Download as pdf or txt
Download as pdf or txt
You are on page 1of 38

Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

The use of ‘Field Intensification Factors’


in calculations for lightning protection of structures

F. D’Alessandro

ERICO Lightning Technologies, GPO Box 536, Hobart, Tasmania. 7001, Australia.

Abstract

The spatial distribution of the electric field ni the vicinity of a prospective strike point on a
structure is an important factor in the determination of the likelihood of a lightning strike to that
point. Hence, the assessment of the effectiveness of various positions of lightning rods for the
protection of structures ideally requires three dimensional calculations of Laplace’s equation to
obtain the electric field, not only for the various trial positions of the lightning rods, but also for
various positions of the downward lightning leader. This paper shows that such calculations can
be greatly simplified by the use of “field intensification factors”. These factors are calculated for
likely lightning attachment points such as corners, edges and lightning rods, but using an imposed
electric field which is uniform in the absence of the perturbing structures. It is shown that these
factors can be applied in the general case of any lightning downward leader at any arbitrary
position, with errors of less than 20% compared with exact calculations made without the use of
the approximation. Examples are given of calculated field intensification factors for various shapes
and dimensions of typical structures, including composite structures, and protective lightning rods
placed on structures. Quantitative relations for the spatial values of the field intensification factor
as a function of structure and rod dimensions are obtained, leading to a simple method for the
estimation of conditions for corona and upward leader inception. The results are discussed in
terms of practical aspects of lightning protection, such as the use of such calculations for the
placement of lightning rods in the optimum locations on structures.
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

Keywords: Lightning protection; Structures; Lightning rod; Finite Element Method; Electric field
intensification; Field Intensification Factor; Electric field modelling.

1. Introduction

1.1. Basis

It is widely recognised that the lightning attachment process is a complex phenomenon, with
many variables and unknowns. Hence, by necessity, practical models of the process are relatively
simple.
An important practical aspect is the design of protection against lightning. Lightning rods or
“air terminals” are placed on structures and other grounded objects with the objective of
capturing lightning strikes and protecting the structure and its contents. A lightning rod is
positioned on a structure using a particular “placement method”. Examples of existing methods
include geometrical constructions, such as the “Cone of Protection” or “Protective Angle” method
[1], “Faraday Cage” concepts, in which a “meshwork” of conductors or air terminations is placed
at set intervals over a structure [2], and “Electrogeometric models” (EGMs), in which empirical
relationships for striking distance and lightning peak current are invoked [1]. The most common
example of the latter is the “Rolling Sphere Method” [3,4], which is also partly a geometric
construction.
In recent times, more physical models have been proposed, and these have the potential to
improve existing placement methods [5-11]. These models account for air breakdown
mechanisms which are applied to the lightning scale. Most of these models have been derived
from laboratory investigations of long sparks, but some have also used the results of field studies
of natural lightning.
A fundamental requirement in new placement methods [12] is detailed information on the
distribution of the electric field around structures, not only for various positions of trial
placements of protective lightning rods, but also for various positions of a downward leader from
a thundercloud. Such information is needed, firstly, to assess the onset of electrical breakdown
when the electric field exceeds the breakdown field at any given point, but also to assess the
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

spatial extent of the breakdown field, which determines whether an upward-directed leader will
form from the point to meet the downward leader, rather than just produce local corona near the
point. Such requirements would involve multiple solutions of Laplace’s equation in three
dimensions to obtain the electric field, and is impractical for any routine, commercial design for a
lightning protection system.

1.2. Criteria for inception of corona and upward leaders

The most important criterion that must be satisfied by any placement method is that one of the
lightning rods is the preferred strike point, rather than any point on the structure. In the physical
models, this criterion can at least be partly satisfied by calculating the onset point of two important
processes: (i) air ionisation and “streamer corona”, and (ii) upward leader inception and
propagation.
With respect to corona initiation, Peek’s Law is well-known and commonly used for this
purpose. It is applicable to wires. The local electric field at the surface of a wire of radius a
which will result in corona onset is given by

  2 ln(1 / γ )  
0.5

EC = Eo δ  1 +    (1)
  A E oδ a  

where Eo is the value of the electric field at which ionisation balances attachment (3 x 106 V/m), δ
= N/No is the relative number density of the gas (air), and ln(1/γ)/A = 1350 [13]. The value of δ
accounts for variations with temperature T and pressure P, using the gas law P = NkT, where N
is the gas number density, k is Boltzmann’s constant, and No is the number density at normal
temperature and pressure.
One limitation of Peek’s Law is that it only applies to wires. However, Hartmann [14] has
proposed a mathematical-physical model which generalises Peek’s Law for use with points and
rods over a wide range of tip radii. It also accounts for variations in the atmospheric parameters.
According to Hartmann [14], the local electric field, EC, at the tip of a rod for which a stable
corona discharge is self-sustained is given by
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

 A 
EC ( r, δ , H a ) = Eo (δ , H a ) 1 + Boo f ( r,δ , H a ) (2)
 r 

where Eo is the value of the electric field at which ionisation balances attachment (V/m), r is the
radius of curvature of the rod tip (m), δ is the relative air density with respect to standard
conditions (P = 760 Torr, T = 20 °C) and Ha is the absolute humidity of the air (kg/m3). The
terms Ao, Bo, E o(δ ,Ha) and f(r,δ , Ha) are known constants or functions. Under standard
conditions, Equation (2) becomes:

 0.127 
EC ≈ 2.6 × 10 6  1 + 0. 435  (3)
 r 

The corona onset field EC at the rod tip is reached when the external or ambient field, E A,
reaches a value EA = E C / K i. Here, K i is a dimensionless quantity, defined as the degree of
enhancement of the ambient field, i.e., Ki = ES /E A, where E S is the enhanced electric field due to
the grounded object, such as a rod, rod on a structure, structure, etc. Hereafter, Ki will be called
the “field intensification factor”.
With respect to upward leader initiation, the issue is more complicated. After the initiation of
corona streamers and the closer approach of the lightning downward leader, one of these
streamers may develop into a stable, propagating, upward leader that intercepts the downward
leader. The inception criteria for the conversion of a streamer into an upwardly-propagating
leader are rather involved and are outside the scope of this paper. The main details can be found
in [9,15-21]. However, two criteria will be reviewed here briefly – the “critical radius concept”
(CR) [15] and the “critical range of field intensification” (CRFI) [9]. These two criteria can be
applied relatively easily to most practical lightning protection problems because static electric field
computations can be used.
In the CR concept, the initiation of a stable leader requires the attainment of the ionisation field
of ~ 3 MV/m over a “critical corona radius”. Corona inception is not sufficient – the radius of the
corona sheath around the rod tip must increase to a critical size before an upward leader is
initiated. From a series of laboratory measurements, Carrara & Thione [15] found this critical
radius to be ~ 38 cm for large air gaps.
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

In the CRFI criterion, streamers initiated from a rod or other grounded structure must reach a
critical length, Ls, before an upward leader can be initiated. Furthermore, the enhanced electric
field ES must exceed a critical value over the streamer zone length L s. For a minimum charge per
unit length of 20 µC/m, the critical streamer length is ~ 0.7 metres. This minimum condition has
been verified experimentally [22,23], as have the values E s ≈ 500 kV/m and 1 MV/m for positive
and negative upward leaders respectively [16,21,24].

1.3. Electric field intensification

It follows from the preceding discussion that the electric field is one of the most fundamental
parameters in lightning protection studies. In particular, the dominant variables in the strike
process are the temporal evolution and spatial distribution of the electric field between the
thundercloud or downward lightning leader and the prospective strike point. Simultaneous
calculations of these variables and the other processes described above are very complex and
hence not easy to put into practice.
As previous work has shown, e.g., [5,9], to a good first order approximation these
calculations can be done more or less independently. The temporal evolution can be determined
by using analytical models of the electric field due to a given charge distribution on the downward
leader [5,9,10], resulting in EA(t). To date, very little research effort has been directed toward the
computation of the spatial distribution E S(x,y,z) and, hence, Ki(x,y,z), for practical structures.
One of the reasons for this is the fact that the problem is difficult or, in most cases, impossible to
treat analytically. Furthermore, even though field intensification factors have been used in the past
for slender structures, there has been no comprehensive assessment of the general validity of such
an approach. Following is a short review of the main efforts found in the literature.
Eriksson [25] used the work of Anderson [26] to compute the approximate field
intensification factor at the top of a research mast. The mast was approximated by a cylindrical
structure of height H and radius R. Anderson obtained an analytical solution for Ki in this simple
case by assuming that the charge distribution in the structure is partly exponential and partly linear.
Based on this assumption, he was able to calculate Ki as a function of the “slenderness ratio”
H/R. He obtained a quasi-linear variation of K i with H/R on a log-log plot, for slenderness ratios
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

ranging from 20 to 900. In fact, the logarithmic slope n, where K i ∝ (L/R) n, varies from ~ 0.6 at
lower ratios to ~ 0.3 at the higher ratios.
The problem with this analytical approximation is that it only applies to a specific geometry. It
could be argued that corona emission would result in a space charge envelope that modifies the
effective geometry to one tending toward a cylindrical or ellipsoidal shape. However, for most
structures other than very slender ones, this argument is not valid. In any case, until the geometry
is actually modelled and the field intensification factor is computed, it is not known whether
corona inception has actually occurred under typical thunderstorm conditions. Even with the
existence of space charge, the corona envelo pe is drastically modified or even removed
completely in the presence of wind. Finally, it is not possible to apply the approximation to
practical buildings where, in many cases, the slenderness ratio is < 20.
Moore [27] has obtained field intensification factors for cylindrical, hemispherically-capped
lightning rods by approximating them as semi-prolate ellipsoids with vertical, semi-major axes of
length c, horizontal, semi-minor axes of length b and tip radii of curvature a. Petrov & Waters [9]
used a similar approximation for slender structures in their striking distance analysis. The semi-
ellipsoid approximation was used because it is not possible to treat vertical cylinders of a finite
length analytically. The result of Moore’s analysis is a reasonably linear log-log plot of Ki vs c/a,
for values of c/a ranging from 1 to 10,000. The logarithmic slope of the lower and upper parts of
this range is ~ 0.7 and ~ 0.8 respectively. Therefore, this curve is more steep than that obtained
in [25].
Neither of the approximations described above are well-suited to problems involving
structures, such as buildings, where the slenderness ratio is not high, nor can they be used to
compute Ki for lightning rods placed on those structures. The semi-ellipsoid approximation has
been used successfully in lightning strike models that deal with slender structures [8,9]. However,
any assumption that the “width” can be neglected fails when practical, extended structures are
considered.
Izraeli & Braunstein [28] proposed a ne w computation model for lightning attachment to
structures, with an emphasis on the determination of striking points. They used an electromagnetic
model for the lightning discharge and evaluated the critical field strength at various points on the
structure using a Conformal mapping technique. This technique can only be used to solve
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

problems in two dimensions. The authors presented one example, deferring further application
notes and examples to a future paper.
To the author’s knowledge, there is little other published research on this topic, presumably
because of past difficulties involved with any analytical treatment of the problem. However, with
the advent of powerful desktop computing systems and software at a reasonable cost, it is now
possible to solve such problems numerically in two and three dimensions (2D and 3D) using
computer modelling techniques.

1.4. Paper overview

The remainder of the paper describes a simple method for carrying out the calculations of the
spatial distribution of electric field near practical structures and protective lightning rods. In
particular, an approximation involving the use of field intensification factors is outlined. The use of
the approximation avoids the need to make three dimensional calculations. Section 2 describes
the computational method and modelling techniques. Section 3 outlines the approximation method
involving field intensification factors, and also presents calculations which demonstrate the
accuracy of the method. Section 4 presents the results of a series of numerical computations of
field intensification factors for key points on and above rectangular, cylindrical and gable
structures, and lightning rods, across a practical range of physical dimensions encountered in
lightning protection. Section 5 discusses the results in terms of practical issues in lightning
protection and Section 6 presents the conclusions.

2. Modelling techniques

2.1. Overview

The use of static modelling or “snap shots” in time of the temporal and spatial variations of the
electric field (as opposed to iterative methods) to determine leader inception and lightning striking
distances is not new, e.g., [5,9]. The approach has been used successfully by these and other
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

authors to simplify the calculations. The alternative is extremely calculation-intensive, making it


difficult to achieve widespread, practical implementation.
The use of a field approximation is a further step in reducing computational intensity. Instead of
attempting to determine field intensification factors using full downward leader models, with a
given charge distribution over a length of, typically, 3 km, these factors are determined in the
presence of a uniform ambient field of nominal value. Qualitatively, equipotential plots of the
electric field due to and beneath a downward-progressing lightning leader show that the field is
quasi-uniform over typical “striking distances” of 50-100 m. Quantitatively, numerical calculations
in Section 3 will show that Ki can be computed independently of the specific downward leader
field, i.e., it is purely a function of the geometry of the structure. The calculations show that the
results are in agreement with a “full” solution within an uncertainty of < 20%. Hence, full
downward leader modelling, requiring very large problem regions and long computation times, is
not essential for the estimation of field intensification factors.
Two of the most common numerical techniques for electric field computation are the “charge
simulation” and “finite element” methods [29,30]. Modern desktop computers, along with
modelling software that utilises these methods, have made it relatively easy to compute the electric
field distribution over and around a grounded structure and its microgeometry. This can be done
in either 2D XY plane (Cartesian geometry), 2D RZ plane (axisymmetric geometry), or full 3D,
depending on the particular scenario that is to be modelled. In the present paper, the finite
element method (FEM) was used. The details of the modelling will now be described.

2.2. Specific modelling issues

The FEM software uses Delaunay Triangles for 2D problems and tangential vector elements
(tetrahedra) for 3D problems. In order to obtain the best possible results, “adaptive mesh
generation” [31] was used. Good first approximations to the problems described in this paper
were obtained with ≤ 15,000 triangles in 2D and a minimum of 50,000 tetrahedra in reduced 3D
models. The transition from 2D to 3D simulations presents many additional problems. The main
one is that much greater computing resources and time are required. As a general rule of thumb,
at least N 3 tetrahedra, where N3 ∝ (N2)1.5, are required to solve a problem in 3D with the same
accuracy as the equivalent 2D scenario requiring N2 triangles.
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

For 3D simulations in which a field approximation was not used, a two-step procedure was
needed to significantly reduce the computing time required to converge to a solution. This
involved “scaling down” the problem. General trends in the electric field and potential around the
structures were established with an initial solution of the whole problem and then an equipotential
cut-plane was identified whereby the scale size of the problem could be reduced. This technique
is essential in lightning models, which include cloud and downward leader scales sizes of
kilometres, compared with lightning rod tips and building edges and corners with a reduction in
scale size of more than three orders of magnitude. The software used in the present work can
easily deal with a “dynamic range” of 1000:1. In some cases, this ratio can be extended by
another order of magnitude with careful mesh refinement.
One of the dilemmas encountered when modelling structures is how to treat 90 ° angles such
as those that exist at edges and corners. Mathematically speaking, this situation results in a
singularity, because the electric field at such a point tends to an infinite value. Also, larger pointed
angles result in weak singularities. These larger angles occur when curved surfaces, necessarily
“digitised” in numerical analyses, are approximated by a series of linear segments. If a large
number of segments is used so that the curve is “smooth”, meshing problems are encountered
because the mesh dimensions are disproportionate. Hence, a compromise must be reached.
Numerical experiments performed as part of the present work have shown that a circle, for
example, can have as little as 12 segments and solution convergence can still be achieved. This is
because the effect of the weak singularity contracts to the immediate vicinity of the discontinuity
once a critical mesh density is exceeded.
The aim of modelling “sharp” features is to obtain the field intensification factor which is used
to determine whether the inception criteria are met for the launch of a stable upward leader. The
authors of the “leader progression model” [6] have used the critical radius concept [15] to
simplify the determination. This approach has meant, essentially, that all geometrical features on a
structure smaller than the critical radius are represented by a spherical surface with a radius equal
to the critical value.
The main aim of the present paper is to present a basic methodology for the determination of
field intensification factors. As such, the computation of leader inception is outside the scope of
the present work and, since there are several different leader inception criteria, a generic,
criterion-independent approach is taken in the paper.
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

Since it is not possible nor realistic to model the edges and corners of buildings and other
structures as perfect 90° angles, any such features in the models were rounded off to a radius of
curvature of a few centimetres. Such radii were considered to be representative of many practical
situations, e.g., concrete parapet edges, ha nd rails around the top perimeter of buildings. In this
way, solution convergence is achieved with a realistic radius of curvature without necessarily
having to invoke a “broad brush” approach such as the critical radius or any other specific
criterion.
Finally, it is assumed there is no space charge present, i.e., Laplace’s equation is solved
rather than the Poisson equation. Solution of the latter is not a problem if the space charge
function ρ (x,y,z,t,…) is well-defined or assumed to take a particular form. In terms of lightning
protection problems, this function is very complex. The solution of a complete model involving
space charge has not yet been achieved. It will require several years of research, although
Aleksandrov et al [32] recently made a promising start from a theoretical viewpoint. However,
under real thunderstorm conditions, there is often a significant amount of wind. The wind blows
the slowly-moving space charge ions away from the immediate vicinity of the structure or tip of
the lightning rod. Thus, the effect that the space charge may have had on the original field
distribution is significantly reduced.

2.3. Application

The electric field distribution around a structure is due to the presence of the structure in an
ambient electric field, E A. The ambient field is due to the thundercloud charge and, more
importantly, to the dynamic field of the approaching downward leader. The normalised value for
the field, or the field intensification factor K i = EP/EA, can be computed at any point P on the
structure, including at or near any lightning rods installed on the structure.
Models were created so that the problem region was a least five times the size of the
structure to be modelled. Dirichlet boundary conditions were applied to the grounded structure,
lightning rod and the lower plane boundary (the ground) in the model. These boundaries were all
assigned a zero potential, whilst the upper plane boundary of the problem region was assigned a
potential such that, in the absence of the structure, a uniform ambient field of magnitude 10 kV/m
was created throughout the model region. This magnitude of ambient field was chosen because it
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

is typical of real thunderstorm fields. However, the magnitude is not important when calculation
values of Ki as it is a “normalised” quantity.
With the structure present, Laplace’s equation, ∇2V = 0, was solved using the finite element
method. This solution provided contours of potential over the problem region. The magnitude of
the voltage gradient, E P = − ∇V, was computed from the contour values. Finally, the value of Ki
at the required locations (x,y,z) was determined using Ki = E P / EA. In lightning protection studies,
the vertical distribution of the field is of most interest and so, in general, only K i(z) was needed in
each particular case.
Using the above methodology, a series of parametric studies was carried out, i.e., hold all
parameters constant except for one which is varied across a range of values of practical interest,
and then repeat for other variables. In this way, values of K i were calculated for different
structure and rod-on-structure configurations.
The main variables of concern to this study were as follows: structure height, H, and width,
W; lightning rod height, h, and tip radius, r; and the spatial distribution in a vertical direction, z.
The point of interest (POI) on the structure, e.g., near a corner or edge, or in the middle of a
plane surface, is also important, but rather than treating these as variables, the three positions
were treated as separate case studies. In the work on shielding effects of adjacent structures, the
separation, d, of the structures is an additional parameter. Multiple, non-linear regression fits were
carried out on the Ki data to obtain general relations that are useful for calculation purposes.
Given the above information, it can be seen that K i = f(H,W,d,h,r,z), as well as the POI .
The choice of whether to model in 2D or 3D was determined by the nature of the problem.
For example, for structures, the upper flat surfaces and, to some extent, the edges can be
modelled as a 2D XY-plane problem, whilst a 3D model is required in order to obtain
information on corners. For isolated rods, i.e., placed on the ground, or those centrally -placed on
a cylindrical structure, the problem can be solved using a 2D RZ model (axisymmetric). For the
general case of lightning rods on structures, it is necessary to solve 3D models.
Some checks were performed prior to the commencement of any parametric modelling.
Well-known cases which have analytical solutions were modelled and the results compared to
ensure they were in agreement. One example is that of a hemisphere of any given radius, placed
on a large, plane surface, with both at zero potential and subjected to a uniform electric field. It is
well-known that the field intensification factor at the uppermost point of the hemisphere is 3.0.
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

The numerical result obtained for this model was in agreement within an uncertainty of 10%,
which is sufficient accuracy for lightning protection applications.

3. Analysis and validity of the approximation method

The fundamental assumption is that K i is a function of shape (radius of curvature) and height
only. Therefore, its value can be determined numerically with any ambient field. Obviously, a
uniform ambient field is the most simple from the point of view of computations. Other workers
have also employed uniform fields, e.g., [9,25,27].
The following case studies confirm that the use of a uniform field is a good approximation to
full simulations of the field enhancement created when the downward leader provides the ambient
field.

Case 1: 3D structure with lightning rod and parapet, full downward leader model
A solution was obtained for a 3D model comprising a: (i) rectangular structure of height,
width and breadth 20, 50 and 50 m respectively; (ii) parapet 1.5 m high and 0.3 m wide along
one edge of the structure; (iii) lightning rod of height 4 m and spherical tip radius 0.35 m, centred
on top of the structure; and (iv) downward leader approaching the structure, frozen at a point in
time when the vertical distance between the structure and the downward leader is 100 m.
A relatively large tip radius was chosen for the lightning rod so that the enhanced field would
be much lower than the corona threshold in an ambient field of 10 kV/m. The downward leader
was assigned a total charge of 1 C, distributed linearly along its length of 3 km. The electric field
obtained from the downward leader model was checked against an analytical solution [25]. Close
agreement was found between the two methods.
Solutions were obtained for three scenarios, namely, with the downward leader: (I) above the
(centre) axis of the structure, (II) above the parapet, and (III) a lateral distance of 25 m from the
side of the structure which has the parapet. A low resolution equipotential plot in the y-z plane of
the 3D model of the structure is shown in Fig. 1 for scenario III. Table 1 displays the electric field
values measured at some critical points on the structure for the three downward leader scenarios,
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

namely at the: (i) top of the lightning rod (TLR), (ii) inner and outer edges of the mid-point of the
parapet (IEP & OEP), and (iii) middle top of the parapet (MTP).
The results show that the lightning rod is close to or just over the corona threshold of ~ 3
MV/m and so with a suitable further increase in the ambient field, the rod may launch an upward
leader. On the other hand, no point on the parapet has exceeded the corona threshold. Table 1
also shows that there is only a small variation in the value of the field at the points of interest for
the three downward leader scenarios. This is an indication that the field due to the downward
leader is relatively uniform at the height of the structure. A visual inspection of the “worst-case
scenario” (position III) shown in Fig. 1 also confirms that the field is, at least, quasi-uniform.

Case 2a: 3D structure with lightning rod, Ki in presence and absence of structure
In this model, a spherical charge of magnitude 10 mC and diameter 10 m is located at a
distance of 45 m above the corner of a rectangular structure of height, width and breadth 20, 50
and 50 m respectively. A lightning rod of height 4 m and tip radius 0.175 m is placed centrally on
the structure. In the absence of the structure, the overhead charge gives rise to an ambient field of
90-120 kV/m in the vicinity of the structure top. This field was checked against an analytical
model of a downward leader [9] in which the charge is distributed uniformly along its length of 5
km. In this case, the equivalent scenario is a downward leader with a charge per unit length of
0.32 mC/m. Figure 2 shows an equipotential plot in the (diagonal) cut-plane common to the
lightning rod, overhead charge and corner of the structure. Table 2 summarises the electric fields
and Ki values computed from the model.

Case 2b: 2D RZ (axisymmetric) structure with lightning rod


In this model, a spherical charge of magnitude 60 mC and diameter 10 m is centrally located
at a distance of 45 m above a cylindrical structure of height 20 m and radius 25 m. A lightning rod
of height 4 m and tip radius 0.175 m is placed centrally on the structure. In the absence of the
structure, the overhead charge gives rise to an ambient field of 280-380 kV/m in the vicinity of
the structure top. This field was checked against an analytical model of a downward leader [25]
in which the charge is distributed linearly along its length of 5 km. In this case, the equivalent
scenario is a downward leader with a total charge of ~ 3 C. Figure 3 shows an equipotential plot
for the model and Table 3 summarises the electric field and Ki computations from the model.
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

In both of the examples in Case 2, the important comparison is between the “Ki (full)” and
“Ki (FAM)” columns. These compare the Ki values obtained using an electric field that is
representative of the distorted field arising from a descending charge and the field approximation
method. The present experience with FEM modelling of electric fields shows that the uncertainties
in the values are typically 10-15% for a first pass of the solution, i.e., < 50000 tetrahedra for
Case 2a and < 15000 triangles for Case 2b. This is particularly so for 3D models, where
tetrahedra must be meshed rather than simple triangles. The uncertainties can be improved if the
problem region is meshed to a much higher degree, but this requires a significant increase in
computing time. In general, a 10% to 15% accuracy is tolerable in lightning protection studies. As
can be seen from Tables 2 & 3, the Ki values are in agreement, within this level of uncertainty.

4. Results for structures and lightning rods

This section presents the results of the following modelling studies, which were carried out
with the methodology outlined in Sections 2 and 3:
1. Parametric investigations of the dependence of K i on structure dimensions (H,W) and
geometry (rectangular, cylindrical and gable);
2. Parametric investigations of the reduction in K i as a result of shielding effects from taller,
adjacent or adjoined structures; and
3. Parametric investigations of the dependence of Ki on lightning rod height and tip radius, for
rods on a large plane surface such as the ground, and rods placed on structures;

4.1. Modelling of structures

A parametric study was carried out on the field intensification created by a variety of
structures, using a uniform field of 10 kV/m. Three structure geometries were modelled –
rectangular, cylindrical and gable. For each geometry, two parametric series were undertaken: (i)
with a constant width of 10 m and variable height in the range 5 - 400 m, and (ii) with a constant
height of 20 m and variable width in the range 2 - 100 m. The calculations were carried out on
2D-XY, 2D-RZ and 3D models as appropriate.
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

Figure 4 shows a typical 2D XY-plane plot of the equipotential lines after solution
convergence of a model of a gable structure, using the field approximation method. Figure 5 is a
plot of the field intensification factors computed for the surface of a rectangular structure of width
30 m and varying height, for three different POI’s – on the flat (top, central) surface, at the
midpoint of a horizontal edge and at the corner. Hereafter, for brevity, these POI’s will be called
the “middle”, “edge” and “corner” positions respectively.
The spatial distribution of electric field above a structure, particularly in the vertical or z-
direction, is also of practical importance. Figure 6 is a plot of Ki(z), for a rectangular structure of
height 20 m and width 50 m, for three points of interest (x,y). Figure 7 is a plot of Ki(x,z) for two
planes of the same structure, viz. from one corner to another along an edge, and from one edge to
another across the middle.
The modelling described above produced a parametric set of K i data that revealed some
common “power law” trends, e.g., as shown in Fig. 5. Hence, multiple, non-linear regression fits
were performed on these data. The fits produced generalised, monotonic relations for Ki =
f(H,W,z), for the three different POI’s. These are useful for calculation purposes, e.g., semi-
analytical treatment with empirical relations rather than requiring numerical solutions on a case-by-
case basis.
For the field intensification at the surface (z = 0) of structures with a predominantly flat top,
e.g., rectangular and cylindrical, these relations are of the form

Ki(H,W) ~ A Hα W β + 1 (4a)

for the key POI’s, where α > 0 and β < 0.


For gable-roofed structures, the “pitch” is an additional parameter. The pitch, P, is defined as
the rise of the gable roof divided by its half-width. A value of P = 2 was chosen for the purpose
of illustration. In this case, the surface K i was found to follow a general relation of the form

Ki(H,W) ~ A (H+W)α W β + 1 (4b)

where (H+W) is the overall height of the structure, α > 0 and β ≤ 0.


Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

In all cases, the factors A, α and β are “constants” obtained from the regression fits, except that
their values depend on the POI. The standard uncertainties in these factors were < 15%.
The spatial distribution of the field intensification factor for structures with a predominantly flat
top, follows relations of the form

Ki(H,W,z) ~ [A (H/W)α] exp[−B(z/W)β] + 1 (5a)


and
Ki(H,W,z) ~ A Hα Wβ zδ + 1 (5b)

for middle and edge / corner POI’s respectively and z ≥ 0.25 m, where α , β > 0 and δ < 0. For
the corners and edges of gable -roofed structures, the K i relation for spatial distribution is of the
form
Ki(H,W,z) ~ A (H+W)α Wβ zδ + 1 (5c)

where α > 0 and β, δ < 0.

4.2. Modelling of shielding effects

The issue of shielding effects between structures is an important one. It can be broken into
two parts, namely the: (i) minimum separation that is required between two adjacent (but
separate) structures in order to achieve independence of the fields; and (ii) degree to which the
lower structure in a composite (adjoined) structure is shielded by the taller part. These scenarios
are illustrated together in Figure 8.
Field intensification factors were again computed in a parametric study, using rectangular
structures and a uniform field of 10 kV/m. For the purpose of regression fitting, the variables in
this study were H1, W1, H2 and W2 (the structure heights and widths), d min (the minimum
separation for independence of fields, and the POI (middle, edge or corner location on the lower
part of a composite structure for which the field reduction is to be calculated).
The general relation for d min was found to be of the form

d min ~ A Hα Wβ (6)
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

where H, W, α , β > 0 and H, W are the dimensions of the taller structure. Interestingly, the
parametric study showed that dmin is relatively insensitive to the relative heights of the two isolated
structures.
A “reduction factor” can be obtained as a quantitative measure of the degree of shielding of a
taller structure on an adjoined lower structure. The general relation for the reduction factor F R,
where FR ≥ 1.0, was found to be of the form

F R ~ A (H1 − H2)α W2β (7)

for all POI’s, where α > 0 and β < 0. This factor is applied to the unshielded Ki values, i.e., Ki’
= Ki / F R.

4.3. Modelling of lightning rods

The electric field modelling of lightning rods can be broken into two parts, namely: (i) rods on
the ground (a very large plane surface), and (ii) rods placed on structures.

4.3.1. Rods on the ground


A parametric study was carried out on the field intensification created by lightning rods on the
ground, i.e., on a very large surface at zero potential. Once again, a uniform field of 10 kV/m was
used. Two parametric series were undertaken on hemispherically- and spherically-capped rods,
namely with: (i) constant tip radii of curvature r = 0.0125 and 0.35 m, with variable height h in the
range 0.5 – 10 m, and (ii) constant heights of 2 and 4 m, with variable tip radius of curvature in
the range 0.01 – 0.5 m. The simulations were carried out in 2D using RZ plane geometry.
Figure 9 shows a typical equipotential plot for a lightning rod, obtained after solution
convergence. Figure 10 is a log-log plot showing how the field intensification at the tip of a rod
varies with its height above the ground plane for three tip radii of curvature.
The general relation for the field intensification at the tip of the rods (z = 0) was found to be of
the form
K i(h,r) ~ A (h/r)α + 1 (8)
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

where α > 0. The dependence of Ki on the ratio h/r is quasi-linear over a reasonable range of
practical values. The spatial distribution of the field intensification factor for lightning rods on the
ground follows a relation of the form
K i(h,r,z) ~ A h α rβ zδ + 1 (9)

for z ≥ 0.25 m, where α , β > 0 and δ < 0.

4.3.2. Rods on structures


The modelling of lightning rods on structures was a more complex undertaking. Firstly, 3D
models must be solved, as rods can be positioned anywhere on a structure. Secondly, in terms of
non-linear regression fits to the data, the placement of a rod on a structure introduces a
multiplicative component in the overall field intensification factor.
A parametric study was carried out on the field intensification created by lightning rods placed
in the three key POI’s on rectangular, cylindrical and gable structures with dimensions as
specified in Section 3.2. To reduce the total number of models to be solved, a subset of the rod
height and tip radius combinations solved in Section 3.4.1 was used, namely, heights of 1, 2 and
4 m, and tip radii of 0.02, 0.08 and 0.175 m. These combinations were found to be adequate to
highlight the general trends and relationships in the data, and to do regression fits which gave
uncertainties in the parameters of < 20%. In a small number of cases in the parametric study,
extremely tall structures (requiring large problem regions in the model set-up) could not be solved
for rods with small tip radii, since the dynamic range of the software was exceeded. Figure 11
shows a typical equipotential plot obtained after solution convergence of a 3D model of a rod
placed at the corner of a rectangular structure.
The general relation for the field intensification at the tip of a rod on a structure was found to
be of the form
Ki(H,W,h,r) ~ Km.K i(H,W).Ki(h,r) (10)

Here, Ki(H,W) and Ki(h,r) have an identical form to relations (4) & (8) respectively, and Km =
f(H,W) is the “multiplicativity factor”. Regression fitting of the Km data gave a general relation of
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

the form Km ~ A Hα Wβ + 1 for flat-topped structures and Km ~ A (H+W)α W β + 1 for gable


structures, where α > 0 and β < 0.
The general relation for the spatial distribution of field intensification above the tip of a rod
on a structure, i.e., the vertical decrease in field, was found to be of the form

Ki(H,W,h,r,z) ~ Km.Ki(H,W,z).Ki(h,r,z) (11)

where Ki(H,W,z) and Ki(h,r,z) have an identical form to relations (5) & (9) respectively. The
factor Km is of the form Km ~ A Hα Wβ zδ + 1 for flat-topped structures, and K m ~ A (H+W)α Wβ
zδ + 1 for gable structures, where α > 0 and β,δ < 0.
Figure 12 is a log-log plot showing how the field intensification at a vertical distance of 1.0
m from the tip of a rod of radius 0.02 m varies with rod height above a rectangular structure, for
three key POI’s.

5. Discussion

A number of important trends and generalisations emerge from the results of the parametric
analyses. The general and specific aspects are discussed in the paragraphs which follow.
Past analyses involving very slender structures have assumed the field intensification factor is a
function of geometry only, with no dependence on the magnitude or distribution of the pervading
ambient field. Based on the definition of the field intensification factor, this assumption follows
from both physical and intuitive arguments. Upon detailed investigation with full 3D models, it has
been shown with the approximation method that this assumption is valid, with an uncertainty of
less than 20%, for practical, extended structures. This level of accuracy is adequate for
lightning protection problems, since some lightning parameters vary over several orders of
magnitude and other are not known with any accuracy, e.g., only to within a factor of 2-3. The
advantage of using this approximation is that is greatly simplifies the computations and makes the
widespread implementation of the technique much easier.
The parametric investigations highlight the dominant K i dependences. These relate to the
structure geometry, structure dimensions, and rod geometry, together with shielding effects in the
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

case of non-isolated structures. Note that the constants in the regression relations (4)-(11) also
depend solely on geometry and not, for example, on any lightning parameters.
There is a significant difference in Ki between the three key POI’s or locations on a structure.
The field intensification near the corners is highest, followed by the horizontal and vertical edges,
and lastly the top, flat surface. Also, for a given structure height, the K i’s vary, albeit mildly, with
the width of the structure. Any analysis of practical, extended structures must take these
differences into account. This is one reason why the (analytical) semi-ellipsoid approximation is
not suitable for practical structures– it can only assign one field intensification factor, that of a
fixed, rounded top, and the only variable is the structure height. The present (numerical) approach
allows for all of the different points on a structure and the effect of structure width. A rule of
thumb that emerges from the present work is that, for any POI on a structure, the surface Ki
varies roughly as the square root of the height of the structure.
Another trend to emerge from analyses such as the one shown in Fig. 7 is that the field
intensification factor across the top of a structure quickly returns to the value of the “middle top”
position. A rule of thumb for the size of the border region in which a large variation (decrease) is
seen in the field before it settles close to the middle value is “~10% of the structure width”.
The analysis has also shown it is possible to quantify the “shielding effects” between structures
of differing height. For complete independence of the fields due to two “isolated” structures, the
minimum distance between them is a positive power function of the height and width of the taller
structure. For adjoined structures, the taller structure reduces the field around the lower structure
as a positive power of the height difference and a negative power of the width of the lower
structure. The latter result is easily understood in terms of the field at the most distant edges and
corners of the lower structure – the further these are away from the taller structure, the less is the
degree of shielding. For non-adjoined structures, but those in close proximity (inside the minimum
distance), the width relation described above can be replaced by an equivalent distance.
The semi-ellipsoid approximation is useful when applied to cylindrical lightning rods placed
on the ground. As reviewed in Section 1, previous researchers found that the logarithmic slope
or power for the Ki ~ h/r relation was found to be ~ 0.8 over a wide range of ratios. In the
present work, this index was found to be ~ 0.9, i.e., in reasonable agreement, given the different
origins of the index. Hence, for rods placed on the ground, the field intensification factor varies
quasi-linearly with the ratio of rod height to the tip radius.
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

Perhaps the most important result from the present study is the significant difference (in terms
of field intensification) that is seen when a lightning rod is placed on top of a practical structure.
The degree of electric field intensification created by a rod of a given height placed on the
ground is significantly different to (less than) that created by the same rod placed on a
structure, irrespective of the position of the rod on the structure. This is due to the
multiplicative effect of the individual rod and structure electric fields on the overall field
intensification. This point is clearly illustrated in Figs. 13(a,b). Furthermore, the degree of electric
field intensification created by a rod on a structure is significantly different to that created by a
“tall” rod of the same overall height. This is demonstrated in Fig. 13(c).
Another aspect of the electric field modelling results relates to the spatial distribution near or,
more precisely, the decrease of the electric field in a vertical direction from the top of a rod,
structure, or rod on a structure. This is particularly important in calculations involving upward
leader inception. In the present study, for distances z ≥ 0.25 m, the vertical decrease in the field
from a point on a structure was found to vary roughly as 1/√ z. For rods, the decrease is more
pronounced, varying roughly as 1/z. The field decrease for 0 < z < 0.25 m, the decrease is much
more steep than indicated above for both structures and rods.
Finally, from a practical point of view, even though the application of the field approximation
makes case-by-case calculations relatively straightforward, the general regression relations
obtained from the data are useful because they are easily coded into a computer program that
must take many different Ki values in order to compute other quantities.
There are many examples of the usefulness of these types of electric field computations in
lightning protection studies. The general regression relations that can be derived from the
calculations are particularly useful from a practical viewpoint. For example, as mentioned in
Section 1, the method by which lightning rods are placed on structures is an important step in the
design of a lightning protection system. Such a system can be made more “efficient” or cost-
effective with a quantitative consideration of the points of highest electric field intensification and
the influence these points have on the probability of lightning attachment. An example of a fully-
developed, practical methodology for the placement of lightning rods on structures was presented
in [12]. This placement method has also been assessed through two separate studies under
natural lightning conditions [33,34].
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

6. Conclusions

This paper has presented a practical methodology for the computation of the electric field
intensification that is created by practical structures and protective lightning rods as a result of the
perturbation of thunderstorm ambient fields. The method requires the numerical solution of
electric fields, in 2D and 3D, using a uniform ambient field as an approximation of a full solution.
The latter would otherwise involve many trial computations of downward leader positions and
different grounded structure geometries and configurations. The approximation can be used to
greatly simplify the analysis of electric field intensification at any point near or on the surface of a
structure or protective lightning rod. The validity of the approximation using field intensification
factors has been verified by extensive computations on case studies. Importantly, the paper
demonstrates the utility of the method for practical structures and protective rods.
Some of the specific conclusions from the analyses and results include:
• Field intensification factors have been evaluated for a wide variety of geometric parameters of
practical structures and lightning rods. From these data, generalised regression relations of the
form Ki = f(H,W,d,h,r,z,POI) can be obtained for the spatial distribution of electric field
above lightning rods and structures;
• There is a significant difference in field intensification between the three main points of interest
on a structure, i.e., the field seen near corners, along edges and on flat surfaces.
• The surface electric field of a structure subjected to a given ambient field varies roughly as the
square root of its height;
• There is a narrow perimeter region around the top of a structure in which a large spatial
variation (decrease) is seen in electric field intensification before it becomes more uniform and
approximates that of the horizontal, upper surface of the structure. The size of this region is
typically ~ 10% of the width of the structure.
• The shielding effects of adjoined structures and isolated structures in close proximity can be
quantified in terms of a reduction factor and a minimum distance for independence of the
fields of the form FR = f(∆H,W) and dmin = f(H,W) respectively, where ∆ H is the height
differential of composite structures;
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

• The numerically computed results for the electric field around cylindrical lightning rods placed
on the ground are in close agreement with those determined by via an analytical, semi-
ellipsoid rod approximation;
• The degree of electric field intensification created by a rod of a given height placed on the
ground is significantly different to that created by the same rod placed on a structure;
• The degree of electric field intensification created by a rod on a structure is significantly
different to that created by a “tall” rod of the same overall height;
• Over a practical spatial range z, the vertical decrease in the field varies roughly as 1/√ z and
1/z from a point on a structure and from the tip of a lightning rod respectively;

• The methodology presented in this paper provides a simple means for the estimation of
electric field conditions leading to corona and upward leader inception and can be used
directly in methods that determine the placement of lightning rods on structures.
Finally, the computer modelling of electric fields presented in this paper can be applied to a
wide variety of additional scenarios. For example, it is possible to model the effect of different
orographics such as structures located in valleys and on hills, more elaborate geometries such as
peculiar buildings, towers and multi-component structures, and multiple buildings such as a
complete section of a city block. All of these possibilities provides a means for better
quantification of the locally-enhanced electric fields present during a thunderstorm, leading to
improvements in the application of lightning protection systems on structures.

Acknowledgements

The author wishes to thank Dr John Lowke from the CSIRO, for his helpful suggestions
regarding the paper, and the two anonymous reviewers, for their feedback which helped improve
the clarity of the paper.

References

[1] R.H. Golde (Ed.), Lightning. Vol II: Lightning Protection, Academic Press, N.Y, 1977.
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

[2] M. Szczerbinski, A discussion of ‘Faraday cage’ lightning protection and application to real
building structures, J. Electrostat. 48 (2000) 145-154.
[3] R.H. Lee, Protection zone for buildings against lightning strokes using transmission line
protection practice, IEEE Trans. on Indust. Appl. IA-14 (1978) 465-469.
[4] R.H. Lee, Lightning protection of buildings, IEEE Trans. on Indust. Appl. IA-15 (1979)
236-240.
[5] A.J. Eriksson, The incidence of lightning strikes to power lines, IEEE Trans. Pow. Del.
PWRD-2 (1987) 859-870.
[6] L. Dellera, E. Garbagnati, Lightning stroke simulation by means of the leader progression
model. Part I. Description of the model and evaluation of exposure of free-standing
structures”, IEEE Trans. Power Delivery 5 (1990) 2009-2022.
[7] F.A.M. Rizk, Modeling of lightning incidence to tall structures. Part I: Theory, IEEE Trans.
Power Delivery 9 (1994a) 162-171.
[8] F.A.M. Rizk, Modeling of lightning incidence to tall structures. Part II: Application, IEEE
Trans. Power Delivery 9 (1994b) 172-193.
[9] N.I. Petrov, R.T. Waters, Determination of the striking distance of lightning to earthed
structures, Proc. R. Soc. Lond. A 450 (1995) 589-601.
[10] N.I. Petrov, G. Petrova, R.T. Waters, Determination of attractive area and collection
volume of earthed structures”, Proc. 25th ICLP, Rhodes, Greece, 2000, pp. 374-379.
[11] P. Lalande, A. Bondiou-Clergerie, P. Laroche, A. Eybert-Berard, J.P. Berlandis, B.
Bador, A. Bonamy, M.A. Uman, V.A. Rakov, Leader properties determined with triggered
lightning techniques, J. Geophys. Res. 103 (1998) 14109-14115.
[12] F. D'Alessandro, J.R. Gumley, A ‘Collection Volume Method’ for the placement of air
terminals for the protection of structures against lightning, J. Electrostat. 50 (2001) 279-
302.
[13] J.J. Lowke, R. Morrow, Theory of electric corona including the role of plasma chemistry,
Pure & Appl. Chem. 66 (1994) 1287-1294.
[14] G. Hartmann, Theoretical evaluation of Peek’s Law, IEEE Trans. IAS 20 (1984) 1647-
1651.
[15] G. Carrara, L. Thione, Switching surge strength of large air gaps: A physical approach,
IEEE Trans. Pow. App. Sys. 95 (1976) 512-524.
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

[16] Les Renardières Group, Positive discharges in long air gaps at les Renardières - 1975
results and conclusions, Electra 53 (1977) 31-153.
[17] F.A.M. Rizk, A model for switching impulse leader inception and breakdown of long air-
gaps, IEEE Trans. Power Delivery 4 (1989a) 596-606.
[18] F.A.M. Rizk, Switching impulse strength of air insulation: Leader inception criterion, IEEE
Trans. Power Delivery 4 (1989b) 2187-2195.
[19] A. Bondiou, I. Gallimberti, Theoretical modelling of the development of the positive spark in
long gaps, J. Phys. D: Appl. Phys. (1994) 27 1252-1266.
[20] N. Goelian, P. Lalande, A. Bondiou-Clergerie, G.L. Bacchiega, A. Gazzani, I. Gallimberti,
A simplified model for the simulation of positive-spark development in long air gaps, J.
Phys. D: Appl. Phys. 30 (1997) 2441-2452.
[21] E.M. Bazelyan, Yu.P. Raizer, Spark Discharge, CRC Press, New York, 1998.
[22] E.N. Chernov, A.V. Lupeiko, N.I. Petrov, Investigation of spark discharge in long air gaps
using Pockel's device, Proc.7 th Int. Symp. on High Volt. Eng., Dresden, 1991, pp. 141-
144.
[23] N.I. Petrov, V.R. Avanskii, N.V. Bombenkova, Measurement of the electric field in the
streamer zone and in the sheath of the channel of a leader discharge”, Tech. Phys. 39
(1994) 546-551.
[24] Les Renardières Group, Negative discharges in long air gaps at Les Renardières, Electra 74
(1981) 67-216.
[25] A.J. Eriksson, ELEK 189 The lightning ground flash - an engineering study, PhD thesis,
University of Natal, Pretoria, South Africa, 1979.
[26] R.B. Anderson, Measuring Techniques, Chapter 13 in Lightning, edited by R.H. Golde,
Academic Press, London, 1977.
[27] C.B. Moore, Improved configurations of lightning rods and air terminals, J. Franklin. Inst.
315 (1983) 61-85.
[28] I. Izraeli, A. Braunstein, Analysis and design of protection systems for structures against
direct lightning strokes , IEE Proc. 130 Part A (1983) 140-144.
[29] H. Singer, H. Steinbigler, P. Weiss, A charge simulation method for the calculation of high
voltage fields, IEEE Trans. PAS 95 (1974) 1660-1668.
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

[30] M.D.R. Beasley, J.H. Pickles, G. d’Amico, L. Beretta, M. Fanelli, G. Giuseppetti, A. di


Monaco, G. Gallet, J.P. Gregoire, M. Morin, Comparative study of three methods for
computing electric fields, Proc. IEE 126 (1979) 126-134.
[31] L. Janicke, A. Kost, Error estimation and adaptive mesh generation in the 2D and 3D finite
element method, IEEE Trans. Magn. 32 (1996) 1334-1337.
[32] N.L. Aleksandrov, E.M. Bazelyan, R.B. Carpenter, M.M. Drabkin, Yu.P. Raizer, The
effect of coronae on leader initiation and development under thunderstorm conditions and in
long air gaps, J. Phys. D: Appl. Phys. 34 (2001) 3256-3266.
[33] N.I. Petrov, F. D’Alessandro, Assessment of protection system positioning and models
using observations of lightning strikes to structures”, Proc. Roy. Soc. Lond. A 458 (2002)
723-742.
[34] F. D’Alessandro, M. Darveniza, “Field validation of an air terminal placement method”,
Proc. of 6th Int. Sympos. Lightn. Prot., Santos, Brazil, 2001, pp. 234-239.
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

Tables

Table 1
Electric field strength at several critical points on the y-z plane of the 3D model, for the three
downward leader scenarios.

Critical Electric field (MV/m)


point I II III
TSAT 3.4 2.9 2.7
IEP 1.1 1.1 1.2
MTP 0.7 0.7 0.8
OEP 1.4 1.3 1.5

Table 2
Electric fields and Ki’s computed from the 3D model in Case 2a. The results of the field
computations near the lightning rod and corner are quoted for two points, viz. at the tip / surface
and at a distance of 1 m from the tip / surface. K i (full) and Ki (FAM) are the field intensification
factors determined from the “full” solution (case study) and the “field approximation method”
respectively.

Overall e -field (kV/m) Intensif. Factors


Feature Calc. Structure Structure Ki Ki Diff.
point absent present (Full) (FAM) (%)
Lightning Tip 90 3100 34 40 15
rod 1m 90 205 2.3 2.6 5
Corner Surface 120 3640 30 35 14
1m 120 450 3.8 3.6 5

Table 3
Output from the 2D-RZ model in Case 2b. Definitions are the same as in Table 2.

Overall e -field (kV/m) Intensif. factors


Feature Calc. Structure Structure Ki Ki Diff.
point absent present (Full) (FAM) (%)
Lightning Tip 380 12500 33 38 13
rod 1m 380 920 2.4 2.5 4
Edge Surface 280 2500 9 10 10
1m 280 700 2.5 2.9 14
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

Figure captions
Fig. 1. Plot of equipotentials around a rectangular structure and lightning rod under the influence
of an electric field due to an approaching lightning channel of length 3 km and total charge 1 C.
This example shows the y-z plane from the 3D model.

Fig. 2. Plot of equipotentials around a rectangular structure and lightning rod under the influence
of an electric field due to a spherical charge of 0.01 C and 10 m in diameter, located 45 m above
the corner of the structure. In this example, a diagonal cut-plane common to the rod, overhead
charge and corner of the structure is shown from the 3D model.

Fig. 3. Plot of equipotentials around a cylindrical structure and lightning rod under the influence of
an electric field due to a spherical charge of 0.06 C and 10 m in diameter, located 50 m above
the centre axis of the structure. This example utilises a 2D RZ model for the structure, rod and
charge due to the axisymmetric nature of the problem.

Fig. 4. Equipotential plot for a gable structure of overall height 40 m and width 20 m. The roof
has a “pitch” of 2.0.

Fig. 5. Variation of surface field intensification, Ki(z=0), with the height of a rectangular structure.
In this example, the structure width is fixed at 30 m.

Fig. 6. Plot of the vertical decrease in field intensification, K i(z), above a rectangular structure of
height 20 m and width 50 m, for three points of interest, (x,y), on top of the structure.

Fig. 7. Plot of Ki(x,z) for two planes of a rectangular structure (H = 20 m, W = 50 m), viz. from
one corner to another along an edge, and from one edge to another across the middle. In this
example, z is the structure height, i.e., K i is the surface field.

Fig. 8. Equipotential plot for two separate structures, one of which is a composite (adjoined)
structure.

Fig. 9. Equipotential plot for a hemispherically-capped lightning rod of height 6 m and tip radius
12.5 mm. Only a small region near the tip of the rod, measuring ~ 170 x 240 mm, is shown, due
to the density of the equipotential lines (high potential gradient) near the tip.

Fig. 10. Variation of Ki(z=0) with rod height above the ground for tip radii of 1, 12.5 and 100
mm.

Fig. 11. Equipotential plot in the y-z plane of a 3D model. In this case, a lightning rod of height 4
m and tip radius 0.175 m is placed at the corner of a rectangular structure of height 20 m and
width 10 m.

Fig. 12. Variation of K i(z = 1.0 m) with rod height above and position on a rectangular structure.
The rod tip radius in this example is 20 mm, the structure height is 40 m and width is 60 m.
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

Fig. 13. (a) Field intensification, Ki(z=0), as a function of rod height, for a rod placed on the
ground and the same rod placed in the middle and on the corner of a low, wide structure (H = 10
m, W = 50 m); (b) Same as (a), but for a tall, slender structure (H = 100 m, W = 20 m); (c) Field
intensification, Ki(z=0), as a function of overall height, for a “tall” rod on the ground and a 1 m
rod in two positions on a structure of width 35 m. In all examples, the rod tip radius was fixed at
20 mm.
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

Figures
Fig. 1.

Fig. 2.
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

Fig. 3.

Fig. 4.
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

Fig. 5.

1000
Corners
Horizontal edges (mid-point)
Middle top

100
Ki

10

1
1 10 100 1000

Structure height (m)

Fig. 6.

100
Corner
Horizontal edge (mid-point)
Middle top
log(K i)

10

1
0 2 4 6

Distance from structure, z (m)


Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

Fig. 7.

100
Line between two corners along an edge
Line between two edges across the middle
log(Ki)

10

1
0 10 20 30 40 50

Horizontal position on structure (m)

Fig. 8.
Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

Fig. 9.

Fig. 10.

10000

1000
K i at rod tip

100

10
r = 100 mm
r = 12.5 mm
r = 1 mm
1
0.1 1 10 100

Rod height above ground plane (m)


Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

Fig. 11.

Fig. 12.

6
Ki at z = 1.0 m

2
Rod near corner
Rod on horizontal edge (mid-point)
Rod on middle top
0
0 2 4 6 8

Rod height above structure (m)


Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

Fig. 13(a).

1000
Rod on structure (corner)
Rod on structure (middle top)
Rod on ground
K i (z=0)

100

10
0.1 1 10

Rod height (m)

Fig. 13(b).

10000
Rod on structure (corner)
Rod on structure (middle top)
Rod on ground

1000
Ki (z=0)

100

10
0.1 1 10

Rod height (m)


Published in Journal of Electrostatics, 2003, vol. 58, pp. 17-43.

Fig. 13(c).

10000

1000
K i (z=0)

100

10
Tall rod on ground
1 m rod on structure (corner)
1 m rod on structure (middle top)
1
5 10 20 50 100 200

Overall height (m)

You might also like