You are on page 1of 10

Review of claimed enhanced lightning protection of buildings by early streamer emission air terminals

D.Mackerras M.Darveniza A.C.Liew

Indexing terms: Lightning protection of buildings, Early streamer emission air terminals

Abstract: Simple air terminals are earthed conducting objects able to launch a connecting streamer or leader discharge that intercepts a downward lightning leader discharge and diverts to itself the lightning strike. Early streamer emission (ESE) air terminals are claimed to initiate the connecting streamer earlier in time than would a simple air terminal in the same position, and are therefore claimed to be able to attract the lightning discharge from a larger distance than would a simple air terminal. The chief objection raised is that this implies that a streamer or leader discharge from an ESE air terminal is able to continue propagating when the electric field ahead of the advancing tip of the streamer or leader is below the minimum value that would apply to a streamer or leader from a simple air terminal. Once a streamer or leader discharge has propagated into the space remote from the air terminal, its further progress depends upon the supply of energy from the electric field in the space near the tip of the discharge and upon the dielectric properties of the air undergoing breakdown. As neither of these factors can be influenced by the air terminal, it is concluded that it is not possible to gain a significant improvement in lightning interception performance by causing the early emission of a streamer from an air terminal.

the lightning interception performance of an air terminal by causing the early emission of the connecting streamer or leader discharge from the air terminal. A recent proposal in the USA to produce a National Fire Protection Association (NFPA) Standard on early streamer emission air terminals for lightning protection (Draft Standard NFPA 781 - F93TCR) has made it necessary to examine the basis for the claims of enhanced lightning protective properties. In examining these claims we have studied an extensive literature [1, 41] relevant to lightning phenomena and to both standard and nonstandard lightning protection methods. The situation at the end of 1995 was that the NFPA had decided not to issue NFPA 781. However, there are proposals to consider recognition of ESE air terminals in the IEC Standards on lightning protection that are likely to be based on material similar to that contained in Draft Standard NFPA 781 - F93TCR. It is our view that there are at least three incorrect assumptions in the procedures that lead to claims of improved lightning interception performance, for ESE air terminals.

Fig.1 Stages in lightning attachment process: commencement of connecting streamer

Introduction

At time t1 a connecting streamer commences from building

Our objective is to review critically the claims for enhanced lightning protection characteristics of early streamer emission (ESE) air terminals for lightning interception. This review is intended to be a contribution to the discussion of the possibility of improving
IEE, 1996 IEE Proceedings online no. 19960649 Paper received 15th March 1996 D. Mackerras and M. Darveniza are with the Department of Electrical and Computer Engineering, University of Queensland, St. Lucia, 4072, Australia A.C. Liew is with the Department of Electrical Engineering, National University of Singapore, Kent Ridge, Singapore, 0511
IEE Proc.-Sci. Meas. Technol., Vol. 144, No. 1, January 1997

2 Review of the physics of the lightning attachment process for simple air terminals Descriptions of the physics of the lightning attachment process have been given [1, 14, 15, 22, 40]. The term air terminal refers to any metallic object with an electrical connection to earth to which a lightning strike may attach. The attachment process starts when the downward first leader of the ground flash has reached a point about 100 200m above the ground or the building that is about to be struck. The first leader typically takes between about 10 and 40ms to progress from cloud to ground, and the attachment process
1

occupies the last 1ms or less of that time. The situation in the last stage of the downward progression of the leader is shown in Figs. 13.

[1215, 31]. The following relationship between ds in metres and I in kA is attributed in the IEEE Working Group Report [24] to Love [31]: This and other relationships between ds and I and experimental points from Eriksson [13] are shown in Fig. 6.1 of [40]. The significance of the striking distance for the attachment process is that if a downward leader tip goes within a distance ds from a point on a building, and that point is capable of launching a connecting upward leader, then that point will be the one to which the lightning channel connects. Since the point of attachment of the lightning channel on the building has served as the collection point for the strike, the term collection distance could appropriately be used for the distance denoted ds in Fig. 3 when discussing building lightning protection. The basic assumption in conventional lightning protection is that the collection distance is equal to the striking distance, whereas the essential claim made for ESE air terminals is that their collection distance can be caused to be greater than the collection distance for simple air terminals. The term streamer is used here to denote a spatially diffuse discharge process with relatively low current density and high resistivity. Details of positive streamer phenomena, which are extremely complex, have been investigated by the Les Renardires Group [29]. There will usually be a transition from a streamer process to a leader process with a thermally ionised core of high current density and low resistivity. There is considerable uncertainty concerning the extent to which the discharge processes in a positive laboratory streamer/leader a few metres long correspond to those in a positive upward streamer/leader tens of metres long in natural lightning. We assume that the processes are similar, at least in the initial parts of the discharge, provided that the initial value and rate of change of electric field for the laboratory test are similar to those in natural lightning. 3 Electric eld aspects of attachment process

Fig.2 Stages in lightning attachment process: connecting streamer


approaching leader
At time t2 connecting streamer and leader approach each other

Fig.3 Stages in lightning attachment process: connecting streamer meeting leader


At time t3 connecting streamer and downward leader meet at junction point. Striking distance ds is defined as shown

The leader channel current results in the transfer of charge (usually negative) from the charged regions of the thundercloud towards earth. The lower end of the leader channel is at a negative potential somewhat below the potential of those regions, probably in the range 10 100MV. The attachment process terminates when an upward connecting leader meets the downward leader and the return stroke propagates up the leader channel. The peak current I in the return stroke is mainly derived from the charge on the leader channel, so these two quantities are related. Berger [5] has shown that if Q is the charge transferred to ground in the first 1 ms or so of the return stroke, presumably mainly derived from the leader channel, then the best fit to the available data for 89 negative first strokes is where I is in kA and Q is in coulombs. Returning to the situation just before the connection is made, the electric field at parts of the building rises as the leader approaches, and reaches a value at time t1 when streamers arise from points on the building, as shown in Fig. 1. Shortly after, one of the streamers leads to the development of an upward leader, denoted the upward or connecting leader which propagates towards and finally reaches the tip of the downward leader, shown as occurring at time t3 in Fig. 3. The existence of other unsuccessful upward leaders is sometimes seen in photographs of lightning striking the ground [14, 40]. The striking distance ds shown in Fig. 3 is described [14, 40] as the distance between the object to be struck and the tip of the downward leader at the instant (time t1 in Figs. 13) when the connecting leader is initiated from the object. Since the electric field between the leader and the building just before the connection is made depends on the leader charge, and the leader charge is related to the peak current in the return stroke, there is a relationship between ds and I
2

The direction of progression of the downward leader and the initiation of streamers leading to the development of an upward connecting leader are probably guided mainly by the electric field in their vicinity. Thus, if the electric field in the space between the downward leader and the building could be calculated for successive positions of the leader as it progressed towards earth, then the path of the leader could be predicted. The launching of the connecting leader could be predicted from the electric field around the building. Such a procedure is generally considered impractical for routine application to the lightning protection of buildings. However, a consideration of the electric field between a building and a downward leader in geometrically simple cases can provide valuable insight into the attachment process. Dellera and Garbagnati [11] have described a method of simulating the progression of the downward leader and the upward connecting streamer/ leader for natural lightning to tall objects such as masts and electric power lines using electric field calculations based on the charge simulation method. Rizk [36] has modelled the process in even more detail. Fig. 4 shows the approximate form of the equipotential lines in the situation when the tip of the downward
IEE Proc.-Sci. Meas. Technol., Vol. 144, No. 1, January 1997

leader is 300m above ground and the leader is descending vertically along the axis of a cylindrical building 100m high and 50m in diameter. This is the situation 1 to 2ms before the return stroke commences. The leader, represented as a charged object 10m in diameter and 100m long, is shown projecting from a horizontal sheet of charge at a height of 400m. Both the leader and the upper charge sheet are at a potential of 20MV. The equipotential lines are 0.5MV apart. The plot shows how the electric field is enhanced at the upper outer edges of the building, and indicates an electric field enhancement factor of about two near the edge of the roof, compared with the average field (20/200 = 0.1MV m1) between the leader and the roof of the building.

As the leader approaches the building, say to within 100 m, the electric field at the upper parts will increase, and streamers will be launched upward from points of electric field enhancement on the building roof. This will involve a transition from a corona type of discharge to a streamer. One or more of these streamers will undergo transition into one or more upward leaders. These leaders will advance into regions of diminishing field, and will be unable to obtain sufficient energy from the field to continue their progress. There exists a minimum electric field required for the continued successful propagation of a leader. The criterion has been stated by Golde [14] in the form that the minimum average field is about 0.5MV m1. However, Dellera and Garbagnati have restated the criterion (private communication, 1995) in the form that a field of 0.5MV m1 or more is required in the region ahead of the advancing leader to ensure continued propagation. The field remote from the advancing tip of the discharge is not considered relevant to its progression. Thus, in the example shown in Fig. 4, the tip of the downward leader will have to advance to within about 40m of the building top before the average field reaches 0.5MV m1 (using the Golde criterion) and a connecting leader can be launched. When this happens, the leader will probably originate from the edge of the building roof, in a region of electric field enhancement, despite the fact that the midpoint of the roof of the building is the closest point to the leader tip. This should be kept in mind when determining strike attachment points and priority should be given to placing air terminals at points of electric field enhancement such as the upper outer edges and corners of buildings. Selection of a critical average electric field lower than 0.5MV m1 can be accommodated by reducing the assumed potential of the leader and the electric field. Otherwise the discussion is unchanged. 4 National codes of practice for lightning protection Most countries have a national standard or code of practice on lightning protection. Examples include: AS1768-1991 for Australia [1], BS6651-1992 for the UK [6], the NFPA (National Fire Protection Association) code used in the USA [33], the Norm Francaise in France [34] and CP33:1985 for Singapore [37]. The International Electrotechnical Commission (IEC) has produced codes IEC 1024.1-1990 and IEC 1024.1.11993 [25] for general use in countries that do not have their own code, or as a guide to revision for those countries that do have their own code. There are major aspects of interception lightning protection in which the recommendations of the various codes are in agreement. There is general agreement that the major components of a lightning protection system are: air terminals, downconductors, earth electrodes, equipotential bonding and overvoltage protection. Most codes are confined to describing only conventional or standard methods of protection, avoiding reference to nonstandard methods. 5 Rolling sphere method of predicting lightning strike attachment points and the collection surface concept Many recent codes have adopted the rolling sphere method of determining probable strike attachment
3

Fig.4 Approximate form of equipotentials at instant when a downward


leader tip is 300 m from the earth and is 200 m from the top of a building represented by a conducting circular cylinder 100 m high and 50 m diameter
The leader is represented by conducting circular cylinder 100m long and 10m diameter, is at a potential of 20MV and the equipotentials are 0.5MV apart. (Field map provided by W. Zaengl)

It is to be expected that the electric field at a point in the space close to the upper outer corner of the building will increase as the point approaches the corner. This could be visualised in terms of the electric field lines (normal to the equipotential lines) crowding closer together as the corner is approached. Electric field enhancement at exposed upper parts of a building will generally result in dielectric breakdown of the air nearby. This type of dielectric breakdown is known as a corona, and will occur anywhere that sharp edges are present at exposed upper parts of a building or on the ground under the thundercloud. Consequently, the air near the building protuberances and edges will be partly ionised, and the air surrounding the building will carry a net positive charge. Soula [38] estimated that the air up to about 600m height under a thundercloud had a charge density of about 1nC m3. A net positive charge density of this magnitude will cause the electric field to increase with height at the rate of about 340Vm1 per metre increase in height and will therefore have only a second-order effect on the electric field in the situation shown in Fig. 4.
IEE Proc.-Sci. Meas. Technol., Vol. 144, No. 1, January 1997

points. This method has been described by Lee [1, 27] where a sphere radius of 45m is used, corresponding to the striking distance and collection distance (for simple air terminals) for a peak lightning current of about 10kA. The rolling sphere method is derived directly from the electrogeometric method of predicting lightning strike attachment to phase wires and shield wires of electric power transmission lines. The validity of the rolling sphere method for predicting lightning strike attachment to buildings is supported by the success of the electrogeometric method in explaining shielding failure rates in transmission lines. The essence of the argument in favour of the rolling sphere method is as follows. We consider a particular peak lightning current, for example 10kA, and the corresponding striking distance and sphere radius of 45m. We allow the sphere to roll over and around the building in any direction with the sphere surface touching the building. Each radial line from the point of sphere contact with the building to the centre of the sphere represents a possible path of the lowest portion of the leader channel. The surface traced out by the centre of the sphere defines all possible points in space from which a leader can attach to the building. This surface, shown in Fig. 5, could be regarded as the lightning collection surface for the building.

distance for strikes to flat surfaces smaller than the collection distance for corners and edges. 6 Comparison of standard and nonstandard methods of lightning protection In all national standards or codes on lightning protection, there is a convergence of views on the function of air terminals, namely, that any simple metallic object will serve the purpose provided it is adequately earthed or bonded to an earthed part, and has adequate metallic cross-sectional area to carry the full lightning current. That is, the effectiveness of the air terminal in launching streamers and a connecting leader and acting as the connection point for the lightning channel, is attributed entirely to its position on the building, and is independent of the fine details of its shape or structure on a scale of a few tens of centimetres. Thus the basic philosophy in designing the interception system is to determine all probable strike attachment points on the building using a method such as the rolling sphere method. Then, simple air terminals are required at all such possible attachment points, unless existing metallic structures will serve adequately as an air terminal. Inconspicuous air terminals are usually preferred and rods protruding above the building are avoided except where needed to protect an adjacent surface or volume. Such systems will have a large number of points on the upper parts of the building that will serve as launching points for upward streamers and connecting leaders and therefore for the attachment of the lightning channel to the building. They therefore cater for the arrival near the building of lightning leaders from all possible directions. The air terminals connect directly to downconductors that lead the lightning current to the earth electrodes. In contrast, special-purpose or proprietary interception lightning protection methods are generally based on the claim that it is possible to enhance the protective properties of an air terminal by various design features. These protection systems are denoted nonstandard because they are not included in major standards for lightning protection. Reports favourable to the use of nonstandard systems include [24, 1720, 23, 35]. Reports arguing against the effectiveness of nonstandard systems include [710, 14, 26, 32, 39, 41]. Depending on the origin of the particular device, various means of enhancing the protective property have been used including radioactive ionisation devices, nonradioactive ionisation devices, special shapes of the air terminal, creating small sparks between two parts of a device and causing voltage pulses on some part of the device. All special purpose proprietary air terminals are claimed to provide enhanced lightning protection by causing the emission of an upward streamer/leader that will propagate towards the tip of the downward leader at an earlier stage in the attachment process than would occur for a simple air terminal in the same position. The enhanced protective property may be expressed in the form that the special purpose device is equivalent to a simple air terminal of greater height than its physical height. Alternatively, the advantage may be stated in terms of an increased collection distance compared with a simple device in the same position. Thus, the special purpose device is claimed to attract the lightning channel to itself from a greater distance than would a simple air terminal in the same
IEE Proc.-Sci. Meas. Technol., Vol. 144, No. 1, January 1997

Fig.5 Lightning leader collection surface for flashes with a particular


Lightning downward leaders reaching portion of collection surface marked aaa will terminate at point A etc.

striking distance and collection distance, ds

Once a leader tip has touched the collection surface, the electric field between the tip of the leader and the building will be dominated by the charges on the leader channel and the charges induced on the building. Charges elsewhere will have relatively little effect on the electric field in this space. Once the leader tip has reached this surface, the average electric field will have reached about 0.5MV m1, ensuring that a connecting leader will occur. Thus, the rolling sphere method does account for electric field requirements in an indirect manner. The procedure can be repeated for any other peak lightning current and corresponding sphere radius. The main deficiency of the method is that it tends to lead designers to expect strikes to flat building surfaces that have, in actuality, a very low probability of being struck. This deficiency is partly compensated for in [1] by recommending that priority be given to placing air terminals at exposed corners and edges of buildings, usually in the form of a conducting strip. An alternative approach would be to assign a collection
4

position; thus the special purpose device will compete successfully with the simple device even though it is further from the tip of the downward leader than the simple device. A characteristic feature of a complete lightning interception system using special purpose devices is that only one, or at most, a very small number, of prominent air terminals will be used. These systems will sometimes employ one insulated downconductor per air terminal, with earthing systems separate from the building, or insulated downconductors for part of the height of the building connecting to reinforcing steel at an intermediate height [35]. It may be claimed that there is value in keeping the lightning current away from the fabric of the building. Darveniza [8] has expressed major reservations about the claim that an insulated downconductor can isolate the lightning current from the metalwork of the building. 7 Types of early streamer emission air terminals

situations show that the thickness of the corona sheath around prominent elevated objects will be of the order of centimetres. The fine details of the shapes of devices will be surrounded by a blanket of ionised air and any ions emitted by the special purpose device will be in competition with a much greater quantity of ions from all nearby prominent objects. Any possible external effect of sparking on or within the device will be blocked off by the shielding effect of the corona sheath of ionised and weakly conducting air. It is therefore physically unrealistic to expect that the lightning protection properties of the device will be enhanced.

7.3 Early streamer emission devices


In the class of devices specifically labelled ESE, voltage pulses are produced on one part of the device by means that are a commercial secret. Air terminals that are claimed to operate by producing small sparks also must presumably produce voltage pulses on some part of the device. From the limited information available in the literature on one type of ESE device, it appears that the energy to produce the voltage pulses and sparks must be derived from the increasing electric field during the approach of the downward leader.

7.1 Radioactive air terminals


Heary et al. [23] have presented the results of HV laboratory tests of air terminals with and without radioactive sources which show that the radioactive devices have a height advantage of the order of 10cm where discharge path lengths are in the order of a metre. It is argued that this effect can be extrapolated to a height advantage substantially greater than 10cm in effective height of the air terminal under natural lightning conditions. However, Wu et al. [41] were unable to detect any height advantage for radioactive terminals in HV laboratory tests of a proprietary radioactive air terminal using discharge lengths of about 5m. Possibly a small effect would have been observed if a constant electric field with superimposed impulse had been used as in [23]. Assuming that a height advantage of a few centimetres can be achieved by a radioactive device, it is then necessary to question whether this effect can be extrapolated to a larger height advantage under natural lightning conditions. The mechanism for the height advantage appears to be the emission of ionising (such as alpha) particles from the radioactive source causing ionisation and increased conductivity of the air for a distance of a few centimetres (the range in air of the emitted particles) from the air terminal. This mechanism is the same for laboratory and natural lightning conditions, so there is a sound reason to believe that the height advantage will be the same for the two situations, and that the height advantage will be of the order of centimetres.

Fig.6 HV laboratory test used to compare ESE air terminal with simple
air terminal (Franklin rod)
Negative switching impulse is superimposed on a steady negative field. Time advantage T is the time difference between the times of initiation of current from the two terminals. See Berger [2] for details of the procedure

7.2 Nonradioactive ionising devices, sparking devices and special shapes of air terminal
There is a class of special purpose air terminals based on the creation of ionised air around the device by sparking or otherwise, or on the special shape of the device. However, during the close approach of the downward leader, all prominent conducting earthed objects on the top of a building will be in a high ambient electric field and there will be local electric field enhancement. Consequently, many objects will be emitting ions in corona discharges in sufficient quantities to prevent the local field adjacent to the objects from rising above the dielectric breakdown field for air, about 3MV m1. Order of magnitude calculations in typical
IEE Proc.-Sci. Meas. Technol., Vol. 144, No. 1, January 1997

8 Methods of testing ESE devices in a HV laboratory to determine the collection distance In HV laboratory tests comparing a particular type of ESE air terminal with a simple air terminal (Franklin rod), Berger [2] found that the ESE air terminal emitted a positive streamer between 10 s and 50 s earlier in the front time of the impulse voltage than the simple air terminal. A simplified illustration of one form of test is shown in Fig. 6 where a time difference, T, is measured between the times of initiation of the currents in the ESE air terminal and the simple air terminal. In another form of the comparative test, the two types of
5

air terminal were tested alternately, with a total of 100 impulses applied to each type. The value of T was the difference between the averages of the times of discharge initiation over all the tests of each type of air terminal. Negative switching impulse voltages were used with a superimposed steady electric field. The rate of increase of electric field during the switching impulse front was about 1kV/m per microsecond. The tests were carried out using a very large upper electrode, area about 300m2, with a steady applied negative voltage on which was superimposed the negative switching impulse voltage. It was stated that voltage pulses were produced and applied to an undisclosed part of the ESE terminal, but no information was given about the voltage pulses in terms of their timing, polarity, waveshape, repetition frequency or point of application, making it very difficult to assess the significance of the result obtained. In the tests carried out at Les Renardires High Voltage Laboratory, the speed of propagation of streamers and leaders could be measured. Measured initial speeds were about 2cm/s. The procedures noted above were also specified in Draft Standard NFPA-781 - F93 TCR. Berger [3] has argued that the observed time advantage T would result in the streamer from an ESE terminal, under natural lightning conditions, propagating (presumably after transition to a leader) between 10m and 50m further than a streamer/leader from a simple air terminal. This implied an assumed constant streamer/leader propagation speed of 106 m s1. This assumed propagation speed is stated explicitly without supporting evidence in Draft Standard NFPA 781F93TCR, where the distance advantage, L, for an ESE air terminal is calculated in metres as
where T is in seconds. Berger [3] argued from the above that the electrogeometrical model would be modified by increasing the striking distance by L above the normal striking distance for a given peak current when using an ESE terminal, implying that an ESE air terminal could attract lightning from a greater distance than a simple air terminal, the difference between the two distances being between about 10m and 50m.

9 Examination of claimed collection distances for ESE air terminals Consider the physical situation of an ESE air terminal on the roof of a building during the last stages of the approach of a downward leader towards the building. If the claims noted above are correct, then the ESE air terminal will be able to launch streamers and a connecting leader from point A in Fig. 7 at an earlier time than point B on the building can launch a connecting leader. It is implied that the tip of the connecting leader from point A will continue to advance at the assumed speed and will maintain a distance advantage over the tip of the connecting leader from point B. We use here the Golde criterion, although it will be shown that our conclusions are not dependent on the particular criterion adopted. We assume that the tip of the downward leader, point C, is at a potential of 20MV and has arrived at a particular instant of time at a point 40m from point B and 60m from point A. The average field between B and C will be 0.5MV m1, so the required condition for the launching of a connecting leader from B has been met. The leader from B will therefore be able to obtain enough energy from the electric field to maintain its progress even though the ambient electric field is diminishing as the distance from B increases. Thus, the striking distance and collection distance for simple air terminals for this stroke is ds = 40m. Let us assume that the ESE air terminal at A is claimed to have a 20m advantage in collection distance over a simple air terminal, so L = 20m in Fig. 7. Then, at this moment, the ESE air terminal should be able to launch a connecting leader that has the same probability of completing the connection to the downward leader as the leader from B. So its collection distance should be about 60m. Once the leader from A has been launched, it will be progressing into a region of diminishing ambient electric field strength. The leader will be subject to the same laws governing progress as any other leader. The average field between A and C will be 0.33MV m1 which is less than the 0.5MV m1 required to ensure the progress of the leader on the basis of the Golde criterion (see Section 3). The ESE air terminal will be unable to modify the dielectric characteristics of the air between A and C. The leader from A will therefore be unable to obtain sufficient energy from the electric field to maintain its progress towards C. It will therefore be unable to compete successfully with the leader from B. The above discussion is independent of the actual numerical values chosen. Stated in more general terms, it is physically unreasonable to expect an upward leader to continue its progress towards the tip of the downward leader if it is unable to obtain enough energy from the electric field to do so. The ability to obtain this energy is related to the average field between the downward leader tip and the point launching the upward leader. On the basis of the criterion stated by Dellera and Garbagnati (see Section 3), the ability to obtain the energy is related to the electric field ahead of the advancing leader. The actual numerical value of the field and the criterion selected do not affect the conclusion. All leaders, once they have progressed into air beyond their launch point, are subject to the same laws governing their progress. It follows that the striking distance and collection distance is a direct consequence of these laws and the properties of
IEE Proc.-Sci. Meas. Technol., Vol. 144, No. 1, January 1997

Fig.7 Building equipped with ESE air terminal at A and simple air terminal at B
Tip of downward lightning leader at C is within normal striking distance and collection distance (ds) for B (earthed metallic object at edge of building) and within claimed collection distance (ds + L) of A 6

the air and is independent of any special features of the air terminal launching the leader. 10 Comments on selected streamer/leader propagation speed Berger [4] in his Section 4 stated that: ... Early Streamer Emission (ESE) air terminals ... have been designed to trigger an upward leader earlier than that issued from a Franklin rod ... and further states that: ESE conductors ... use an electrical triggering device which reduces strongly the duration of the transition from streamer to leader (this dead time limits the efficiency of any Franklin rod). We therefore consider that early streamer emission air terminals are actually intended to operate by early emission of a leader and that the relevant propagation speed is that of the leader since the streamer-to-leader transition has taken place early in the process. We consider that the selection of 106 m s1 (100cm/s) as a constant streamer/leader speed is not supported by the available evidence. The speed of 106 m s1 is about the average speed of propagation of certain types of downward stepped leader in natural lightning (see reviews in Chaps. 5 and 12 of [40]) but is larger than propagation speeds in the early stages of laboratory positive streamer/leader discharges. Measurements at Les Renardires High Voltage Laboratory [2830] and by Gorin and Shkilev [16] indicate leader speeds in the range 1 3cm/s in the early stage of leader propagation during long gap breakdown under switching impulse conditions. Fig. 4.1.11 of [29] shows real leader velocity to be in the range 1 2cm/s for gap spacings in the range 2 10m. Fig. 4.2.1 of [29] shows the real velocity of the leader to be in the range 1.5 2.5cm/s for the time interval 200 400s during the discharge. Fig. 3.4.7 of [28] shows the leader length increasing with time at a rate of about 1.5cm/s over a period of about 400s. Fig. 3.6.2 of [28] shows the axial leader velocity above 20cm/s only during the last 4s before breakdown, and exceeding 100cm/s only during the last 2s before breakdown. This is in general agreement with Fig. 6 of [16] where the progress rate of the streamer zone boundary is plotted against the steepness of the front of the voltage impulse waveshape. Progress rates reach about 100cm/s only for a front steepness of about 1000kV/s. These HV laboratory tests are considered to simulate conditions at an air terminal under natural lightning conditions, as the rate of rise of electric field is reasonably close to that occurring under natural lightning conditions. The simulation is improved if the negative switching impulse is superimposed on a steady negative field [4]. For these reasons we consider that the laboratory measurements indicate at least the correct order of magnitude of the initial speed of the upward connecting leader under natural lightning conditions. The speeds noted above agree with those reported by Berger [4]. He reported a leader speed of 20km s1 (2cm/s) for leaders from both ESE air terminals and Franklin rods (simple air terminals) in HV laboratory tests. These tests were carried out using switching impulses with a superimposed steady electric field, as noted in Section 8. Thus, the available laboratory measurements indicate that leader speeds are typically a few cm/s initially and do not reach 100cm/s until the final stages of the attachment process, when the electric field across the
IEE Proc.-Sci. Meas. Technol., Vol. 144, No. 1, January 1997

unbridged part of the gap is rising rapidly. For these reasons we conclude that an increased collection distance for an ESE air terminal under natural lightning conditions cannot validly be calculated by measuring a time advantage as described by Berger [2] and in Draft Standard NFPA 781-F93TCR, under HV laboratory conditions, and multiplying this time by 106 m s1. If all other assumptions were valid, the distance advantage L in metres would be more correctly calculated with T in seconds as

11 Proposed alternative HV laboratory and eld tests for ESE air terminals We consider that a more appropriate way to establish a collection distance advantage for an ESE air terminal would be to place an ESE air terminal and a simple air terminal a few metres apart in a HV laboratory test as described in Section 8. The ESE air terminal should initially be at the same height as the simple air terminal and be lowered in stages until the probability of initiation of discharges from each terminal became equal. The proposed test is illustrated in Fig. 8. The difference in height of the two terminals would then be a physically realistic measure of the increase in collection distance for the ESE terminal.

Fig.8 Proposed HV laboratory test to compare performance of an ESE


air terminal and a simple air terminal (Franklin rod)
Height difference (H) is adjusted until there is equal probability of a discharge to each terminal

Fig.9 Proposed ESE terminal field test


Test aims to determine whether or not terminal can protect a surrounding object (represented by suspended wire) within claimed distance (L) by which ESE collection distance exceeds striking distance for simple air terminals

A similar requirement should be placed on comparative field tests of simple and ESE air terminals under natural lightning conditions. Tests of this type have been described by Gumley [19] and are proposed in the Draft Standard NFPA 781-F93TCR. We propose a type of field test shown in Fig. 9 where an ESE air terminal is required to protect a surrounding elevated
7

conductor in the form of a suspended wire insulated from ground and placed at a radial distance of about 0.9 L from the ESE terminal. Currents from the ESE air terminal and the suspended wire would be recorded. The objective would be to record the streamer and leader currents for close lightning that does not strike the ESE air terminal or the suspended wire, and to record return stroke current for strikes to either the ESE air terminal or the suspended wire. If the ESE claims are correct, the ESE air terminal will have larger (and earlier) emission currents than the suspended wire for all nearby lightning strikes that do not strike either the wire or the ESE terminal, and only direct strikes to the ESE air terminal should be recorded.

point B on the surface is equidistant from A and the ground. The radial distance from the centre line to B is Rpn, the protective radius for a simple air terminal at A. Vertically descending leaders within a radial distance Rpn of the centre line will strike point A, leaders outside this distance will strike the ground. With ESE terminals having a claimed collection distance ds + L, the situation shown in Fig. 10b applies, and the protective radius is increased to Rpe. Values of Rpe as a function of height h of the ESE air terminal and the value of L are shown in Fig. 11, taken from Draft Standard NFPA 781-F93TCR. For example, for an ESE air terminal 10m high, having L = 30m, Rpe is about 66m. This concept is then transferred to buildings of arbitrary shape as shown in Fig. 12. However, a concept developed for tall slender objects cannot validly be transferred to objects of arbitrary shape, because the boundary conditions for the electric field (or for electrogeometrical modelling) have been changed.

Fig.12 Reproduction of Fig. B-2.2 from Draft Standard NFPA 781F93TCR comparing protective zones of a Franklin rod with those of an ESE air terminal whose effective height is claimed to be increased by L

Fig.10 Geometrical construction used to establish the protective radius for a tall slender object
a Simple air terminal at A b ESE air terminal at C

Fig.13 Building with ESE air terminal which fails to protect edge of
building at B
Claimed collection distance 75m (L = 30m) Protective radius 66m Edge at B has normal striking and collection distance, 45m

Fig.11 Reproduction of Fig. 4-2.3(b) from Draft Standard NFPA 781F93TCR giving, for protection level II, values of Rpe as functions of height h for various values of L
L = 0, this curve has been added to original

12 Comments on the protective radii claimed for the ESE air terminals Draft Standard NFPA 781-F93TCR specifies a method of calculating the protective range of an ESE air terminal based on the protective radius concept. This concept is a corollary of the collection volume concept for tall slender objects illustrated in Fig. 10. According to this concept, for a simple air terminal as shown in Fig. 10a, a spherical surface is drawn with centre A, the tip of the slender object, and radius equal to the striking distance, ds, for a selected peak current. The
8

Furthermore, assuming for the moment that the claimed collection distance for the ESE air terminal is correct, the ESE air terminal in Fig. 13 could only protect the building edge to a distance of about L since the edge will have its own normal collection distance. Applying the rolling sphere method to this situation as shown in Fig. 13 shows that a 10-m high ESE air terminal with a claimed collection distance of 75m (45m + 30m) cannot protect the edge of the building with its own collection distance of 45m when the edge is 66m from the ESE terminal. The lightning collection surface for point A (the ESE terminal) is shown as aaa. The lightning collection surface for the building edge (B) is shown as bbb. All leaders reaching surface bbb, from whatever angle of approach, will connect with the building edge rather than the ESE terminal. Support for the view that nonstandard air terminals in general and ESE air terminals in particular cannot protect buildings in accordance with the claimed
IEE Proc.-Sci. Meas. Technol., Vol. 144, No. 1, January 1997

enhanced zones of protection (as defined by the claimed protective radii noted above) is provided by the observations of Hartono and Robiah [21] who have observed lightning strike attachment points on buildings in Malaya well within the claimed zones of protection. Their findings are as follows. It was found that the majority of buildings which were installed with the non-standard lightning protection systems have been repeatedly struck by lightning with a significant number of these strikes falling well within their claimed enhanced zone of protection. The failure of the non-standard lightning protection system to prevent leader attachments to buildings is well known and has been documented. Some countries have even banned the use of the non-standard lightning protection systems for safety reasons. The observed protection failures support our view that the collection distances of nonstandard (including ESE) air terminals is not significantly different from the collection distances of simple air terminals. The observations of Hartono and Robiah [21], mainly based on tall buildings in Kuala Lumpur, show that neither standard nor nonstandard air terminals mounted away from the building edge are able to protect the building edge. A conductive strip mounted along the upper outer edge, with bonding to building reinforcing, appears to be the essential minimum requirement for protecting tall buildings. 13 Conclusions

Our conclusion that nonstandard air terminals cannot have collection distances up to 50m greater than the collection distances of simple air terminals is supported by observations in Malaya by Hartono and Robiah [21] of lightning strikes to unprotected parts (usually upper outer edges) of buildings well within the claimed zones of protection of the nonstandard air terminals. 14 Acknowledgments

An earlier draft version of this paper was sent to several experts worldwide. We gratefully acknowledge their comments and contributions, many of which have helped us to improve the paper. In particular, we acknowledge the assistance and information provided by the following persons: L. Dellera, A.J. Eriksson, E. Garbagnati, Z.A. Hartono, A.E. Pedersen, H. Steinbigler, M.A. Uman, J. Wiesinger and W. Zaengl who provided the equipotential plot in Fig. 4. C.B. Moore supplied evidence supporting our conclusions from field tests of ESE air terminals in New Mexico. J.J. Lowke and R. Morrow supplied evidence from theoretical analyses of electrical discharges supporting our statements concerning the propagation of streamers and leaders. We acknowledge the helpful comments of three referees, one of whom suggested the use of the term collection distance as distinct from striking distance. 15 References

The arguments presented above lead us to the following conclusions regarding the claimed collection distances for ESE air terminals. There is no sound physical foundation for claims of enhanced lightning protective properties for ESE air terminals. Even if a streamer from an ESE air terminal can be launched at an earlier time than a streamer from a simple air terminal, it will be subject to the same laws governing streamer to leader transition and upward propagation as any other leader, and will not complete its path to the downward leader tip unless the electric field is sufficiently high. Thus it is not physically reasonable to assign a collection distance to a special purpose terminal that is larger than the collection distance of a simple terminal in the same position. The choice of 106 m s1 (100cm/s) as an assumed constant upward positive leader propagation speed is not in agreement with the available HV laboratory observations. It would be more reasonable to assume an initial leader propagation speed of about 2cm/s, increasing as the gap between the leader tip and the downward leader diminished, and reaching 100cm/s in the later stages of the attachment process. There appears to have been a misconception regarding the application of the protective radius concept to building lightning protection. Even if it is assumed that the claimed collection distances for ESE air terminals are correct, a simple analysis shows that the edge of a building will not be protected by an ESE air terminal when the edge is at a distance equal to the protective radius from the ESE air terminal as stated in Draft Standard NFPA 781-F93TCR. There is so much doubt about the validity of the claims for ESE air terminals that we consider that the claims should be subjected to independent scrutiny and independently conducted field tests before they are accepted and embodied in any standard.
IEE Proc.-Sci. Meas. Technol., Vol. 144, No. 1, January 1997

1 Australian Standard AS 1768-1991, Lightning protection. Standards Association of Australia, 1991 2 BERGER, G.: Testing to show a time advantage in production of a lightning up leader laboratory simulation of the connecting discharge from a lightning conductor. Lightning protection workshop, November 1992, Hobart 3 BERGER, G.: The application of upward leader initiation time advantage into an electrogeometrical model. Lightning protection workshop, November 1992, Hobart 4 BERGER, G.: Determination of the inception electric field of the lightning upward leader. Proceedings of 8th international symposium on High voltage engineering, 1993, Yokohama, Japan, paper 70.02 5 BERGER, K.: Methoden und Resultate der Blitzforschung auf dem Monte San Salvator bei Lugano in den Jahren 19631971, Bull. Schweiz. Elektrotech. Ver., 1972, 63, pp. 14031422 6 British Standards Institution, British Standard code of practice for protection of structures against lightning. BS6651, 1992 7 BURROWS, B.J.C.: Review of alternative systems. E.R.A. Lightning protection seminar proceedings, 1988, Leatherhead, Surrey, pp. 3.3.13.3.7 8 DARVENIZA, M.: Analysis of a non-standard co-axial downconductor for lightning protection. Proceedings of symposium on Non-conventional lightning protection, 1986, Sydney, paper 4 9 DARVENIZA, M.: Integrated lightning and overvoltage protection. Proceedings of 6th international symposium on High voltage engineering, 1989, New Orleans, LA, paper 10.01 10 DARVENIZA, M., and MACKERRAS, D.: Integrated lightning protection for large modern buildings. Proceedings of Inst. Eng. Aust. Electric energy conference 1989, October 1989, Sydney, pp. 131135 11 DELLERA, L., and GARBAGNATI, E.: Lightning strike simulation by means of the leader progression model. Part I: Description of the model and evaluation of exposure of free-standing structures, Part II: Exposure and shielding failure evaluation of overhead lines with assessment of application graphs, IEEE Trans., 1990, PWRS5, (4), pp. 20092029 12 ERIKSSON, A.J.: A discussion on lightning and tall structures. CSIR special report ELEK 152, National Electrical Engineering Research Institute, Pretoria, July 1978 13 ERIKSSON, A.J.: The lightning ground flash an engineering study. CSIR special report ELEK 189, Pretoria, South Africa, 1979 14 GOLDE, R.H.: Lightning protection (Arnold, London, 1973) 15 GOLDE, R.H.: Lightning; Vol. 1: Physics of lightning, Vol. 2: Lightning protection (Academic Press, London, 1977) 16 GORIN, B.N., and SHKILEV, A.V.: Discharge development in long gaps in the presence of impulse voltage of positive polarity, Elektrichestvo, 1974, (2), pp. 2939
9

17 GUMLEY, J.R.: Non-conventional methods of lightning protection. Proceedings of symposium on Non-conventional lightning protection, 1986, Sydney, paper 3 18 GUMLEY, J.R.: Lightning interception techniques. Proceedings of 20th international conference on Lightning protection, 1990, Interlaken, Switzerland, paper 2.8 19 GUMLEY, J.R.: Comparative performance of lightning air terminals under natural storm conditions. Lightning protection workshop, 1992, Hobart 20 GUMLEY, J.R.: The lightning stroke and fundamental protection concepts. Lightning protection and earthing seminar, first annual technical meeting, 1993, Centre for Management Technology, Kuala Lumpur 21 HARTONO, Z.A., and ROBIAH, I.: A method of identifying the lightning strike location on a structure. Proceedings of the international conference on Electromagnetic compatibility, 1995, Kuala Lumpur, paper 4.5, pp. 112117 22 HAYDON, S.C.: The physics of lightning. Proceedings of symposium on Non-conventional lightning protection, 1986, Sydney, paper 1 23 HEARY, K.P., CHABERSKI, A.Z., GUMLEY, S., GUMLEY, J.R., RICHENS, F., and MORAN, J.H.: An experimental study of ionising air terminal performance. IEEE/PES 1988 summer meeting, 1988, Portland, Oregon 24 IEEE: Working Group report Estimating lightning performance of transmission lines II Updates to analytical models. 92 SM 453-PWRD, 1992 25 INTERNATIONAL ELECTROTECHNICAL COMMISSION: IEC 1024.1-1990 Protection of structures against lightning, Part1: General principles, IEC 1024.1.1-1993, Section 1: Selection of protection levels for lightning protection systems 26 KARMZYN, H., and YEO, T.: Aspects of structural lightning protection. Lightning protection and earthing seminar, first annual technical meeting, 1993, Centre for Management Technology, Kuala Lumpur 27 LEE, R.H.: Protection zone for buildings against lightning strokes using transmission line practice, IEEE Trans., 1978, IA 14, pp. 465470 28 LES RENARDIRES GROUP, : Research on long air gap discharges at les Renardires 1973 results, Electra, 1974, (35), pp. 49156

29 LES RENARDIRES GROUP, : Positive discharges in long air gap discharges at les Renardires 1975 results and conclusions, Electra, 1977, (53), pp. 31153 30 LES RENARDIRES GROUP, : Negative discharges in long air gaps at les Renardires 1978 results, Electra, 1981, (74), pp. 67216 31 LOVE, E.R.: Improvements on lightning stroke modelling and applications to the design of EHV and UHV transmission lines. MS thesis, University of Colorado, Boulder, CO, 1973 32 MACKERRAS, D., DARVENIZA, M., and LIEW, A.C.: Standard and non-standard lightning protection methods, J. Electr. Electron. Eng. Aust., 1987, 7, pp. 133140 33 NATIONAL FIRE PROTECTION ASSOCIATION, USA: Lightning protection code, NFPA 780, 1978 34 NORME FRANCAISE: Protection contre la foudre, installations de paratonerres, NF C17-100, 1987 35 RIVA, D.: Australian approach to protection of buildings and computer rooms. Lightning protection workshop, November 1992, Hobart 36 RIZK, F.A.M.: Modeling of lightning incidence to tall structures, Part I: Theory; Part 11: Application, IEEE Trans., 1994, PWRS9, (1), pp. 162193 37 SINGAPORE STANDARD CP33: 1985: Code of practice for lightning protection, Singapore Institute of Standards and Industry Research, 1985 38 SOULA, S.: Transfer of electrical space charge from corona between ground and thundercloud: measurements and modeling, J. Geophys. Res., 1994, 99, pp. 10,75910,765 39 TRUUPOLD, E.: Analysis of a co-axial lightning conductor. Proceedings of symposium on Non-conventional lightning protection, October 1986, Sydney, paper 6 40 UMAN, M.A.: The lightning discharge (Academic Press, Orlando, FL, 1987) 41 WU, P.-S., TANG, H.-S., JIAN, X.-J., WANG, S.-S., and YAN, Y.-J.: Testing research on effectiveness of radioactive lightning conductors. Proceedings of 6th international symposium on High voltage engineering, 1989, New Orleans, LA, paper 27.19

10

IEE Proc.-Sci. Meas. Technol., Vol. 144, No. 1, January 1997

You might also like