See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/256658218

Evaluation of lightning voltages base on experimental results: propsal for

revision of IEC 60060-1 and IEC 61083-2

Article · October 2002

CITATION READS

1 1,257

5 authors, including:

Sonja Monica Berlijn Fernando Garnacho

Statnett Universidad Politécnica de Madrid
40 PUBLICATIONS   179 CITATIONS    50 PUBLICATIONS   313 CITATIONS   

SEE PROFILE SEE PROFILE

Ernst Gockenbach Peter Werle

Leibniz Universität Hannover Leibniz Universität Hannover
269 PUBLICATIONS   3,323 CITATIONS    116 PUBLICATIONS   472 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

power transformer condition evaluation View project

Overvoltages in GIS View project

All content following this page was uploaded by Sonja Monica Berlijn on 30 May 2014.

The user has requested enhancement of the downloaded file.

 ELECTRA PAPER
7-th of February 2002

EVALUATION OF LIGHTNING IMPULSE VOLTAGES BASED ON

EXPERIMENTAL RESULTS.
PROPOSAL FOR THE REVISION OF IEC 60060-1 AND IEC 61083-2
Fernando Garnacho Ernest Gockenbach Sonja Berlijn
Pascual Simón Klaus Hackemack Peter Werle

Summary in the evaluation of lightning Considerable work has been

impulse voltages. devoted during recent years to
solve the problems associated
The procedures for evaluation with the evaluation of lightning
of lightning impulse wave- impulse parameters. This work
shapes in IEC 60060-1:1989 1. Introduction has been focussed mainly on
allow different interpretations the definition of a mean curve,
to be made of crucial wave Tests with lightning impulse but up to now no satisfactory
shape characteristics particu- voltages are specified in IEC solution has been found 3.
larly when there are Publication 60071-1 1 to
oscillations on the peak of the demonstrate the response of The techniques now available
impulse. In extreme cases these high voltage power system for the evaluation of digital
differences can lead to apparatus to transient records of waveform make it
inconsistencies in insulation overvoltages due to natural sensible to question whether the
dimensions with a consequent lightning occurring under existing impulse parameters
difficulty in achieving proper normal operating conditions. and their definitions 2 are the
insulation co-ordination bet- Definitions for lightning most suitable ones to use with
ween items of equipment. impulse tests are given in IEC digital recording and current
Publication 60060-1:1989 2. insulating materials.
An investigation, funded by the A standard lightning impulse is
described as a full impulse Whilst mathematical algorithms
European Community, was
having a front time T1 of 1.2 µs can be used to define various
conducted in the research
and a time to half value of 50 mean curves, it was decided
laboratories of KEMA, FFII-
µs. The value of the test voltage that a new investigation of
LCOE, Schering Institute, for an impulse without evaluation procedures should
NGC and TU Graz during the oscillations is its peak value Up. be done on the basis of the
period 1997 to 1999. The physical processes involved in
conclusions from this The recommended procedure in dielectric breakdown. This
investigation show that a the standard 2 for the report describes the
modification to the procedures determination of the peak value experimental studies of the
for evaluating lightning of a standard lightning impulse European Project to establish
impulse waveshape parameters with an overshoot or an the physical basis for a revision
can improve the reproducibility oscillation on the peak is of the requirements in the
in tests on high voltage ambiguous. The evaluation of standards 2 4. It includes a
equipment. an impulse when the proposal for a new evaluation
application of a standard procedure for lightning
A proposal is made for an lightning impulse results in a impulses.
amendment to the requirements non-standard waveform, as in
of IEC 60060-1:1989 and IEC tests on transformers, is not
61083-2:1999 that would addressed in the standard.
remove the present ambiguities

Page 1 of 6
 ELECTRA PAPER
7-th of February 2002

2. Experimental
Studies
Table -I. Test program
The aim of the project was to
study the effects of Determination of the U50- Breakdown
superimposed oscillations on Parameter
U50 impulse breakdown Test Parameter to vary to keep
voltages in air, oil, paper-oil, Group constant
SF6 gas and XLPE dielectric Configuration Limits Fixed value
media. Oscillations up to 5 Polarity (+ and -) 0.5s  T1  2.0s T2 = 50s
MHz and 20% amplitude were I Homogeneous and
superimposed on smooth 1.2/50 non-homogeneous 40s  T2  60s T1 = 1.2s
standard lightning impulses. field
Polarity (+ and -) 0.2MHz  f  5MHz A = Ai
The test program shown in II Homogeneous and 0.2s  d  5.0s T1 = 1.2s
Table I was divided into two non-homogeneous Ai = 5%, 10%, 20% T2 = 50s
main groups. Group I deals field
with the influence of the time Note: f = oscillation frequency; d = overshoot duration;
parameters T1 and T2. Group II A, Ai = oscillation or overshoot amplitude.
deals with the influences of the
duration of the frequency/-
overshoot, and the amplitude of
the frequency/overshoot on the
breakdown voltage.
1 2
Each laboratory developed its 1 2
own test generator and
measuring system. The test 6
arrangements comprising 5 7
Control Recorder
generating circuit, test vessel, 7
5 6
measuring system and evalu-
ation software are described in
detail in 5. Using the general
schematic diagram of the test 4
circuit shown in Fig. 1, test 3 3
voltages up to 120 kV were
achieved for all dielectric
media. Tests on air gaps were
made with voltages up to 500 Figure 1. Schematic test circuit used in the European project: 1- Generator of
kV using a conventional smooth LI. 2- Measuring system of LI according IEC 60060-1. 3- Generator of
lightning impulse generator oscillating waves shapes. 4- Measuring system with an adequate bandwidth for
modified by means of series the oscillation measurements. 5- Trigger control and impulse analysis. 6- Digital
inductance. recorder. 7- Test vessel.

Both measuring systems in shape, with oscillation or to achieve sufficient accuracy

Fig. 1 shared the same digital overshoot, applied to the test for superimposed high fre-
recorder with two channels, one specimen. quency oscillations of 2 MHz
for each measuring system. The and 5 MHz. All laboratories
difference between the two A special design of the used the same software as
waveshapes recorded on each measuring system for the described in 6.
channel represents the impulse oscillating generator was made

Page 2 of 6
 ELECTRA PAPER
7-th of February 2002

3. Results of the is the peak value of the 3.2 Influence of time

measured impulse when parameters
researches superimposed noise is removed.
The conclusion from the results
3.1 Influence of the Values of these two relevant of tests in air, oil, SF6 and
superimposed oscillations parameters for a homogeneous XLPE is that the 50%
and overshoot field in air are shown in Fig. 2 breakdown voltage does not
for two overvoltage levels ß1= change significantly when the
For all of the dielectrics tested 3% and ß2= 13% (of the impulse front time varies from
(air, oil, oil-paper, SF6, XLPE) reference value). The results 0.5 s to 2 s, and when the
no difference was detected show that for frequencies time to half value varies from
between the influence of higher than 2 MHz, Upmc(50) is 40 µs to 60 µs. This suggests
overshoot and oscillation of closer to the reference value 1 that the tolerance for the front
equivalent amplitude and pu. For oscillation frequencies time defined in [2] as 30%
frequency (f=0.5/d). between 0.3 MHz and 2 MHz could be increased to 50%.
the reference value 1 pu lies
The same test vessel and between Ue(50) and Upmc(50). For Oscillations on the front of an
specimen characteristics and frequencies lower than impulse were shown to have a
properties were used for all 0.3 MHz, Ue(50) is closer to the negligible influence on the
tests carried out for each reference value. breakdown voltage. Therefore
dielectric medium and with oscillations between 30%
electrode configuration with The overvoltage ß is defined as: and 90% of the extreme value
impulses smooth and having of the impulse (Ue), the front
superimposed oscillations. The ß = Ue(50) - Upmc(50) (1) time was determined by using a
50% breakdown voltage for regression line of that part of
each dielectric medium and In general ß is not identical to the recorded impulse between
electrode configuration using a the amplitude of the oscillation the first crossing of 0.3Ue and
smooth 1.2/50 impulse was according to the definition in the last crossing of 0.9Ue.
taken as a reference value IEC 60060-1:1989. It is only
Uref(50)= 1pu. The 50% break- when Upmc and Ue occur at the
down voltage for impulses with same time that ß is identical to
oscillations were evaluated the amplitude of the oscillation.
from the extreme value of the 4. Application of the
impulse Ue(50) and from the
peak value of the mean curve test voltage factor k
Upmc(50). The extreme value Ue

4.1 Test voltage according to

Ue(50)   = 3% ,  2 = 13% the present IEC 60060-1
1,15

Ue(50) Overshoots or oscillations in

1,10
Ue(50) the region of the peak of an
1,05 impulse, measured by means of
an approved measuring system
1,00
7, are tolerated provided their
0,95 single peak amplitude is less
Upmc(50)  Ue(50) than 5% of the extreme value
0,90
 Ue(50)(1) 2. Depending on the
 Upmc(50)(2)
frequency of the oscillation or
0,85 Upmc(50) Upmc(50)(1)
the duration of the overshoot,
0,80
0 1 2 3 f(MHz) 5
the test voltage of a standard
lightning impulse is the
Figure 2. Influence of the frequency of the oscillations on the extreme value if the frequency
breakdown voltage for air in homogeneous field.

Page 3 of 6
 ELECTRA PAPER
7-th of February 2002

is less than 0.5 MHz, and it is means of the formula: k for the dielectrics tested can
the peak value of the mean be described by the following
curve for frequencies equal to Uref(50) - Upmc(50) equations:
or higher than 0.5 MHz. k = ---------------------- (4)
Ue(50) - Upmc(50) - k= 1 when f  f1
- k  log f 2  log f (5)
4.2 Test voltage according to The test voltage factor k has log f 2  log f1
the new research also been analysed for with-
when f1 < f < f2
stand voltages (breakdown
The conclusions from the probability about 10%) and the - k = 0 when f  f2
experimental work reported in results show a constant test
The evaluation criteria in the
this paper lead to different voltage factor k50%  k10%  k.
criteria for the evaluation of the present standard 2 for
test voltage, based on a certain The empirical test voltage impulses with oscillations or
frequency interval for the factors k have been calculated overshoot can also be expressed
superimposed oscillations as using double exponential using the test voltage factor
introduced in 8. The results functions as mean curves. which changes from 1 to 0 at a
show that the test voltage lies frequency of 0.5 MHz, as
between the extreme value of The relation between the test shown by the dotted line in Fig.
the impulse Ue and the peak voltage factor and oscillation 4.
value of the mean curve Upmc. frequency is not the same for
The actual value can be all dielectrics. The regression k
evaluated by using a test line shown in Fig. 3 is the best
voltage factor k which varies fit for all of the measured 1 a)
from 1 to 0 according to values and represent the
whether the oscillations are of average value for the different b)
lower or higher frequency. dielectrics tested.
f(MHz)
0
0,3 0,5 1,6
The 50% breakdown voltage k 1
Figure 4. Test voltage factor k
can be calculated using the test 0,8 a) Current IEC 60060-1
voltage factor k from: 0,6 b) New Proposal
0,4
Uref(50)= Upmc(50) + k . ß (2)
0,2

where ß = Ue(50) - Upmc(50) 0

10 100 f (kHz) 1000 10000
5. Implementation
and examples
Note: this formula has been shown to Figure 3. Test voltage factor k as
apply for values of ß  20% of Uref(50). function of the frequency.
The evaluation method
The reasoning applied to the discussed in this paper is
The maximum difference
50% breakdown voltage can be suitable for both manual and
between the test voltage factor
applied also to other probability software analysis. It improves
of the average curve and the
levels e.g. the withstand level: the reproducibility of test
test voltage factor for any of the
results particularly as the value
dielectrics is not higher than
Ut = Upmc +k . ß (3) of the test voltage is more
0.3, which means that the
directly related to the physical
influence on the test voltage is
where ß = Ue - Upmc stress of the dielectric. The new
relatively small (±3%) even for
proposal has been used to
overvoltages with ß up to 10%.
evaluate the TDG examples in
If greater accuracy is needed, or
4.3 Determination of the test the standard IEC 61083-2 4.
if ß is more than 10%, then the
voltage factor actual test voltage factor for the
dielectric can be used.
The test voltage factor was 5.1. Determination of the
determined from the 50% extreme value Ue
For values of f1= 0.3 MHz and
breakdown tests results by f2= 1.6 MHz, the voltage factor
The extreme value of a

Page 4 of 6
 ELECTRA PAPER

measured waveform Ue is of the extreme value. In the Ut = Ue - (1-k) . ß (6)

regarded as the 100% reference third and fourth columns of
value for time parameters Table II the peak values of the The calculated values of test
evaluation and it can be double exponential mean voltage according to the two
evaluated manually or by curves and corresponding procedures are given in
automatic algorithm. Any overvoltages are given. columns seven and eight of
superimposed noise should be Table II.
removed either manually or 5.3 Determination of the
better by using a moving frequency/overshoot 5.5 Alternative calculation of
window filter or a better FIR duration test voltage
filter [8]. The second column of
Table II shows the extreme The frequency can be Taking into account that if a
values for five full lightning determined manually as the short smooth lightning impulse
impulses of the Test Data time difference between two (T1=0.84s) is filtered by
Generators given in the consectutive oscillation peaks means of the function of the
standard IEC 61083-2 4. on the crest of the impulse test voltage factor k, its peak
waveform. Alternatively the value would change less than
5.2 Determination of the frequency can be evaluated by 0,1%, equation (3) could be
overvoltage ß applying DFT or FFT software transformed into the following:
to the difference between the
The most accurate way to recorded waveform and the Ut  k .( Upmc + ß ) (7)
determine the overvoltage ß double exponential mean curve.
requires the calculation of the Therefore, the test voltage Ut
previous mean curve using In order to determine the can be determined applying this
model-based curve fitting 10. overshoot it is necessary to find filter to the recorded impulse
The mean curve should follow the points of intersection and then the test voltage Ut is
a double exponential model between the recorded impulse given by the extreme value Ue
because the empirical test and the double exponential of the filtered curve.
voltage factor k has been mean curve, and special care
obtained from a double must be taken. The fifth column Ut = Ue(filtered curve)
exponential function. The of Table II shows the calculated
double exponential function frequency or overshoot When a filter is applied the
should allow fitting around the duration for each TDG value of the overvoltage ß is
crest, from a fixed amplitude waveform 4. unknown. Therefore the equi-
(e.g. 50%) in the front to a few valent requirement of ß 10%
microseconds after the extreme 5.4 Calculation of the test of Ue is limiting the difference
value. voltage between Ue and Ut as shown in
Fig.5.
In accordance with the Using the equations (5) the test
application of results the voltage factor k was determined
maximum permitted over- from the equation:
voltage ß should be about 10%
Table II Test voltage of the TDG cases of the IEC 61083-2
Case Ue Upmc  f k Ut Ut' (kV)
(kV) (kV) (kV) (MHz) (kV) IEC 61083-2
4/9 + 1024 + 975 + 49 0.66 0.53 + 1001 + 975
3/8 + 1050 + 993 + 57 0.37 0.88 + 1043 +1050
13 - 1072 - 957 - 115 0.39 0.84 - 1054 - 1070
14 - 1071 - 960 - 111 0.59 0.60 - 1027 - 960
Case (Uextr/Upmc' ) Upmc (1 / 2) (f1 / d2) (k1 / k2) Ut Ut' (kV)
(kV) (kV) (kV) MHz / s (kV) IEC 61083-2
11 960 / 954 950 10 / 4 3,7 / 1 0 / 0.69 953 950
Note: Case no 11 has oscillations superimposed on an overshoot; therefore the determination of two k- factors can be applied:
one for oscillations (k1) and the other for the overshoot (k2). The Upmc' is the peak value of the mean curve that removes
only the oscillations but not the overshoot.

Page 5 of 6
 ELECTRA PAPER

standards and should betaken lightning impulse tests for

Ue – Ut into account during the next high voltage equipment”,
revision. Contract nº PL-951210-
SMT-CT96-2132, 17 Sept.-
1999
0,1 Ue Acknowledgments
6 F. Garnacho, P. Simón.
The authors wish to thank Mr. "Software for the
Robert Hughes for his valuable evaluation of the lightning
f1 f2 editorial comments and writing impulse parameters". 1-st
help. In addition we would like Workshop on Insulation
Figure 5. Voltage difference between to thank the European
Ue and Ut as function of the filter
Co-ordination, Madrid,
Community for the financial November 1999
characteristic
support.
In many cases an easy way to 7 IEC 60060-2 “High-voltage
obtain good approximations is test techniques. Part 2:
to use the single exponential
References Measuring systems”, 1994
function that fits the tail of the 1 IEC 60071-1 "Insulation
impulse 11. 8 S. Berlijn, F. Garnacho,
co-ordination. Part-1:
E. Gockenbach, M. Muhr,
Definitions, principles and
K. Hackemack, M. Kvarn-
7. Conclusions rules", 1993
gren, P. Simón, P. Werle
2 IEC 60060-1 “High voltage “Influence of Different
Recent research has studied the test technique. Part 1: Lightning Impulse Shapes
physical behaviour of air, oil, General Specifications and on the breakdown
oil-paper, SF6 and XPLE Test Requirements” 1989 behaviour of insulating
dielectrics subjected to impulse materials” Proposal of
voltages having superimposed 3 F. Garnacho, M. Aro, K. Modification of IEC
oscillations or overshoot. Schon and others 60060-1, ISH 2001
“Evaluation Procedures for
The results show that no lightning impulse 9 Yi Li, J. Rungis “Precision
difference exists in the break- parameters in case of digital filter for high
down behaviour for overshoot waveform with oscillations voltage impulse
or oscillations superimposed on and/or an overshoot”, measurements systems"
a smooth lightning impulse. Trans. on IEEE Power IEEE Trans. on Power
Delivery, Vol. 12, No. 2, Delivery, Vol. 14, No. 4,
The influence of the 1997, p. 640 - 649 1999, p. 1213-1220
superimposed oscillation
depends on the frequency and 4 IEC 61083-2 "Digital 10 T.R. McComb “Calculating
the proposed frequency depen- recorders for measure- the parameters of full
dent test voltage factor k allows ments in high-voltage lightning impulse using
a reproducible evaluation of the impulse test. Part 2: model-based curve fitting”
test voltage. Evaluation of software IEEE Trans. on Power
used for the determination Delivery, Vol. 6, No. 4,
The test voltage evaluation can of the parameters of 1991, p. 1386-1394
be done automatically as well impulse waveforms”, 1991
as manually and the test voltage 11 E. Gockenbach,
represents the stress in the 5 S.M. Berlijn, F. Garnacho, K. Hackemack, P. Werle,
dielectric more closely. P. Simón, E. Gockenbach, “A contribution to the
P. Werle, K. Hackemack, evaluation of lightning
The proposed method can M. Watts, “Final report, impulse with oscillations or
improve the evaluation digital measurement of overshoot near the peak”,
procedures in the relevant IEC parameters used for ISH 2001

Page 6 of 6

View publication stats