Download Report

2015 EWEA Offshore, Copenhagen, Denmark
Lightning Protection Zoning and Risk Exposure
Assessment of Wind Turbines
S.F. Madsen, A.C. Garolera and K. Bertelsen
Global Lightning Protection Services A/S
HI Park 445
7400 Herning, Denmark
sfm@glps.dk
Abstract — The present paper discusses attachment processes
on wind turbines and wind turbine blades in particular, to set
up a simple set of guidelines to achieve effective lightning
protection of blades.
The first part discusses how downward and upward initiated
strikes are intercepted by blades to identify the most likely
attachment zones on the blade. An extension of the general
principles in IEC 61400-24 [1] for identifying the number of
strikes to the wind turbine is made, such that a distinction
between upward and downward strikes is suggested for the
exposure risk assessment. Besides discussing which current
pulses are likely to attach the different zones of the blade, the
probabilities of getting such strikes are assessed using the
probability density functions in IEC 62305-1 [2] to complete
the risk assessment. Different literature discussing lightning
attachment distribution is reviewed, which eventually
supports the necessity of focusing the LPS effort sin the tip
region of the blade.
Based on the analysis, the lightning protection Zoning
Concept [3] for wind turbine blades published in 2012 is
presented, and a refined concept is developed based on
findings for blade lengths exceeding 80m.
The final part of the paper comments on natural steps in the
good blade design process, which ensures that the two main
features of LPS design (interception efficiency and current
capability) are considered.
Keywords – Wind turbine blade, Lightning protection,
lightning zoning, exposure risk assessment.
I.
INTRODUCTION
Lightning protection of wind turbine blades can be
divided into two main concerns, lightning interception, and
lightning current conduction. The first process – and the
focus in this paper - defines where the blade is struck, and
ultimately forces the lightning protection engineers to think
of how to ensure timely and correct inception of the upward
or connecting leader.
In 2010 the analysis of lightning attachment to wind
turbines resulted in the development of the zoning concept
for lightning protection of wind turbine blades, refined
slightly in 2012 [3]. Based on numerical modeling and field
surveys, the analysis considered which areas of the wind
turbine blades that were exposed to direct lightning strikes,
and which amplitudes were most likely to be experienced
on the different parts of the blade. The information from the
different means of analysis was structured, and a special
zoning concept dividing the blade into zones with different
exposure was derived.
The concept has been used in numerous wind turbine
blade designs, to foresee where to place air terminations,
and which parameters to use when simulating or testing the
lightning behavior on components in the different zones.
In [3] the sketches and descriptions were covering
blades up to 40m lengths, in which it was found that direct
strikes to inboard sections of the blades were very unlikely
and of limited amplitude. Since the blades are only getting
longer, the zone where direct strikes are not expected
according to the previous zoning concept from 2012 needs
an update to consider direct strikes of limited amplitude.
Besides the distribution of strikes on the blades, this
paper also presents a simple engineering mean of estimating
the number of strikes occurring to a wind turbine of certain
height, accounting for upward and downward strikes, local
ground flash density, estimated turbine lifetime, etc.
Knowledge of this information enables a definition of
lifetime testing of critical components.
II.
ATTACHMENT PROCESS TO WINDTURBINES
Direct lightning attachment to any structure can be
triggered by processes initially formed at the cloud, or by
the development of an upward leader at the grounded
structure. These important processes when determining
which parts of the blade are exposed to the direct attachment
will be treated in this section.
A. Downward initiated strikes
Along with the electrification processes where the
development of the charge regions in the cloud appears, the
field close to the cloud charge centre and at elevated objects
at the ground will increase. For flat terrain with only limited
variations in the structure elevation, the field will typically
reach the ionization field in the cloud before it happen at the
ground (due to the concentrated charge, and the lower field
1
strength required to ionization at higher altitudes = lower
pressure), and the inception of a downward moving stepped
leader will be the consequence.
The downward moving stepped leader will propagate
with a velocity in the range of 105m/s towards the ground,
with a potential more or less similar to the cloud potential.
As the leader progresses, the field around grounded objects
will increase, and once the downward moving stepped
leader gets close enough to exceed the ionization field
strength at ground, an upward moving connecting leader is
formed.
The downward moving stepped leader and the
connecting leader from the ground will then approach one
another, and connect to form the final path for the lightning
current. Once the path is completed, the charge deposited in
the leader branches (as the stepped leader has descended)
will be discharged to ground through the wind turbines, and
can be measured as a high amplitude current pulse in the
structure known as the first return stroke. The attachment
process has been completed, the path between cloud and
earth has been defined, and the following long stroke current
and subsequent stroke pulses will flow in that path.
The distance between the tip of the downward moving
stepped leader and the structure at the moment when the
upward moving leader is incepted is known as the ‘striking
distance’ and depends on the field strength between the two.
The field strength is determined by the potential at the leader
tip, and eventually linked to the charge distribution on the
leader - and particularly at the leader tip. Empirical
relationships have been defined to correlate the striking
distance with the prospective peak current, eventually to
define engineering methodologies to place lightning air
terminations. The principles are explained by Berger (1972)
and Golde (1977) and described in [4]. The relationships are
based on investigations of strikes to power lines, and the
Electro Geometrical Methods (EGM) are then in lightning
standards extended to be used on all ground structures not
exceeding 60m in height [2].
Concerning wind turbines and other tall and slender
structures, it has been found that the field is enhanced by the
structure itself, which plays an important role in the
attachment process. It does not only increase the likelihood
of upward strikes as discussed in section C, but it can also
change the striking distance locally on the structure. The
consequence is that when assessing striking distances for
downward (DW) strikes of certain amplitudes, the striking
distances differ around the structure, making the simple
EGM methods with constant striking distance inadequate
During the past ten years, numerical methods have been
developed to assess the locations where turbines can be
struck by strikes with different current amplitude [5], all
based on the charge distribution and leader inception
criterion proposed by Becerra and Cooray [6] and [7]. The
results have led to the zoning concept for wind turbine
blades describing which parts of the blades that are exposed
to which lightning current amplitudes [3].
B. Upward initiated strikes
For tall and exposed structures, a situation may occur
during the cloud electrification process where the ionization
field is reached at the structure before it is reached in the
cloud. This can cause the formation of an upward initiated
strike where the leader is incepted at the structure and
propagates upward towards the charge centers in the cloud
[8]. Some researchers have suggested that the formation of
an upward strike may be triggered by inter cloud discharges
or distant discharges to earth, such that a rapid redistribution
of charge within the cloud can trigger the formation of the
upward leader. The predominant upward strike are initiated
by a positive upward leader, and lowers negative charge
from the cloud to the structure, corresponding to what is
known as negative lightning.
Besides the structure height determining the likelihood
of upward initiated strikes, the elevation of the surrounding
area is also important. In this sense, if a turbine is located on
a mountain top, the effective height used for assessing the
fraction of upward vs. downward strikes must consider the
field enhancement due to the presence of the mountain. An
example of the 70m tall instrumented tower on the Mount
Salvatore used for the lightning research of Berger, is
estimated to have an effective height of 270m – 350m due
to the elevation of the mountain top in relation to the
surrounding area [8]. The principle of effective height
assumes that there is a definitive gap between the top of the
structure and the cloud base, whereas situation where the
cloud bases are at the same elevation as the structure (winter
lightning in Japan) the term effective height is not defined.
There are several approaches for estimating the effective
height, but no universal principle has been agreed. Recent
research activities reported at the CIGRÉ 4.36 meetings and
published at ICOLSE 2015.as well as in the maintenance
C. Empirical Ratio of Upward and Downward strikes
Comparing different lightning incidences to structures
of various heights, it is found that for structures below 100m
of effective height almost all strikes appear to be downward
initiated, whereas structures with a height of more than
500m tend to trigger only upward strikes.
In 1987, Eriksson published a research on lightning
attachments to structures in different countries with heights
ranging from 20m to 540m, and derived an empirical
relationship between the total number of strikes to a certain
structure (including both upward and downward initiated
strikes) and the ground flash density [8].
𝑁 = 24 ∙ 10−6 ∙ 𝐻𝑠2.05 ∙ 𝑁𝑔 

In the equation Hs [m] is the structure height and Ng
[1/yr·km²] is the annual ground flash density for the site in
question. The relationship is believed valid for objects of
height taller than 60m.
Later Eriksson and Meal (1984) fitted the data with an
equation defining the percentage of upward strikes relative
to all strikes, valid for structure heights ranging from
Hs=78m – 518m where the percentage of upward strikes
according to the equation is 0% and 100% respectively [8].
𝑃𝑢 = 52.8 ∙ ln(𝐻𝑠 ) − 230

Rakov and Uman [8] suggest to estimate the number of
strikes to a certain structure, and differentiating between
DW and UW strikes, by starting with the general equation
for assessing the number of DW strikes to the structure vs.
the equivalent collection area, and then subtracting this
2
number from the overall number of strikes. In this sense, the
three equations are:
NDW = Ae ∙ Ng 
NTot = 24 ∙ 10
−6
∙
Hs2.05
∙ Ng 

originally published in CIGRE reports [11], and currently
being revised by a new CIGRE working group.
The accumulated distributions are seen in Fig. 1,
indicating the often mentioned mean value of the first short
stroke peak current of app. 33kA.
NUW = Ntot − NDW
Where Ae is the equivalent collection area, which for a
slender and tall structure is defined as:
Ae = π ∙ R2a 

β
R a = α ∙ Hs
Where α and β are empirical constants, where values
suggested by CIGRE [9] are 𝛼 = 14 and 𝛽 = 0.6 .
Considering a wind turbine with a 150m tip height located
on flat terrain with an annual ground flash density of 𝑁𝑔 =
2 as an example, the following results apply:
R a = 471m
Figure 1. Accumulated probability density function of the First negative
short, the subsequent negative short, and the first positive short stroke.
Ae = 696,769m 
2
NTot = 1.39yr −1 

NDW = 0.50 yr −1 
NUW = 0.88yr −1
If this wind turbine was located as one out of 50 in a
wind farm, and the lightning exposure during 3 years of
operation was to be estimated, the following overall number
of strikes to the wind turbines would be:
NTot_50 = 208
NDW_50 = 75

NUW_50 = 133
In reality the total number of strikes will most likely be
lower than 50 ∙ 𝑁𝑡𝑜𝑡 , since the collection areas of the
individual turbines often overlap.
As experienced with the Japanese winter lightning in
section IV, the fraction of UW initiated lightning using the
analytical approach above may only be valid for ‘normal
lightning environments’ with cloud heights of 5-10km,
whereas the winter lighting environment with cloud bases
of 1-1.5km will result in a much larger fraction of UW
discharges [10].
III.
PROBABILITY AND RISK
One of the issues which needs to be considered when
designing blade LPS, is the probability of having strikes of
different amplitudes. This means that even if a strike of a
certain type may cause very severe damage, it might be that
the probability of having such a strike is insignificant.
Evaluating the risk may therefore lead to the decision that
the installed LPS doesn’t need to intercept such a strike. To
assess these circumstances, the probability density functions
as presented in the IEC 62305-1 [2] are discussed.
A. Probability Distribution
In IEC 62305-1 [2] the probability distributions are
identified for the different strike parameters, which are
shown in Fig.3. The three proposed distributions are
As seen on the probability distributions, the probability
of the first short stroke (positive or negative) being smaller
than 10kA is only app. 3% for the first positive stroke, and
even less for the first negative stroke. The probability for the
subsequent negative short stroke is of less importance, since
the attachment process is determined by the leader charge
defining the first short stroke current.
The information of the probability density functions of
the strike parameters can be used to assess the risk of
attachment and hence damages to the blades in question.
B. Risk Interpretation
In section 8.2 of IEC 62305-1 [2] the four lightning
protection levels LPL1 to LPL4 are described, and
associated with the probability distributions as plotted
below. Here it is explained how the efficiency of the
different LPL is assumed equal to the probability with
which lightning current parameters are inside such range.
This means that the efficiency of a lightning protection
system made to comply with LPL1 is 98%, where the 2%
outside the range either have an amplitude which is lower
than 3kA (the LPS fails in intercepting the strike - it misses
the air termination) or is exceeding 200kA (the LPS is struck
correctly, but destroyed by the current).
However, it does not state clearly that the LPS must
withstand everything between 3kA and 200kA to be in
compliance with LPL1, only that the efficiency of the LPS
is expected to be equal to the percentage of strikes falling
within the same range for LPL1.
In IEC 61400-24, covering lightning protection of wind
turbines [1], all references concerning LPL1 points towards
IEC 62305-1 [2], hence there is no guidance in how to
interpret the risk or the range defined for each LPL.
Based on the above, there are two typical means of
addressing the expected efficiency of LPL1 and hence how
to claim responsibility at blade damages, although not
explicitly defined in the standard:
1.
The blade protection designed (and tested) according to
LPL1 is claimed having an efficiency of 98%, meaning
3
2.
that the blade must tolerate all strikes with amplitudes
between 3kA and 200kA, whereas LPS failure and
blade damages are tolerated for peak currents outside
this range (Ip<3kA or Ip>200kA).
The blade protection designed (and tested) according to
LPL1 is claimed having an efficiency of 98%, meaning
that for every possible strike occurring to the blade
during the turbine lifetime, the blade must remain
unharmed in 98% of the events disregarding the
amplitudes.
The discussion on how to evaluate the LPL1 protection
concept and how to describe the principle more clearly is
the target for the revision and maintenance of the IEC
61400-24 Ed. 1.0.
IV.
LITTERATURE REVIEW
Many researchers and manufacturers around the world
work in fields related to lightning protection of wind turbine
blades. In the following sections, a couple of interesting
papers are addressed, and the important aspects related to
attachment processes to wind turbines are highlighted.
A. Winter lightning characteristics, M. Ishii et al.
The paper presented at the ICOLSE conference in 2013
[11], describes a measurement campaign on 27 Japanese
wind turbines conducted in the past 5 years. The
measurements were set up as part of the CIGRE working
group WG C4.4: Lightning Protection of Wind Turbines,
and the measurements were conducted from 2008 to 2013
on 21 wind turbines on the west coast of Japan (winter
lightning area), and 6 turbines in land and on the south/east
coast of Japan. The following details are outlined in the
paper:
1.
2.
3.
4.
5.
A total of 676 strikes were measured using Rogowski
coils, preferably to characterize the waveforms of the
lightning strikes occurring to wind turbines in Japan.
The 676 strikes were captured during the five lightning
seasons from October to April (2008-2013), and further
10 strikes were captured outside this timeframe – but
left out of the analysis.
Of the 676 strikes, 69% were of negative polarity, 17%
were of positive polarity and the remaining 14% were
bi-polar flashes.
674 of these 676 strikes were upward initiated
corresponding to more than 99.7%. Only a single
downward positive and a single downward negative
strike were identified. All the negative upward flashes
and 95% of the bipolar flashes started with a positive
upward moving leader. At a presentation at
ICOLSE’2013 it was stated that 252 strikes were
recorded on video cameras, and of these 252 strikes, all
but one strike attached to the extreme tip (99.6%),
whereas the remaining single strike attached 1m
inboard the tip. These last data is not included in the
ICOLSE paper.
Concerning the peak current, only 5 strikes (2 negative
and 3 positive) out of 676 strikes (<0.8%) exhibited
peak amplitudes in excess of 50kA, corresponding to
less than 1% of the strikes.
Concerning the specific energy, only 2 strikes out of the
676 strikes exhibited specific energies in excess of
10MJ/Ohm, corresponding to less than 0.3%. The
6.
7.
8.
maximum specific energy measured was 19MJ/Ohm
with a peak current of +39kA
Concerning the charge, 27 strikes (<4%) contained
charge levels exceeding 300C, 2 strikes (<0.3%)
contained charge levels exceeding 1000C. Most of the
charge in the latter case was transferred by positive
current, the largest charge content associated with
negative current was 606C. Of the 27 strikes, only 11
strikes exceeded a duration of 300ms corresponding to
1.6% of all strikes. The median values of charge levels
for the strikes reported are: 45C, 47C and 140C for
negative, positive and bipolar flashes.
Subsequent strokes were only detected in 13% of the
strikes, hence the discharging had a more continuous
manner i.e. higher charge levels, longer duration and
less peak current.
For winter lightning the cloud base is only 1-2km above
ground, compared to the typical cloud base height
during summer of 8km. This also explains the larger
number of strikes, the larger fraction of UW strikes and
the lower amplitudes.
Consequently the following conclusions apply in
relation to the lightning environment defined in IEC 623051 [2]:
Attachment
point
Peak current
Charge
Specific
Energy
99.7% upward initiated strikes, 99.6% of these
attached directly to the blade tip.
<0.8% exceeds 50kA.
IEC 62305-1 [2]: <5% of positive flashes exceeds
250kA, <5% of negative flashes exceed 90kA,
fixed LPL1 value of 200kA
<4% exceeds 300C, 0.3% exceeds 1000C.
IEC 62305-1 [2]: <5% of positive flashes exceeds
350C, <5% of negative flashes exceed 40C, fixed
LPL1 value of 300C
<0.3% exceeds 10MJ/Ohm.
IEC 62305-1 [2]: <5% of positive flashes exceeds
15MJ/Ohm, <5% of negative flashes exceed
0.55MJ/Ohm, fixed LPL1 value of 10MJ/Ohm
B. Mitigation of Rapid redistribution of Charge at FAA
facilities
The paper presented at ICOLSE 2013 by C.M. Graves
[13], discusses some incidences happening to air traffic
control towers in US airports. In several cases the
equipment within the control tower was damaged, but
without experiencing a direct strikes. Correlation with
NLDN data indicated that large amplitude cloud to ground
strikes will trigger upward discharges from the tall towers.
The explanation is rapid redistribution of the charge centres
within the cloud, enabling triggering of upward discharges
from the control towers.
Due to the crucial importance of flight safety and
reliable air traffic control, the Federal Aviation
Administration (FAA) had this type of lightning issues in
focus.
Attachment
point
Peak
current
Charge
Specific
Energy
Upward triggered strikes to tall structures,
comparable to attachment to the tip regions of the
blade in the Wind turbine analogy
Not commented, besides that it is a large but remote
stroke triggering the upward discharges from the tall
structures.
Not commented.
Not commented.
4
2.
3.
4.
5.
The majority of strikes to wind turbines are expected to
be upward initiated. If the tools used today therefore
only consider downward strikes, the number of strikes
will be heavily underestimated.
The rotation of the blades may play an important role,
since they may trigger strikes themselves due to their
size. The triggering will be upward initiated and hence
occur form the tip.
When designing protection of electronic systems, one
must consider the reflection occurring when current
pulses are injected in structures with a large extension.
This might be a problem if the rise time is very short,
and the structure very large.
Bonding between CFC and down conductors must be
considered carefully, to avoid internal voltages creating
flashovers, and also to consider the path for the static
discharge current.
It should be considered wether eddy currents induced
into the CFC laminate by the lightning current will
dissipate energy enough to weaken the structure.
Consequently the following conclusions apply in
relation to the lightning environment defined in IEC 623051 [2]:
Attachment
point
Peak current
Charge
Specific
Energy
Upward triggered strikes are expected to be
dominant, attachment to the tip of the blades.
Not commented.
Not commented.
Not commented.
D. Proposal of new Zoning concept for wind turbine
blades
In 2010 at the ICLP and again in a Journal paper in 2012,
the idea of a zoning concept for wind turbine blades was
published [3].
The Zoning concept was developed to present an
engineering tool for assessing which lightning strikes attach
to the different regions of the blade. Historically the
lightning protection standards have described how the blade
should be protected to LPL1 (200kA strike) from the tip and
down to 20m from the blade root (radius 20m or R20),
whereas no evidence of such exposure was presented. Both
numerical simulations defining the attachment point
distribution [2] as well as extensive field data provided
evidence that the exposure from direct strikes was focused
on the tip of the blade, and that the peak current of strokes
expected inboard were of limited amplitude.
The suggested concept therefore defined regions of the
blade that would be exposed to certain peak currents, and
then based on the standardized waveforms in IEC 61400-24
[1], the test requirements and input for further numerical
modeling could be derived. The concept does not define
accurate positioning of air terminations or requirements to
The Zoning Concept from 2012 have been used in
several design projects for blade lightning protection, and
has also been adopted by some of the certifying bodies.
100,0%
80,0%
27,6%
40,5%
1.
The zoning concept for wind turbines is derived based
on numerical simulation and site inspections. The numerical
simulations are conducted for generic blades of 40m
lengths, and in this process the distribution of strikes
attaching to the turbines at different amplitudes was
investigated. The results showed clearly that for downward
initiated strikes the majority of strikes attach to the blade
tips, and that for lower amplitude strikes, the attachment can
move inboard on the blade and attach to other and less
exposed parts of the wind turbine (Hub, Nacelle, tower,
etc.). The findings were correlated with field inspections,
which also pointed in the direction that the blade tips are the
most exposed parts of the turbines. For the field inspection,
the estimation of the size of damage or attachment point vs.
the blade radius was not published, so the tendency from the
simulations that high amplitude strikes attach to the tip and
smaller amplitude strikes attach further inboard could not be
proved. An example of the field data governing the
lightning attachment distribution is seen in Fig. 2.
60,0%
40,0%
9,7%
The major conclusions are:
insulation levels of internal parts, this is identified in a
different process called lightning protection coordination.
Attachnment point distribution
C. Review of current issues in wind turbine lightning
protection
The paper by F. Rachidi et al. reviews some different
issues of wind turbine lightning protection, including
attachment points, current distribution in Carbon Fibre
Composites (CFC), electronics etc. [14].
20,0%
0,0%
Length of blade [m]
Figure 2. Attachment point distribution of 2818 identified lightning
attachment points on 120 blades (45m).
E. Damage statistics and triggered lightning [15]
In the PhD thesis by A.Candela [15], damage statistics
from several wind farms in the US is reported. The data
considered five years of operation of a total of 508 turbines,
all with blades exceeding 35m length. During that period of
time, 304 lightning incidences causing damages to the
blades are reported, and by visual inspection of all damages,
the distributions along the blades are determined.
For blades manufactured using only fibre glass (64.8%
of the population i.e. 197 blades), the damage distribution is
seen on Fig. 3. Concerning blades using CFC structural
components, the damage distribution is comparable. The
population considered 107 blades, and again the tendency
shows that the blade tips are subjected to the largest number
of strikes. The distribution is seen in Fig. 4. By analyzing
the complete population of blades, it is found that more than
60% of all damages occur to the extreme tip, 90% occur to
the outermost 4m, and the remaining 10% is predominantly
located from 5-10m from the tip.
5
Such observations are made by other researchers as well,
one of these measurement campaigns are to be published by
W. Risoon et.al at ICOLSE’2015.
V.
REVISED ZONING CONCEPT
After the first suggestion of the Zoning Concept for
three blade rotors, lightning protection systems on blades
exceeding 80m lengths have been designed. By conducting
the detailed attachment point distribution analysis on such
longer blades, it was found that smaller amplitude strikes
may attach further inboard on the blades.
Figure 3. Location of the lightning damage in percentage for GFRP blades
[15].
Figure 4. Location of the lightning damage in percentage for blades
containing CFC as structural components and GFRP in the shells [15].
These data obviously focus on blade damages, whereas
successful lightning attachments are not registered nor
commented. By experience, such attachments occurs to the
extreme parts of the blades, and since all the blades are fitted
with different air termination concepts in the tip region, it is
expected that the attachment point distribution will move
even further out.
A second finding in [15] shows that large cloud to
ground discharges may result in the simultaneous triggering
of upward discharges from the wind turbines. These
findings were determined by correlating wind turbine
SCADA alarms with the data measured by Lightning
location systems, and proved in five cases how CG strikes
occurring as far as 10km away can trigger the strikes.
F. Lightning discharges produced by wind turbines [16]
In a publication by J. Montanyà in 2013, evidence is
presented explaining how the rotation of wind turbines may
play an important role in the triggering of upward lightning
strikes. By utilizing 3D Lightning Mapping Arrays (LMA)
and video surveillance at a wind turbine site in the northeast
of Spain, repetitive discharges correlating well with the
rotational speed of the turbines are identified. Pictures are
captured showing how multiple upward leaders are incepted
from the blade tips, and it is concluded that rotation, and the
movement of blade tips relative to the static space charge
generation can explain the observations.
The process involved the use of numerical models of
downward leader propagation and the following inception
of upward leaders from the structures proposed by Becerra
[7], to determine which parts of the wind turbine are
exposed to direct attachment of different amplitudes. The
equations outlined in the papers by the Uppsala lightning
research team has been implemented in Comsol and Matlab,
to enable import of a 3D turbine geometry and analyzing the
exposure [5]. For a generic turbine, and considering vertical
DW leaders the difference in amplitude is seen on Fig. 5.
The dots on Fig. 5 indicate the tip of the downward
moving leader at the moment when a successful upward
leader is incepted from the turbines. For this configuration
of turbine size and vertical incoming leaders, the blade tips
will receive all strikes and share the attachments evenly. The
plots also indicate that when lowering the current from
40kA to 20kA, the distance between the turbine and the
downward leader (the striking distance) decreases,
eventually enabling strikes inboard the blade tip at very low
amplitudes of 3-5kA.
The principle using inclined leaders with prospective
peak currents of 3-20kA has been applied on a generic
turbine structure with 60m blades. On Fig. 6, the
percentages of strikes attaching at each blade radius
(averaged over all three blades for different rotor angles) is
plotted for different prospective peak currents. Note that the
peak of the scale is set to 3%, meaning that the actual
fraction of strikes attaching to the tip region for strikes of
higher amplitudes cannot be seen. The results indicate
clearly that for higher current amplitudes, the attachment
tends to move towards the blade tip.
Realizing that the 3kA or 5kA strikes may attach further
inboard on the blade, changed the original Zoning concept
in [6] where strikes only could attach at the outer 20m of the
blades. The revised concept shown on Fig. 7 includes a
Zone 0A4 enabling direct strikes of 10kA for the entire blade
length.
The consequence of extending the direct strike zone and
using the Zoning Concept for blade LPS design, is then that
the inboard sections should also be capable of withstanding
direct strikes of 10kA. This may be achieved quite simply
for blades with CFC in the shells, which can then be
designed to accommodate the direct strikes, but for GFRP
blades, the likely hood of a puncture through the root section
must be addressed.
The revised zoning concept currently used for three
bladed rotors is seen on Fig. 7.
6
Figure 5. Simulation of the attachment point distribution for vertical DW leaders at amplitudes of 40kA (top), 20kA (bottom) [5].
Figure 6. Attachment to the blade vs. the blade radius (60m blades). Comparison between the attachment simulation results for different Ipeak=3, 10kA. For current
larger than 10kA attachment is most likely within 5m from the tip (between 70-90% of the attachments), while currents around 3kA can attach along the entire
length of the blade.
Zone 0A1
Zone 0A4
Zone 0A3
Zone 0A2
1m
XX m
15m
4m
Zone OA1: Tip end to 1m inboard, <200kA
Zone OA2: 1m inboard to 5m inboard, <100kA
Zone OA3: 5m inboard to 20m inboard, <50kA
Zone OA4: 20m inboard to root end, <10kA
Figure 7. Zoning concept for three bladed rotors enabling direct attachment on the entire blade surface.
The individual zones in the revised Zoning Concept is
defined by specific lengths measured from the blade tip, since
simulations have shown that the distributions are unaffected by
the typical ranges of blade lengths (20-80m). The individual
zones are outlined in the following


Zone 0A1: The outermost 1m tip section exposed to the full
threat - direct attachment with a maximum peak current
corresponding to LPL1 in [2] - 200kA, 10/350us
Zone 0A2: The section of the blade from 1m inboard the tip
to 5m inboard the tip, exposed to direct attachment with
current levels of only 100kA, 10/350us

Zone 0A3: The section of the blade from 5m inboard the tip
to 20m inboard the tip, exposed to direct attachment with
current levels of only 50kA, 10/350us

Zone 0A4: The section of the blade from 20m inboard the
tip to the root end of the blade, exposed to direct attachment
with current levels of only 10kA, 10/350us
The zoning concept does not dictate where to place receptors
or air termination systems, it is only used to assess the possible
strike amplitudes to different regions on the blade. Although it
is not strictly formulated how to interpret the LPL1 requirements
in the IEC 61400-24, it can be interpreted such that strikes with
amplitudes between 3kA and 200kA must be safely intercepted
and conducted towards ground, whereas damages are tolerated
7
for strikes outside these extremities. In practice concerning the
attachment process, it means that since strikes may occur to the
inboard sections of the blade (even with a very low probability),
the blade must be capable of handling it. Hence impulse current
tests to inboard blade sections of 3-10kA has been conducted to
provide evidence of only limited damages to the blade at such
an exposure.
If the probabilities of having such small amplitude strikes to
the blades are accounted for by considering the probability
density functions described in the lightning protection standard
[2], one can come to the conclusion that protection according to
strikes of such low amplitudes is unnecessary, because they only
occur very rarely or because the likely damage due to
interception failure is very limited.
The zoning concept do not apply for two bladed rotors,
where the profound shielding effect by the blade tips on a three
bladed rotor is not as clear. Attachments on two bladed rotors
are still favored at the blade tips, but larger strikes may also
attach further inboard.
VI.
DESIGN INPUT
Outlining the research on the different lightning exposure of
wind turbine blades enables a design guideline or a list of best
practices in how to achieve proper LPS on blades. The following
statements are derived:
1.
2.
3.
4.
5.
Tall structures experience a large portion of upward
initiated strikes for normal lightning conditions, a fraction
which approach 100% for low cloud bases during winter
lightning.
Upward lightning are triggered from the blade tips and are
in general more frequent, but of lower amplitude and
specific energy.
Large cloud to ground flashes can trigger simultaneous
upward lightning from many turbines. These triggered
strikes are of low amplitude, but occur very frequently.
The damage distribution experienced on GFRP and CFC
blades seems to be very similar, hence the environment and
the methodologies to assess the attachment point
distribution can be the same.
Damages experienced on CFC blades can occur for the
same reasons as for GFRP blades, but may also be caused
by attachments directly to the CFC structures, explained by
inadequately placed air terminations and poor insulation of
CFC parts especially in the tip region (outermost 3m of the
blade).
Another failure scenario for CFC blades is if equipotential
bondings have not been implemented, if the bondings are placed
wrongly, or if the current carrying capability of the bondings is
inadequate. One may note that the external appearance of
damage due to lack of equipotential bonding may be very similar
to the effects of a direct lightning strike on the blade surface.
Therefore, the damages located far inboard should be carefully
inspected to not misclassify them as direct attachment.
VII. THE GOOD DESIGN PROCESS
point analysis, or by just adopting the Lightning Zoning Concept
for blades, and deciding the severity of the site (the number of
impacts to expect during the lifetime or between scheduled
services).
Secondly, the number and positions of the air terminations
on the blade must be aligned with the blade design, such that
analysis and HV strike attachment test can provide evidence that
strikes will only be intercepted by dedicated air terminations.
Once the design is fixed from an analysis standpoint, the
verification tests as described by the IEC 61400-24 can be
conducted. Concerning the interception effectiveness, the High
Voltage strike attachment test according to Annex D2.1 is
performed. The blade is tested in three angles relative to the
reference plane, and in four different pitch angles eventually to
show that all flashovers attach to the air terminations.
The following steps address the good blade design process:
A. Fibre glass blades (GFRP)
1. Identify the site, and asses the number of UW/DW strikes
following the equations in section 3.4.
2. Assume that all UW strikes happen from the blade tip, and
therefore design a proper tip air termination.
3. Identify the probability distribution of DW strikes (either
via IEC 62305-1 [2] or via future lightning statistics), and
place air terminations within the outermost 5m of the blade
to accompany the blade design.
4. Maximize the insulation level for down conductors in the
outermost 10m of the blade
5. Install air terminations above connections on the down
conductors (i.e. when changing from the insulated down
conductor from the tip to a simpler conductor
configuration).
6. Conduct the HV strike attachment test [1] on the outermost
20% of the blade length, to identify the air termination
effectiveness.
7. Setup a high current test plan based on the expected UW
and DW strikes and their distribution, and conduct the test
to simulate the expected lifetime of the blade according to
[15].
B. Blades containing CFC
1. Identify the site, and asses the number of UW/DW strikes
following the equations given above.
2. Assume that all UW strikes happen from the blade tip, and
therefore design a proper tip air termination.
3. Identify the probability distribution of DW strikes (either
via IEC 62305-1 [2] or via future lightning statistics), and
place air terminations within the outermost 5m of the blade
to accompany the blade design.
4. Maximize the insulation level for down conductors in the
outermost 10m of the blade
5. Install air terminations above connections on the down
conductors (i.e. when changing from the insulated down
conductor from the tip to a simpler conductor
configuration), and where the blade shifts geometry
electrically (tip of CFC parts).
The first step of blade LPS is to define the exposure, either
by doing a site based risk assessment along with an attachment
8
6.
7.
8.
Conduct the voltage/current distribution analysis, to
identify where to place equipotential bondings and which
requirements to apply for each of them.
Conduct the HV strike attachment test [1] on the outermost
20% of the blade length, to identify the air termination
effectiveness.
Setup a high current test plan based on the expected UW
and DW strikes and their distribution, and conduct the test
to simulate the expected lifetime of the blade according to
[17]. This test includes all connection components,
equipotential bondings, additional current paths, etc.
VIII. CONCLUSION
The paper addresses a need for an engineering tool useful for
LPS designers and still accounting for the slightly more complex
lightning exposure experienced on large wind turbines.
The revised Zoning Concept provided along with the short
guideline to achieve proper LPS designs for different blade
types, will ensure that lightning engineers focus the attention
towards the areas of the blades where lightning exposure is
highest. During the design phase and for the final verification,
the Zoning Concept is also used to assess the test parameters.
[11] R. B. Anderson and A. J. Eriksson, ‘Lightning Parameters for Engineering
Applications’, Electra, vol. 69, pp. 65-102, 1980.
[12] M. Ishii, D. Natsuno og A. Sugita, ‘Lightning Current Observed at Wind
Turbines in Winter in Japan’, Proceeding of the International Conference
on Lightning and Static Electricity, Seattle, WA, USA, 2013.
[13] J. C.M. "Chuck" Graves, ‘Assessing and Mitigating Rapid Redistribution
of Charge at FAA Facilities’, Proceedings of the International Conference
on Lightning and Static Electricity, Seattle, WA, USA, 2013.
[14] F. Rachidi, M. Rubinstein, J. Montanyà, J.-L. Bermúdez, R. R. Sola, G.
Solà and N. Korovin, ‘A Review of Current Issues in Lightning Protection
of New-Generation Wind-Turbine Blades’, IEEE Transactions on
Industrial Electronics, årg. 55, nr. 6, pp. 2489-2496, 2008.
[15] A.C. Garolera, ‘Lightning Protection of flap system for wind turbine
blades’, PhD thesis, Tehcnical University of Denmark, September 2014
[16] J. Montanyà, O.van der Velde and E.R. Williams (2014), ‘Lightning
discharges produced by wind turbines, J. Geophys. Res. Atmos., 119,
doi:10.1002/2013JD020225
[17] Bertelsen, K., H.V. Erichsen and S.F. Madsen: ‘New high current test
principle for wind turbine blades simulating the life time impact from
lightning discharges’, Proceedings of the International Conference on
Lightning and Static Electricity 2007, August 28-31, Paris, France.
Following the publication and the ongoing revision of the
IEC 61400-24, initiated March 2014, the revised zoning concept
will be sought implemented in the upcoming version of the
standard. By highlighting the special exposure on wind turbines
relative to regular buildings the lightning protection system
effectiveness will be improved, and most likely also result in a
reduction of the cost of the overall LPS.
Finally having the methodology described in the
international standard, makes the certification process easier for
the certifying bodies.
IX.
REFERENCES
IEC 61400-24 Ed. 1.0, ‘Wind Turbines – Part 24: Lightning Protection,
IEC, 2010-06.
[2] IEC 62305-1 Ed. 2.0, ‘Protection agianst lightning – Part 1: General
Principles’, IEC, 2011-03-22
[3] Madsen, S.F., K. Bertlesen, T.H. Krogh, H.V. Erichsen, A.N. Hansen and
K.B. Lønbæk: ‘Proposal of New Zoning Concept Considering Lightning
Protection of Wind Turbine Blades’, Journal of Lightning Research, 2012,
4, (Suppl 2:M8) 108-117.
[4] M. A. Uman, The Lightning Discharge, Mineola: Dover, 2001.
[5] Madsen, S.F. and H.V. Erichsen: ‘Numerical model to predict attachment
point distributions on wind turbines according to the revised IEC 6140024’, Proceedings of the International Conference on Lightning and Static
Electricity 2009, September 15-17, Pittsfield MA, USA.
[6] Cooray, V., V. Rakov and N. Theethayi, ‘The relationship between the
leader chage and the return stroke current - Berger's data revisited’
Proceedings of the International Conference on Lightning Protection,
Avignon, France, 2004.
[7] Becerra, M. and V. Cooray, ‘A simplified Physical Model to Determine
the Lightning Upward Connecting Leader Inception’, IEEE Transactions
on Power Delivery, vol. 21, nr. 2, 2006.
[8] Rakov, V. A. and M. A. Uman, Lightning - Physics and Effects, New
York: Cambridge, 2003.
[9] CIGRE Report (Doc. 63, 1991)
[10] Berger, K., R. B. Anderson and H. Kroeninger, ‘Parameters of Lightning
Flashes’, Electra, vol. 41, pp. 23-37, 1975.
[1]
9