that lightning striking a single wind turbine can produce damage in wind turbines connected with metallic data cable [13]. The mix of electronic equipment in a wind turbine includes control and measurement sensors distributed throughout the turbine. Two microprocessors are often installed, one in the nacelle and the other at the tower base. These must be linked by some sort of cabling. Turbines also have SCADA connections to permit remote data monitoring and control. 1) Use ber optic lines wherever possible for telephone or SCADA connections because they are not electrical conductors. Use ber optics in both the external SCADA network, and also between the nacelle controller and the base controller. (Be sure not to use ber optic cable that has a metallic wire incorporated only for mechanical strength purposes.) 2) Where twisted copper wire must be employed for telephone or SCADA systems, a good grounding system will help mitigate high transient overvoltages. In addition, protect all I/O ports with a heavy duty (20 kA 8/20 s I-peak) data line protector wired in series and grounded directly to the metal chassis of the equipment that is being protected with a ground wire no longer than 15 cm. (This includes the junction of the SCADA system and the controller interface.) 3) Bond the sheaths of any power cables running between wind turbines solidly to the turbine ground system.

Fig. 4.

Single lightning ash to a turbine.

4) Install grounded bare copper cables in the trenches carrying SCADA cables between wind turbines. 5) Bond the shield of any SCADA cable solidly to earth at both ends. G. Electrical System Generally, output voltage from a wind turbine generator is not more than 690 V. This output voltage is stepped up to 66 kV or more by a transformer and supplied to a transmission line. 1) Transformer: Lightning surge problems of wind power generation take place in the primary (a low-voltage side) not in the secondary (high-voltage side) of a transformer, unlike conventional power distribution transformers. 2) Generator: Lightning strikes that hit the high voltage electrical grid will appear on (couple into) the low voltage (400/690-V) side of the transformer. For that reason the 400/690-V power line that connects the generator to the transformer must have a surge protector installed. Surge protectors specically designed for a 400/690-V system, and with a minimum I-peak handling capacity of 180 kA (8/20 s) should be used to protect the turbines electrical system. H. Specifying Surge Protectors (SPDs) 1) I-peak: A wind turbines main surge protectors must be capable of handling a minimum of 180 kA of surge current and have multiple individually fused back up protection paths on every phase, so that the wind turbine is never left without protection (see Section V-A above). 2) Back-up protection paths: IEC/TR 61400-24 includes a dozen graphics like this one (from page 16), which show the conguration of lightning strikes that are likely to hit wind turbines. Fig. 4 represents a single lightning ash to a turbine. A lightning ash to an object consists of more than one stroke. Although the ash shown in the diagram is composed of four distinct strokes, it is not uncommon for there to be ten or more strokes in a single lightning ash. An SPD needs individually fused redundant (back-up) protection paths so that protection to the wind turbine will be

Authorized licensed use limited to: King Mongkuts Institute of Technology Thornburi. Downloaded on October 16, 2008 at 00:24 from IEEE Xplore. Restrictions apply.

2)

3)

4)
Fig. 5. Individually fused redundant (back-up) protection paths for protection of the wind turbine.

5)

maintained even if one path sacrices itself on the rst stroke. This approach is called out in IEC 60364-5-53 and is particularly relevant to wind turbine surge protection. Fig. 5 is reproduced from IEC 60364-5-53 to show the required design. I. Nacelle External Attachments The anemometer and wind vane mounted on the rear top of the nacelle feed information to the controller and need a surge protector to prevent overvoltages from destroying controller components. The outside obstruction lights that warn aircrafts also need to be protected with a surge protector. J. Inverter Most large wind turbines use asynchronous generators, which allow direct connection to the power grid. These generators do not use inverters. In situations where inverters are used, they would need surge protection. K. Lightning Attachment Points Attention must be paid to all points that are vulnerable to lightning attachment. We should not rely on the rolling sphere method since it has never been evaluated or tested on a wind turbine (see more on this point in Section VI below). Lightning can strike almost any place on a wind turbine. One commonly overlooked area is the meteorological instrument support at the rear of the turbine nacelle. L. Personnel Safety An in-depth study of danger to wind turbine personnel during a lightning storm found three main risk categories: electric shock, exploding wire effects, and ejection of blade material. 1) The rst of these issues, electric shock, applies to personnel in a wind turbine and anyone in the vicinity of the windfarm grounding system. Although an electric shock may not kill, it can be dangerous in other ways such as causing people to lose grip when climbing down a ladder.

6)

The risk from exploding wires is only relevant to windfarm personnel. All wiring used in a wind turbine should be adequately rated to carry safely the transient current produced by a lightning ash. Finally, the hazard from ejected blade material can be prevented by the use of blade lightning protection systems [13]. Standing Order 1 for wind turbine personnel around wind turbines should be: All operatives to be evacuated immediately if lightning has been forecast for the wind farm location, or if lightning is observed in the direction of prevailing weather. Lightning warning signs should be conspicuously posted at wind turbine sites, and personnel drilled on what to do in the event of lightning. Wind farms should construct shelters where personnel can safely wait out lighting storms. Predicting lightning events in advance is key to implementing such a policy. There are instruments available that can give 2-h warning of an approaching lightning storm. When lightning current ows into a windfarm grounding system the potential of that grounding system will rise. This in turn will raise the potential on the soil surface. This is a hazard to people walking around the windfarm in the form of touch and step voltages that result from the potential gradient that will exist on the soil surface. If an emergency requires personnel to be present while there is a threat of lightning, ensure that they are wearing special shoes and gloves, which are capable of insulating them from lightning current and high step voltage. Thick rubber mats installed on the oors of working areas can also be benecial. VI. FURTHER RESEARCH NEEDED

There are a number of areas that could bear further research. A. Blade Protection All major turbine manufacturers now include lightning protection as part of blade design. This improvement has signicantly lessened the incidence of catastrophic damage from direct strikes, but it has not totally solved the problem. As manufactured blades become longer and longer, a more effective and more easily manufactured method of blade receptors must be developed [10]. B. Air Terminals For some time now, a controversy has raged between proponents and critics of the unconventional ion-emitting lightning rods. After a thorough review of available data in 2003, Uman and Rakov concluded that there are no data to support the claim that lightning elimination and early streamer emission techniques are superior to conventional lightning protection systems [15]. However, proponents continue to claim the contrary. If they did work, or could be made to do so, wind turbines would be an ideal application

Authorized licensed use limited to: King Mongkuts Institute of Technology Thornburi. Downloaded on October 16, 2008 at 00:24 from IEEE Xplore. Restrictions apply.

for these products. Interested national bodies should request manufacturers of these lightning eliminators and ESE rods to participate in controlled tests to verify the effectiveness of these products on wind turbines. Manufacturers who chose to participate should guarantee the safety of the wind turbines on which they are being tested. They should be willing to submit to such tests because any products that can demonstrate superiority in this application would have a large upside market potential. C. Main Bearing ProtectionNeed for an Equipotential Bonding Device Protecting the wind turbine bearings from direct hits to the rotor blades is an unsolved problem. At least 80% of the lightning current from a direct hit to a blade ows from the blade conductor into the nacelle bed plate through the main bearing closest to the rotor [12]. As there is evidence that such incidents may badly stress any rapidly moving bearings, it would be a design goal to reduce the amount of this charge that ows through the bearings. Because the path to the ground through the bearings is a very low-impedance path, lightning nds it very attractive. So far, efforts to develop a switching circuitry, which would shunt much of this charge around the bearings, have been unsuccessful. Any type of lightning protection device must be composed of two functional parts: a switch or some type of switching circuitry triggered by overvoltages, and a good ground connection to allow dissipation of the surge energy. The spark gap arrestor is one such switch based on a 75-year-old patent, the original and best use of which was to withstand direct lightning strikes to structures. Spark gaps were never designed to protect electronic equipment and they do not do a very good job of it. They are slower acting and characteristically have large let-through voltages (4 kV or above), which is more than enough to destroy electronic equipment. Although they are not recommended to protect the wind turbine control and electrical systems, there is a unique opportunity here for spark gap manufacturers to employ their technology in its correct role as an equipotential bonding device. It is required to create a path to the ground that can bypass the low impedances of the main bearings. Some imagination from the spark gap engineers and designers as well as engineers and designers of bearings, working in concert, might provide a useful solution to this problem. D. Attachment Points In the absence of any other method, IEC/TR 61400-24 resorted to the rolling sphere method in predicting where lightning may attach to a wind turbine. Since this method has never been tested on either high buildings or on wind turbines, a more precise technique should be developed. See Section VIII-B for more information on rolling sphere method. VII. IEC/TR 61400-24 SUGGESTED IMPROVEMENTS Following are ve suggested improvements to IEC/TR 61400-24 [25].

In the interests of reducing serious damage and death resulting from lightning strikes to wind turbines, it is hoped that these issues are addressed when the document comes up for review in 2007. A. Lightning Parameters IEC/TR 61400-24 places special importance on the parameters of positive CG lightning drawn from two studies conducted in the 1970s, the results of which appeared in Electra Magazine (issues 41 and 69. In fact, the data on positive CG lightning cited in Table I of the IEC/TR 61400-24 document were extracted directly from that rst Electra article [16]. IEC/TR 61400-24 did not realize that its second reference (Electra 69) disavowed the positive CG lightning ndings presented in Electra 41.2 This is important because this is the only reference ever used to support the IECs Class I Tests [17]. These lightning parameters have no basis in scientic facts and have been and continue to be criticized all over the world [18], [26]. Chinese scientists have calculated that if Class I test parameters actually exist in nature, they would appear no more than once every two million years [19]. A reality-based lightning protection system must be founded on a lightning model comprised of accurate lightning parameters. Any reference to Bergers positive CG lightning parameters should be struck off from the standard and a more substantive set of parameters embraced. B. Rolling Sphere Method Paragraph 8.3.1 of IEC/TR 61400-24 introduces the rolling sphere method for determining potential lightning attachment points. The rolling sphere method is often used as a guide for the placement of air terminals on structures. However, it uses a xed striking distance (typically 45 m) irrespective of the height or geometry of the structure. It has never been tested or evaluated on tall buildings and rarely on wind turbines. Dr. C. Bouquegneau, Chairman of the IEC committee which authored the recently released IEC 62305 Lightning Standards, calls the rolling sphere method a crude method which hides our insufcient understanding of the lightning attachment process [20]. The rolling sphere method may be of some use in the placement and height of air terminals, but cannot be relied upon to guarantee where lightning may attach to a wind turbine and where it will not. Reference to it should be struck off from IEC
2 The only data on positive CG lightning at the time these two articles were written had been assembled by K. Berger (a Swiss researcher) who had published his results in Electra 41. One of the co-authors of that original article (R.B. Anderson) was also the author of the Electra 69 article. He unveiled the history of those positive cloud-to-ground data reported by Berger. Page 81 of Elektra 69 explains: Berger has recently pointed out that all positive records from this station should, in fact, be classied as upward discharges. And on page 84, he amplied this as follows: Note: the parameters of positive ashes were originally analyzed by Berger et al.. (Electra 41) in 1975but on the assumption that these were downward ashes. In his new analysis, he has extended the positive ash data sample and classied all these records as upward. In consequence, there is apparently no comprehensive source of data available on the impulse characteristics of positive downward ashes.

Authorized licensed use limited to: King Mongkuts Institute of Technology Thornburi. Downloaded on October 16, 2008 at 00:24 from IEEE Xplore. Restrictions apply.

61400-24 until it has been successfully tested on wind turbines (see Section V-L). C. Lightning Protection Zone System Paragraph 8.3 of IEC/TR 61400-24 seeks to apply the Lightning protection zone (LPZ) system to wind turbines. An LPZ zone of protection is an area considered statistically safe from the direct attachment of a lightning strike as dened by the rolling sphere zone of protection model that is covered in Section VII-B. The LPZ system is difcult to justify for wind turbines because it has never been evaluated on them. Additionally, it is based on the old IEC 61312-1, which used the 10/350-s waveform as its lightning model. Two separate scientic studies directly measuring the duration of lightning strikes to wind turbines in Japan found durations to be 2530 s (nowhere near the 350 s of IEC 61312-1) [7], [21]. The 10/350-s test waveform has received and continues to receive much criticism around the world [16], [22]. It is risky to adopt such a system as the basic approach to protect installations as important as wind turbines. An attendants hazard of using this system can be seen in the next section. D. Personnel Protection Because lightning is a deadly hazard to all personnel on a wind farm, explicit and clear-cut instructions must be given tothem which are not open to interpretation or misunderstandings. Section 10 of the IEC/TR 61400-24 document entitled, Personnel Protection, lacks such instruction. Although the section begins with the observation that work should not be performed on wind turbines during thunderstorms, it then endeavors to rate the relative safety of personnel caught in various parts of a wind turbine during a lightning storm. Section 10.1.1 advises: It is highly recommended that provision is made to shelter any personnel in minimum LPZ OBthe documents reference to this is Clause 8.3.1 which states: . . . LPZ OB is determined by means of the rolling sphere model as shown in Figure 22. . . The areas marked in gray are LPZ OB. (Fig. 22 of the document is reproduced above as Fig. 6 of this paper.) To the unwary reader, Fig. 22 of the abovementioned document makes it appear quite safe for personnel to wait out a lightning storm in the tower under the nacelle or even hanging on to the anemometer on the top rear of the nacelle. However, in reality, it would not be safe at all. A much more realistic and responsible instruction is given in the British Wind Energy Associations Guidelines for Health and Safety: All operatives to be evacuated immediately if lightning has been forecast for the wind farm location, or if lightning is observed in the direction of prevailing weather [23], or from NASA: Personnel around a wind turbine can be in serious trouble during a lighting storm. The safest procedure is for personnel to vacate the site [1]. E. Analysis of Lightning Protection System Costs Sections 5.5 and 5.6 of the IEC/TR 61400-24 document deal with an analysis of the viability of lightning protection costs.

Fig. 6. Rolling sphere method for determining potential lightning attachment points. (Reproduced from Fig. 22 of Section 10 of the IEC/TR 61400-24 document, Personnel Protection).

Several pages of formulas end in the suggestion that 1.7% of the cost of the wind turbine could be justied in a lightning protection system against direct strikes to the blades in an area of low annual lightning ashes (0.2 ashes per km2 per year). As for systems to protect against the far more frequent damage caused to the controller and electrical systems, the document writers expressed their inability to use the IEC 61662 model to come up with a workable estimate. This section should be rewritten to provide more useable data to wind turbine owners. IEC TR 61000-5-2 offers useful guidance on how a grounding network should be structured and how power and control cables should be positioned (routed) with respect to the grounding network [24]. 1) Direct Strike Protection Costs: Almost all direct strikes to a wind turbine will hit the blades. When lightning hits an unprotected blade, the results are often catastrophic and very expensive to repair. It is those strikes to which IEC/TR 61400-24 gave its predominant attention. However, as mentioned above, reputable blade manufactures now include surge protection in the design of their wind turbine blades. Since blades on all new turbines will include built-in lightning protection, this protection can now be considered to be covered in the basic cost of the turbine. While direct strikes to a turbine still pose problems, the Section 5.5 conclusion that it would be feasible to spend up to an additional 1.7% of the cost of a turbine on a direct strike lightning protection system is excessive at this point in time and should be reanalyzed. 2) Cost Calculation Model for Indirect Strike Protection: A wind turbine is three times more likely to be damaged by indirect lightning than direct lightning. These damages account for more than twice the downtime caused by direct strikes to the blades. Blade protection gives no relief to that statistic. The cost of lightning-related repairs to electrical and controller systems will still be less than minor repairs to a blade ($1900) but one must not forget that replacing a generator is more expensive than replacing a blade ($50 000). The following computations are suggested as a reasonable way to determine a justiable budget for the installation of a lightning surge protection system

Authorized licensed use limited to: King Mongkuts Institute of Technology Thornburi. Downloaded on October 16, 2008 at 00:24 from IEEE Xplore. Restrictions apply.

to protect against 99% of damage caused to the controller and electrical systems. The annual cost of any lightning protection system Cam should not exceed the annual cost of lightning damage (the frequency of lightning damage Ftotal , multiplied by its cost Closses ) plus the loss in energy production CLE that results from lightning events and repairs. Cam < (Ftotal Closses ) + CLE where Cam (1)

generating capacity of a power system, missing that contribution during an outage is probably not a major public-relations factor. Nevertheless, the intangible but real-world aspects of lightning-related damage must be taken into consideration in a risk assessment.

REFERENCES
[1] C. Dodd, T. McCalla, and J. G. Smith, How to protect a wind turbine from lightning, Nat. Aeronautics Space Admin., Sep. 1983, DOE/NASA 0007-1, NASA-CR-168229. [2] B. McNiff, Wind Turbine Lightning Protection Project 1999-2001, Nat. Renewable Energy Lab., U.S. Dept. Energy, Golden, CO. [3] M. Durstewitz, C. Ensslin, M. Hoppe-Kilpper, and K. Rohrig, External conditions for wind turbine operationResults from the German 250 MW wind programme, presented at the Eur. Union Wind Energy Conf., May 1996. [4] I. Cotton et al., Lightning Protection of Wind Turbines A Designers Guide to Best Practices. Manchester, U.K.: UMIST, 1999. [5] DEFU, Recommendation No. 25, Lightning Protection of Wind Turbines. Lyngby, Denmark: Res. Inst. Danish Elect. Utilities, 1999. [6] IEA, Lightning protection of wind turbine generator systems and EMC problems in the associated control systems, in Proc. 26th Meeting Experts, Milano, Italy, Mar. 1994, Annex. XI. [7] A. Wada, S. Yokoyama, T. Numata, Y. Ishibashi, and T. Hirose, Lightning damages of wind turbine blades in winter in Japan, presented at the Int. Conf. Lightning Protect., Avignon, France, 2004. [8] D. Sullivan, Wind in the Silos, New Farm Mag., Jun. 10, 2004. [9] I. Cotton, B. McNiff, T. Soerenson, W. Zischank, P. Christiansen, M. Hoppe-Kilpper, S. Ramakers, P. Petterson, and E. Muljadi, Lightning protection for wind turbines, presented at the Int. Conf. Lightning Protect., 2000. [10] I. Cotton, N. Jenkins, and K. Pandiaraj, Lightning protection for wind turbine blades and bearings, Eur. CommissionContract JOR3-CT95-0052. [11] Dept. Transportation, FAA Standard FAA-STD-019d, Aug. 9, 2002. [12] Lightning protection guideline for RBN-DGPS stations, VTS centers, lighthouses and AIS stations, China Transportation Ministry, Oct. 20 2005, Document 411. [13] A. J. Surtees, F. D. Martzloff, and A. Rousseau, Grounding for surgeprotective devices, presented at the PES 2006 Gen. Meeting, Montr eal, Jul. 2006. [14] N. Jenkins and I. Cotton, Lightning protection of wind turbines: Further work, Eur. Commission, Contract JOR3-CT98-0241, to be published. [15] M. A. Uman and V. A. Rakov, A critical review of non-conventional approaches to lightning protection, Am. Meteorol. Soc. J., p. 1809, Dec. 2002. [16] K. Berger, R. B. Anderson, and H. Kr oninger, Parameters of lightning ashes, Electra No. 4, pp. 2337, 1975. [17] B. Glushakow and D. Neri, A call to standardize the waveforms used to test SPDs, presented at the Int. Conf. Lightning Protect., Avignon, France, 2004. [18] Data sheet no. 1, French Lightning Protect. Assoc., Dec. 2000. [19] Kiu Ji, Ma Hongda, Hu Changxin, Wang Yili, Liu Yuanyuan, and Zhang Jianping, Technology: Lightning protection research of high speed trains, China Lightning Protect. Mag., no. 6, p. 16, 2005. [20] C. Bouquegneau, Chairman IEC Technical Committee TC 81, A critical view on the lightning protection international standard , 2005. [21] Y. Y. Shiraishi, T. Otsuka, and H. Matsuura, The observation of direct lightning stroke current to the wind turbine generator system, presented at the 27th Int. Conf. Lightning Protect., 2004. [22] Task Force, Reality check initiative on the equivalency of 8/20 versus 10/350 waveforms for testing surge-protective devices, presented at the IEEE PES Gen. Meeting. Montreal, Canada, Jun. 18 2006. [23] British Wind Energy Association, Best Practice Guidelines For Wind Energy Health and Safety, Apr. 2005, p. 52 [24] IEC, Electromagnetic Compatibility (EMC)Part 5: Installation and mitigation guidelinesSection 2: Earthing and cabling, 1997, TR 610005-2. [25] IEC, Wind turbine generator systemsPart 24: Lightning protection, TR61400-24, Ed. 1.0 en: 2002. [26] IEEE, Recommended practice on characterization of surges in lowvoltage (1000 V and less) ac power circuits, C62.41.2TM-2002, p.32.

annual cost of a lightning protection system (arrived at by taking the cost of installation divided by the number of years the manufacturer warrants the system); Ftotal frequency of lightning event damage to controller and electrical systems; Closses average cost of a damaging incident to controller or electrical system; cost in lost energy production while the turbine is CLE ofine. CLE can be computed as CLE = (FLED + FLEN )CE (2)

where FLED frequency (in hours) of lost production occasioned by damaging lightning incidents; FLEN frequency (in hours) of lost production occasioned by non-damaging lightning incidents3 ; cost of one hours actual energy production by the CE turbine (the amount of money for which one hours worth of the turbines production can be sold, adjusted to take into account expected percentage production losses due to mechanical and other nonlightning associated factors). Using US dollar gures and inputing an average $1000 repair cost per damaging incident, and a CE value of 0.996 per hour adjusted actual energy output of a 1-MW wind turbine we get Cam 10 < (3 $1000) + (266 + 71) $0.78 Cam 10 < 3000 + 208 Cam 10 < 3208 Cam < $3208 for a lightning protection system with a oneyear warranty. (For a lightning protection system with a ten-year warranty, an owner would be economically justied in spending $32 080.) These proposed computations only consider the monetary costs. For a new technology that depends on public acceptance of other nonmonetary aspects, such as landscape spoiled by wind farms or publics perception of its reliability, a conspicuous interruption of service caused by extensive lightning damage can make the technology an easy target for detractors. Granted that as long as wind turbines are only a small fraction of the
3 The NREL 2002 report documents the phenomena of frequent reports of faults where there was no hardware damage. This was attributed to indirect lightning causing controller problems at night. No damage was done but the turbine went ofine and the controller had to be reset in the morning. This resulted in hundreds of lost turbine hours.

Authorized licensed use limited to: King Mongkuts Institute of Technology Thornburi. Downloaded on October 16, 2008 at 00:24 from IEEE Xplore. Restrictions apply.

Bruce Glushakow (M05) was born in the United States but has spent half his life living and working in Europe and Asia. After establishing an electrical contracting company in New York he joined MCG Electronics Inc. His main research interests include standard technology. He has authored a number of articles and books including: Lightning and Surge Protection Guideline for Electronic Equipment, published in Chinese in 2003; A Call to Standardize the Waveforms Used to Test SPDs, presented at ICLP 2004,

Avignon, France; 10/350 under the microscope, EC&M Magazine (online version), 2004; Lightning protection for telecommunications substations in 2004; Multiple-source lightning transients:The cause of microwave and telecommunications equipment damage, 2005; The IEC 63205 Lightning Protection Standards: Can they be xed or is cancellation the only recourse? presented to IEEE SPDC in Clearwater, FL, in 2006.

Authorized licensed use limited to: King Mongkuts Institute of Technology Thornburi. Downloaded on October 16, 2008 at 00:24 from IEEE Xplore. Restrictions apply.