See

discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/268037903

Attenuation and deformation characteristics of

lightning impulse corona traveling along
bundled transmission lines
ARTICLE in ELECTRIC POWER SYSTEMS RESEARCH JANUARY 2015
Impact Factor: 1.75 DOI: 10.1016/j.epsr.2014.07.020

READS

161

5 AUTHORS, INCLUDING:
He Jinliang

Xiao Zhang

Tsinghua University

Tsinghua University

468 PUBLICATIONS 2,941 CITATIONS

4 PUBLICATIONS 5 CITATIONS

SEE PROFILE

SEE PROFILE

Shuiming Chen

R. Zeng

Tsinghua University

Tsinghua University

57 PUBLICATIONS 266 CITATIONS

216 PUBLICATIONS 914 CITATIONS

SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate,

letting you access and read them immediately.

SEE PROFILE

Available from: He Jinliang

Retrieved on: 23 November 2015

Electric Power Systems Research 118 (2015) 2936

Contents lists available at ScienceDirect

Electric Power Systems Research

journal homepage: www.elsevier.com/locate/epsr

Attenuation and deformation characteristics of lightning impulse

corona traveling along bundled transmission lines
Jinliang He , Xiao Zhang, Pengcheng Yang, Shuiming Chen, Rong Zeng
Department of Electrical Engineering, Tsinghua University, Beijing, China

a r t i c l e

i n f o

Article history:
Available online 12 August 2014
Keywords:
Lightning impulse
Bundled conductors
Attenuation
Deformation
Traveling lightning overvoltage

a b s t r a c t
Comprehensive study of the impulse corona characteristics is helpful to understand the transient process
and guarantee a suitable insulation design of power transmission systems. In this paper, rst the attenuation and deformation characteristics of lightning overvoltage traveling along a 500 kV experimental
transmission line are tested. Then the lightning impulse corona characteristics on different bundled conductors commonly used in the 1101000 kV power transmission lines are experimentally investigated,
the impulse corona QV data measured in the corona cage is transformed into those on the corresponding
overhead lines. The inuence of bundle geometry on corona is analyzed. Attenuation and deformation
effects of lightning impulse corona along transmission lines are evaluated by comparing the simulation
and test results. Simulation shows the amplitude and wave-front steepness attenuate remarkably, and
the deformation of the voltage mainly happens when the impulse voltage exceeds the corona inception
one.
2014 Elsevier B.V. All rights reserved.

1. Introduction
When lightning transient travels along an overhead line, the
voltage waveform will be deformed and attenuated by several
factors, including frequency dependent behavior of the conductor impedance, skin effect and the impulse corona discharge [13].
Among them, the corona decreases the conductor impedance and
increases the coupling factors between lines as well. In order to
precisely calculate the lightning overvoltage of power systems, the
characteristics of the impulse coronas on different transmission
bundled lines need to be tested and analyzed comprehensively
[4,5].
Many researchers have studied the physics and discharge process of impulse corona [68]. In 1954, Wagner and Lloyd rstly
carried out impulse corona experiments on an outdoor overhead
line and measured the deformed voltage waveform [9]. Corona
cage experiments were conducted by Maruvada in 1970s to analyze the lightning and switching impulse corona on 4 different
kinds of bundled conductors [10]. During recent decades, several
other experiments have been carried out to obtain the lightning

Corresponding author. Tel.: +86 13601024327/10 62788811;

fax: +86 10 62784709.
E-mail address: hejl@tsinghua.edu.cn (J. He).
http://dx.doi.org/10.1016/j.epsr.2014.07.020
0378-7796/ 2014 Elsevier B.V. All rights reserved.

impulse corona characteristics of conductors in the corona cage or

conductors above the ground plane [1114].
However, there is still a lack of experimental data covering
the lightning impulse corona QV curves on bundled conductors
presently used in power transmission of different voltage grades.
In addition, how the geometry of different bundles inuences the
characteristics of the impulse corona also needs to be tested and
analyzed comprehensively.
This paper provides the latest data and analysis from the continuous experimental research on impulse corona characteristics
of different conductors commonly applied in 1101000 kV power
transmission systems, carried out in China UHV AC Test Base,
Wuhan. Firstly the attenuation and deformation characteristics
of the traveling lightning overvoltage are observed on a segment
of 500 kV overhead transmission line, then the impulse corona
QV curves measured in the corona cage are transformed into
those on the corresponding overhead lines to build the impulse
corona models for transmission conductors used in different voltage levels. The characteristics and inuential factors of the lightning
impulse corona are comprehensively investigated. Finally, the simulated deformation and attenuation effects of impulse corona along
the transmission line are compared with the tested one, and
the lightning transients on an actual UHV AC transmission line
under impulse corona effect is studied. This paper is an extension of our conference paper [15], which has been completely
revised.

J. He et al. / Electric Power Systems Research 118 (2015) 2936

Voltage divider
Measuring point 3

Matched
impedance

Voltage divider
Measuring point 2

Test line

Voltage divider
Measuring point 1

200 m

Lead
150 m
UHV support
insulator
4800 kV Impulse
voltage generator

Impulse voltage (kV)

30

-500
0
-300
-600

-1000

-900
-1200
0.0

-1500
Fig. 1. Experiment setup of lightning impulse corona on the test line.

At head end
At the middle point
At the tail end

0.0

2.0x10

-5

-6

2.0x10

4.0x10

-6

4.0x10

-5

-5

6.0x10

Time (s)
2. Lightning impulse corona experiment on the specimen
transmission line
First, a 200 m testing section of 500 kV transmission line is set
up to investigate the inuence of lightning impulse corona on the
wave deformation over actual high-voltage bundled transmission
lines. In this experiment, lightning impulse voltages are applied by
the generator at the head end of the line, and the noninductive
matching resistor is connected between the line and the ground at
the tail end. Impulse voltage dividers are installed to measure the
traveling instantaneous voltages at different points along the line.
2.1. Equipment setup
The diagram and the photograph of the experiment setup are
shown in Figs. 1 and 2(a) respectively. The experiment setup consists of a 4800 kV/720 kJ impulse generator as shown in Fig. 2(b), the
test line, impulse voltage dividers, the lead wire and the matching
resistor.
The length of the test line is 200 m, with bundled conductors
of 4LGJ630/45 s450, the same as applied on 500 kV AC power

Fig. 3. Tested impulse voltage deformation along the HV line.

transmission in China. The heights of the hanging points on the

towers of both sides are 21.6 m and 18.6 m, respectively.
The tail end of the test line is connected to the ground through
the matching resistor, avoiding the inuence of the reection
waves on the voltage measurement. The non-inductive resistor is
applied to eliminate the voltage drop caused by the high-frequency
components under lightning impulse, which consists of six 50 
non-inductive resistors in series, as shown in Fig. 2(c).
A 150 m HV lead wire is used to connect the generator and test
line. To decrease the inuence of corona around the lead on the
measurement, a corrugated pipe is installed as the sheath outside
the lead wire.
The impulse voltages along the test line are measured by the
voltage dividers at three points: the head and tail end and the middle point, as shown in Fig. 2(d). The divider consists of 6 capacitors
in series, with a ratio of 4747 and a measurement uncertainty of
less than 2%. The transient parameters of the divider include the
step response time t 100 ns, and the overcharge voltage coefcient 20%. A four channels, 100 MHz, and 16 bit data acquisition
system is used to record the signals.
2.2. Impulse voltages along the test line

Fig. 2. Photographs of the experiment sets.

The magnitude of the applied impulse voltage in the experiment

is approximately 50% higher than the corona inception voltage of
the test line, which is 880 kV according to calculation. When a
1320 kV, 2.6/50 s lightning impulse voltage is applied at the head
of the test line, the instantaneous voltages on different measuring
points along the line are shown in Fig. 3.
In this case, the crest of the voltage at the middle point is
1242 kV, 5.7% less than that at the head end. The voltage crest
at the tail end is 1177 kV, 8.5% less than that at the head end. In
addition, the rise time of the surge (the time it takes to rise from 0%
to 100% of the maximum value) increases from 2.81 s at the head
end, to 2.90 s at the middle point, and 3.02 s at the tail end.
First, according to the experimental results, the attenuation and
deformation of the lightning overvoltage by impulse corona could
not be neglected, especially on high voltage bundled transmission
lines. Furthermore, to separately evaluate the deforming inuence
of corona out of other inuential factors such as the frequency
dependent nature of conductors, simulation investigation based on
corona model should be conducted.
Second, the characteristics of the impulse corona on transmission lines are correlated with several factors, such as the structure
and height of the bundled conductors, the bundle spacing, the number and radius of the subconductors, spaces among the phases,

J. He et al. / Electric Power Systems Research 118 (2015) 2936

31

Fig. 4. Electric-eld on the conductor surface: (a) overhead line (b) in the corona cage.

etc. It is useful to comprehensively investigate the impulse corona

effects on different kinds of bundled lines applied in power system
with various voltages.
3. Corona cage experiments and transformation to actual
transmission line
The investment on constructing overhead lines of different voltage level for test is not an economical way to carry out. Besides,
considering the convenience in measuring the space charge, corona
cage is used for the lightning impulse corona tests of different bundled conductors.
The corona process mainly depends on the surface elds around
the conductors. However, the electric eld distribution in a corona
cage is different with that on the actual overhead lines. Moreover,
the streamer of corona grows symmetrically around the conductor
in the cage but on real operating overhead lines it is asymmetrical.
So here, the relation on the surface electric eld between conductors in the cage and actual overhead transmission lines is studied,
and formulas are presented and validated to transform the QV
curve measured in the cage into that of the corresponding overhead
lines.
3.1. Transformation of corona characteristics between the
experimental cage and actual transmission line
According to the electric-eld calculation of the operating UHV
bundle conductors, the difference in the maximum surface eld
among the subconductors is less than 5%. So, Eavmax is dened as the
average maximum surface electric eld on different subconductors
in one bundle. In the corona cage, in contrast, the maximum surface
elds on every subconductor are almost uniform with a common
value of Ecmax , regarding their symmetrical conguration.
Basically, the criterion for corona onset is that the electric eld
on conductor surface exceeds the threshold value. And the inuence of bundle geometry parameters, including the bundle spacing,
subconductor radius, and suspending height, varies fundamentally
by changing the surface electric eld distribution of the conductor. Here, it is assumed that when the surface electrical eld on the
conductors in the cage is similar with that on the overhead line,
the corona characteristics measured in the cage can be transformed
into those for the actual overhead lines.
When the values of Eavmax and Ecmax reach the same as Em ,
according to Mangoldts derivation [16], the surface charge of the
bundle conductor qc is
qc =

20 nrb Em
[1 + (rs /rb )(n 1)]

where rs is the radius of the subconductor, rb is the radius of the

circle on which the centers of the subconductors lie, and n is the
number of subconductors.
The voltages on the conductors in the cage ucage and on overhead
line uline are:
ucage =

qc
Ccage

(2)

uline =

qc
Cline

(3)

where Ccage and Cline are the capacitance to the ground from the
bundle in the corona cage and on overhead line, respectively.
In this way, Eq. (4) is used to determine the relationship of the
voltages applied on the conductors under these two cases when the
corona degrees in both cases reach the similarity:
uline =

ucage Ccage
Cline

(4)

The nite element simulation software ANSYS is applied to

validate the results. Take the 1000 kV transmission line for example, Cline = 9.24 F/m and Ccage = 22.6 F/m. When the voltage on
the overhead line is 2090 kV in amplitude, the equivalent voltage
applied in the corona cage should be 855 kV according to Eq. (4).
The surface electric-eld on conductors under both conditions are
illustrated in Fig. 4. It can be seen that the eld distributions agree
well between both, which validates the formula (4).
3.2. Measurement of QV curves in the corona cage
The experimental setup of the UHV corona cage is previously
introduced in our paper [17], as shown in Fig. 5. The metallic-wire

Corona cage

Shielding
section

measuring
section

Shielding
section

Tested conductors

Coaxial
cable
Impulse Divider
generator

Coaxial
cable
Data
acquisition

(1)
Fig. 5. Experiment setup of HV corona cage.

Measure
capacitor

J. He et al. / Electric Power Systems Research 118 (2015) 2936

mesh cage is composed of a 25 m long measuring section in the

center, and two 5 m long shielding sections at each end. The tested
conductors are suspended inside the internal cage with a cross
section of 8 m 8 m. Three sets of wire connection hardware are
equipped to suspend different bundled conductors in the cage.
The impulse voltage is measured by the voltage divider, and the
space charge is measured by the capacitor connected between the
internal and external cages. Q and V signals are transmitted to the
data acquisition system by the coaxial cable with the same length
simultaneously.

Negative
Positive
Geometric capacitance

30

q(C/m)

32

qp

20

qn
10

4. Corona characteristics under lightning impulse

400

4.1. Corona inception voltage and delay time

u(kV)

800

1200

u
When the impulse voltage exceeds the corona inception value,
charges are generated in the space around the conductors, the
QV curve deviates from the geometrical capacitance and increases
rapidly after the knee point. Therefore, the knee point of the QV
curve is usually reckoned as the corona inception voltage of the
conductors.
It rst appeared in the test results of paper [18] that the knee
points of the same conductors under different amplitudes of positive lightning impulse are different, and the corona inception delay
time is proposed to indicate this phenomenon in our previous paper
[17]. The inception voltage, knee point voltages and corona onset
delay times on ten types of bundle conductors measured under
different magnitudes of positive lightning impulse are presented
in Table 1.
The values of inception voltage in Table 1 are obtained from
switching impulse experiments carried out on the same bundles, because the wave front time of the switching impulse is
250 s, hundred order times of the lightning impulse corona delay
time. The deviation between the knee point voltage on the QV
curves and the corona inception voltage increase with the lightning
impulse voltage amplitude on the same bundled conductors.
4.2. Inuence of the bundle parameters on lightning impulse
corona characteristics
In this section, the specic relation between lightning corona
characteristics and conductor parameters, namely the bundle spacing, number and radius of subconductors are analyzed base on
experimental results.
In general, when the bundle spacing or the number and radius
of the subconductors increase, the knee point of the QV curves is
enhanced, while the steepness of the curve slows down after the
corona inception point, resulting in a decreased area of the whole
curve as well.
When the voltage exceeds the corona inception value, the average dynamic capacitance Cdav is dened as (see Fig. 6):

explained that the equivalent cross-sectional area of the bundle

becomes larger and generates more space charges as a result when
the voltage exceeds the corona inception value. When the applied
voltage on the line is normalized, the corona space charge becomes
the main factor inuencing the relative dynamic capacitance of
the conductor. Similarly, when the radius of the subconductor
increases, the range of the impulse corona charge around the conductor expands, resulting in a larger relative dynamic capacitance,
as shown in Fig. 8. These trends agree with the deduction of impulse
corona capacitance empirical formula of bundled conductors presented by Podporkin [19].
In contrast, the bundle spacing, as another inuential factor, is
not of monotonic positive correlation with the dynamic capacitance. As shown in Fig. 9, for the 6LGJ630 conductors, the relative
dynamic capacitance rst rises when the bundle spacing changes
from 400 mm to 450 mm, but then declines and reaches the minimum when the bundle spacing keeps increasing to 500 mm.
On one hand, the bundle spacing affects the relative dynamic
capacitance of the conductors by inuencing the surface electric
eld. The smaller is the bundle spacing, the stronger is the mutual
inuence between the subconductors, and the more non-uniform
is the surface electric eld around the conductors. Therefore, conductors with greater bundle spacing generate less space charge,
which can explain why the conductors with a bundle spacing of
500 mm have the smallest relative dynamic capacitance among the
experimental results.

2.8

(5)

Since different bundled lines have different corona inception

voltages and geometrical capacitances. For a valid comparison, the
Cdav of different bundles are normalized into the relative corona
dynamic capacitance Cdav /C0 , where C0 is the geometrical capacitance of the conductors. Meanwhile, the lightning impulse voltages
applied during the experiment are also normalized by being divided
by the corona inception voltage U0 of each bundle.
The relative dynamic capacitances and energy dissipated by the
corona are the two main indexes analyzed here when different
types of bundled conductors are tested.
As shown in Fig. 7, the relative impulse corona dynamic capacitance increases with the number of subconductors, which can be

2630 s450
4630 s450
6630 s450

2.6

Cdav/C0

q
Cdav =
u

Fig. 6. Corona QV curve of the same conductor under lightning impulse of different
polarities.

2.4
2.2
2.0
1.8
1.5

2.0

2.5

3.0

umax/u0
Fig. 7. Inuence of bundles to the relative lightning impulse dynamic capacitance.

J. He et al. / Electric Power Systems Research 118 (2015) 2936

33

2.4
2.4

2.2

Cdav/C0

Cdav/C0

2.2

2.0

4300 s450
4500 s450
4630 s450
2.4

2.6

2.8

3.0

6630 s400
6630 s450
6630 s500

1.8

1.8
2.2

2.0

1.6

3.2

1.6

1.8

umax/u0

2.0

umax/u0

Fig. 8. Relative lightning impulse dynamic capacitance under different subconductor radius.

On the other hand, conductors with greater bundle spacing will

inherently have larger geometrical capacitance C0 , and a smaller
Cdav /C0 . That is why the relative dynamic capacitance of conductors
with 450 mm bundle space is smaller than that with 400 mm bundle
space.
The energy dissipated by lightning impulse corona is another
important indicator of the attenuating and deforming effect of
impulse corona during the transient process. It also changes with
the parameters of the conductors signicantly.
Comparatively, the number of subconductors has the most signicant inuence on the dissipated energy by impulse corona,

Fig. 9. Inuence of bundle spacing on the relative lightning impulse dynamic capacitance.

while the radius of subconductor has a weak inuence, whereas

the bundle spacing has the weakest correlation with the lightning
corona dissipated energy. As shown in Fig. 10, the energy dissipated
by the impulse corona increases rapidly when the number of subconductors rises from 2 to 6. This is because more subconductors
bring more spacing charges under the same voltage, and enhance
the energy loss.
As illustrated in Fig. 11, there is a little difference in the energy
dissipated by corona between conductors with different bundle
spacing, specically the energy presents only minor increase with

Table 1
Corona inception delay time of UHV conductors under positive lightning impulse.
Conductor

Corona inception
voltage (kV)

Magnitude of impulse
voltage umax (kV)

Knee point voltage u0 

(kV)

Deviation (u0  u0 )/u0

Corona onset delay

time t (s)

4LGJ300/50 s450

410

1026
1150
1274

620
710
780

51.2%
73.2%
90.2%

0.45
0.59
0.65

4LGJ500/35 s450

430

1039
1166
1298

740
780
820

72.1%
81.4%
90.7%

0.74
0.70
0.67

4LGJ630/45 s450

450

980
1117
1247

710
770
810

57.8%
71.1%
80.0%

0.78
0.76
0.71

6LGJ300/50 s375

530

1046
1169
1291

800
870
920

50.9%
64.2%
73.6%

0.76
0.83
0.81

6LGJ400/35 s400

540

910
1034
1154

770
800
840

42.6%
48.1%
55.6%

0.88
0.75
0.69

6LGJ630/45 s400

560

912
1037
1164

760
810
840

35.7%
44.6%
50.0%

0.80
0.70
0.62

8LGJ500/35 s400

625

935
1052
1180

780
820
850

24.8%
31.2%
36.0%

0.60
0.53
0.45

8LGJ630/45 s400

640

984
1130
1247

800
860
900

25.0%
34.4%
40.6%

0.50
0.47
0.50

10LGJ630/45
s350

700

914
1046
1166

800
860
920

14.3%
22.9%
31.4%

0.72
0.73
0.70

12LGJ630/45
s300

770

918
1039
1155

860
900
970

11.7%
16.9%
26.0%

0.85
0.84
0.86

34

J. He et al. / Electric Power Systems Research 118 (2015) 2936

15

W(J/m)

conductors and another bundle of six conductors). Specically, the

similar conclusion in [10] is obtained that the dynamic capacitance
of corona increases with the increase of conductor radius. Similarly,
the energy dissipated by the impulse corona increases with the
number and radius of subconductors. However, because of the limitation of experimental samples, it is difcult to discuss from paper
[10] the inuence of bundle spacing on corona dynamic capacitance
while the rest of parameters remain the same.

2630 s450
4630 s450
6630 s450

10

5. Model and applications of lightning impulse corona

5.1. Coefcients of the empirical model of impulse corona on
different bundled conductors

0
1.5

2.0

2.5

3.0

umax/u0
Fig. 10. Energy dissipated by lightning impulse corona under different numbers of
subconductors.

6630 s400
6630 s450
6630 s500

W(J/m)

10

q=

1.2

1.6

2.0

umax/u0
Fig. 11. Energy dissipated by lightning impulse corona under different bundle spacing.

the expansion of bundle spacing. In contrast, when the radius of

subconductor increases, the range of the space charges expands,
resulting in more energy dissipated by corona, as shown in Fig. 12.
Comparison with other literature:
Maruvada, et al. [10] carried out tests on four different lines
(two single conductors with different diameter, one bundle of four

W(J/m)

Cgeo u
q0 (u/u0 )

Ccond =

15

This section is mainly concerned with the empirical formula of

negative lightning impulse corona QV curve on bundle conductors,
and it is difcult to propose a uniform equation for positive one due
to the time delay phenomenon during its inception. Based on the
experimental results, it is summarized in our previous paper [17]
that the negative lightning impulse corona QV curve ts the power
function well after the voltage exceeds the corona inception value:

4300 s450
4500 s450
4630 s450

10

2.0

2.5

3.0

umax/u0
Fig. 12. Inuence of subconductor radius on the energy dissipated by lightning
impulse corona.

u < u0
k

Cgeo

q0

dq
k uk1
=
k
du (u0 )

(6)

u u0

Cgeo

u < u0
u u0 ,

u
>0
t

(7)

u
<0
t

where u0 is the corona inception voltage as show in Table 1, q0

is the charge at the corona inception point. Cgeo is the geometrical capacitance of conductors which are easily calculated by either
numerical or analytical methods. Ccond is the capacitance between
the bundled conductors to the ground considering impulse corona
effects. k is the power coefcient of the empirical formula, which
is derived from the regression analysis to the curves of different
bundled conductors, the value of which can be referred in [17].
In the simulation, the impulse corona is modeled as nonlinear capacitance added between the transmission line and ground.
The line is divided into several segments, the length of which is
discussed in [20].
5.2. Inuence of impulse corona on the traveling lightning surge
The models of the generator and tested transmission line are
built in EMTDC software, the frequency-dependent characteristics
of the lines are considered. The impulse corona model on bundled
conductors of 4LGJ630/45-s450 is established in the simulation
according to the experiment and tting results. The simulated
instantaneous voltages at the end of the test line with and without
corona are compared with the measured ones, which is illustrated
in Fig. 13.
The calculated waveform largely ts with the measured one at
the wave-front, especially the amplitudes of the impulse voltages
coincide well between both. The voltage crest at the end of the line
decreases from 1251 kV to 1172 kV when corona effect is considered, according to the simulation. However, a deviation between
the measured and calculated results is noticeable at the wave-tail,
which decreases faster in the measured waveform: this is because
the hysteresis of the Q-V curve is not being modeled during the
simulation.
In addition, the lightning impulse corona models of 8LGJ630/
45-s400 bundled conductors are applied in the simulation

Impulse Voltage on test line (kV)

J. He et al. / Electric Power Systems Research 118 (2015) 2936

Fig. 14 shows the waveform and amplitude change little without

considering the corona effect. When the lag of time due to travel
distance is eliminated, the three waveforms almost overlap, which
indicates that over a short traveling distance, the transmission line
itself has weak inuence on the attenuation of lightning overvoltage. However, the attenuation and deformation of the waveform
become obvious in Fig. 15, when the corona effects are considered.
Specically, the crest decreases to 5720 kV from 6210 kV while the
wave front time of the surge increases to 6.2 s from 5.5 s after
the a traveling distance of 2 km. The corona inception voltage of
the double-circuit UHV line is approximately 1700 kV. It can also
be seen from Fig. 16 that the impulse waveform distorts little when
the voltage is below 1700 kV, yet begins to deform remarkably after
the corona is incepted.

Without corona(Calculated)
With corona(Calculated)
Measured waveform

-500

-1000

-1500

0.0

3.0x10

35

-5

6.0x10

-5

6. Conclusions

Time (s)
Fig. 13. Simulated and measured results of the impulse voltage at the end of the
test line.

0.5 km
1 km
2 km

u(kV)

-2000

-4000

-6000

-8000
0

10

20

30

40

50

t(s)
Fig. 14. Lightning surges along the UHV line without corona.

0
0.5 km
1 km
2 km

u(kV)

-2000

-4000

The impulse corona on the bundled conductors has noticeable inuence on the deformation and attenuation of the lightning
transients. First, during a measurement carried out on a segment of
overhead 500 kV test line, 8.5% decrease of the voltage amplitude
and 7% increase of wave-front time are obtained under a lightning
impulse voltage of 1320 kV, 2.6/50 s.
The QV data measured in the corona cage need to be transformed into that on the corresponding overhead lines, considering
the geometry of the bundles and the cage. Impulse voltage of lower
amplitude should be applied in the corona cage test to generate the
similar surface eld around the bundle as on overhead lines. Based
on comprehensive experiments on different bundled conductors
applied in 1101000 kV power systems, it is found that under a
normalized impulse voltage Umax /U0 ranging from 1.5 to 3.0, the
relative lightning impulse dynamic capacitance increases signicantly with the number and radius of subconductors, but rst rises
and then declines with the bundle spacing. The energy dissipated
by impulse corona increases most signicantly with the number
of subconductors, and subconductor radius presents a weak inuence, whereas bundle spacing has the weakest correlation with the
energy dissipated by lightning impulse corona.
The empirical formulas describing impulse corona characteristics derived from experiments are applied in simulations. The
simulated impulse voltage agrees with the tested result on the
wave-front, but deviates on the wave-tail, since hysteresis in the
QV curve is not modeled. Simulation of an actual double circuit
UHV AC transmission line shows signicant attenuation on both the
amplitude and wave-front steepness under infulence of impulse
corona. The deformation of the voltage mainly happens when the
voltage exceeds the corona inception value.
Acknowledgments

-6000
0

10
t(s)

20

Fig. 15. Lightning surges along the UHV line with corona.

to investigate the lightning surge on the double-circuit UHV AC system operating in East China. The transmission line is represented
by a number of distributed-parameter sections of 50 m in length,
while the model of corona capacitance is connected at junctions
between each section. When a negative 2.6/50 s lightning current
of 50 kA strikes the phase line, the transients on the line at different
distances from the striking point are illustrated in Figs. 14 and 15,
without and with consideration of impulse corona respectively.

The authors are grateful to Wu Xiong, Xu Tao and other staffs

of UHV AC Test Base, for their support on the test elds and equipment.
References
[1] V. Cooray, N. Theethayi, Pulse propagation along transmission lines in the presence of corona and their implication to lightning return strokes, IEEE Trans.
Antennas Propag. 56 (2008) 19481959.
[2] M. Kudyan, C.H. Shih, A nonlinear circuit model for transmission lines in corona,
IEEE Trans. Power App. Syst. 100 (1981) 14201430.
[3] M. Mihailescu-Suliciu, I. Suliciu, A rate type constitutive equation for the
description of the corona effect, IEEE Trans. Power App. Syst. 100 (1981)
36813685.
[4] D.X. Gu, P.H. Zhou, M.H. Xiu, S. Wang, M. Dai, Y. Lou, Study on overvoltage and
insulation coordination for 1000 kV AC transmission system, High Voltage Eng.
32 (12) (2006) 16 (in Chinese).

36

J. He et al. / Electric Power Systems Research 118 (2015) 2936

[5] T.A. Ametani, T. Kawamura, A method of a lightning surge analysis recommended in Japan using EMTP, IEEE Trans. Power Del. 20 (2) (2005) 67875.
[6] T. Noda, N. Nose, N. Nagaoka, A. Ametani, A wave-front-time dependent corona
model for transmission-line surge calculations, Elect. Eng. Jpn. 129 (1) (1999)
68775.
[7] M. Davila, G. Cinvestav, J. Naredo, P. Moredo, A. Ramirez, Practical implementation of a transmission line model for transient analysis considering corona
and skin effects, in: IEEE PowerTech Conference, Bologna, 2003, pp. 2326.
[8] V.V. Bochkovskii, Impulse corona on single and bundled conductors, Elektrichestvo 7 (1966) 2228 (in Russian).
[9] C.F. Wagner, B.L. Loyd, Effects of corona on traveling waves, AIEE Trans. Power
App. Syst. 74 (1955) 858872.
[10] P.S. Maruvada, H. Menemenlis, R. Malewski, Corona characteristics of conductor bundles under impulse voltages, IEEE Trans. Power App. Syst. 96 (1) (1977)
102115.
[11] X.R. Li, O.P. Malik, Z.D. Zhao, A practical mathematical model of corona for
calculation of transients on transmission lines, IEEE Trans. Power Del. 4 (2)
(1989) 11451152.
[12] M. Mohammadi, B. Vahidi, Model of corona for transient study, in: IEEE TENCON2005, Melbourne, Australia, 2005, pp. 24012404.

[13] R. Davis, R.W.E. Cook, W.G. Standring, The surge corona discharge, The Institution of Electrical Engineers Monograph, 1960, No. 415 S.
[14] P.S. Maruvada, H. Menemenlis, R. Malewski, Characteristics of conductor bundles under impulse voltages, IEEE Trans. Power App. Syst. 96 (1) (1977)
102115.
[15] J.L. He, P.C. Yang, S.M. Chen, R. Zeng, Lightning impulse corona characteristic
of 1000 kV UHV transmission lines, in: International Symposium on Lightning
Protection (XII SIPDA), Belo Horizonte, Brazil, 711, 2013, pp. 152156.
[16] W.V. Mangoldt, Electrical fundamentals of bundle conductors, in: Bundelleitungen, Siemens-Schuckert-Werke A.G, Germany, 1942.
[17] P.C. Yang, S.M. Chen, J.L. He, Lightning impulse corona characteristic of 1000 kV
UHV transmission lines and its inuences on lightning overvoltage analysis
results, IEEE Trans. Power Del. 28 (4) (2013) 25182525.
[18] T. Narita, S. Okabe, Propagation characteristics of lightning surge with corona
discharge on UHV designed transmission line, Trans. IEE Jpn. 122-B (2) (2002)
307313 (in Japanese).
[19] G.V. Podporkin, A.D. Sivaev, Lightning impulse corona characteristics of conductors and bundles, IEEE Trans. Power Del. 12 (4) (1997) 18421847.
[20] S.J. Carneiro, J.R. Marti, Evaluation of corona and line models in electromagnetic
transients simulations, IEEE Trans. Power Del. 6 (1) (1991) 334342.