You are on page 1of 6

Manual Evaluation of Lightning Impulses according to the new IEC 60060-1 (measurement

uncertainty and method)


S.M. Berlijn
1*
, M.Gamlin
2
, T.Steiner
3
, E. Pultrum
4
, J. Rickmann
5
, W. Lick
6

1
STRI AB, Box 707, SE-77180 Ludvika, Sweden
2
Haefely, Lehenmattstr. 353, CH-4028-Basel, Switzerland
3
Highvolt, Marie-Curie-Str. 10, D-01139 Dresden, Germany
4
KEMA, PO Box 9035, NL-6800 ET ARNHEM, Netherlands
5
Trench, Lehenmattstr. 353,4052 Basel, Switzerland
6
TU Graz, Inffeldgasse 18, A-8010 Graz, Austria
*Email: Sonja.berlijn@stri.se


Abstract: A new revision of IEC 60060-1 is under
preparation right now. At the CIGRE WG D1.33
meeting in 2005 it was discussed that in this draft both
manual and digital evaluation of lightning impulses are
accepted and described. It was commented at the WG
meeting that in case of manual evaluation of lightning
impulses the overall measurement uncertainty for the
peak value and time parameters will be larger than in
case of automatic evaluation. It was however unclear
how large the difference was. Therefore it was decided
that a round-robin should be organized to find the actual
numbers for the measurement uncertainty in manual
evaluation. The main conclusions of the work
performed were that improvements in the revision were
necessary, that the measurement uncertainty in the
lightning parameters using manual evaluation can be
much larger than when using automatic evaluation
leading to that manual evaluation can only be used as a
rough guide for the evaluation of test parameters and
proper training of the test engineer is necessary.
1 INTRODUCTION
A new revision of IEC 60060-1 is under preparation
right now. At the CIGRE WG D1.33 meeting in 2005 in
Eforie it was discussed that in the draft of the revision
of IEC 60060-1 [3] both manual and digital evaluation
of lightning impulses are accepted and described.
It was commented at the WG meeting that in case
manual evaluation of lightning impulses is accepted, a
note needs be added to the standard. This note should
inform the users of the standard that in case of manual
evaluation an overall measurement uncertainty of 3%
for the peak value is difficult, if not impossible, to
achieve. After a discussion it was decided that instead of
discussing this matter, it would be better to organize a
round-robin and to find the actual numbers for the
measurement uncertainty in manual evaluation.
2 GOAL OF THE ROUND-ROBIN
The goal of the investigation to be held in 2005/2006
was to find the measurement uncertainty in manual
evaluation of lightning impulses.
This should be achieved by sending a number of
impulses to a number of co-ordinators for evaluation.
Each co-ordinator should organize that the impulses
were evaluated by a number of persons.
The results of the evaluation were collected in early
2006 and analysed. The results are presented in this
paper.
2.1. Co-ordinators
The co-ordinators, listed in alphabetical order are:
Haefely Michael Gamlin
Highvolt Wolfgang Haushild
KEMA Edwin Pultrum
STRI Sonja Berlijn
Trench Johannes Rickmann
TU Graz- Werner Lick
The co-ordinators managed to get 14 persons to
evaluate the curves. Of these 14 persons 6 have been
performing manual evaluation before, 10 of them have
been performing many lightning impulse tests and 9 of
them had heard about the k-factor. How this influenced
the results will be discussed later in the paper.
2.2. Task description
The task to be performed by the co-ordinators was
the following:
1. Find a small number of people you think are
capable of evaluating lightning impulses using
manual evaluation. This can be persons who never
have performed manual evaluations or persons who
have done manual evaluations in the past or are
still doing it. The best would be some kind of mix.
2. Show these persons the text of the draft IEC
60060-1 and let them read the text carefully.
3. Show them the curves on paper and ask them to
evaluate these curves by using pen, ruler and paper,
using manual evaluation, according to the draft
IEC 60060-1 text. It is allowed to make copies of
the curves and even enlarge them if this is desired.
But no help of computer is allowed in anyway.
4. Fill in one result sheet for each participant. It is
possible to include the results of more participants
in one excel- file.
5. Send the results for summarizing.
3 USED EVALUATION PROCEDURE
The procedure used was according to the paragraphs
A1, A2 and A3 of the draft of the revision of
IEC 60060-1 at the time of the round robin [3]. A1 and
A3 are presented below.
3.1. A.1 Basis of the Procedures
The procedure is based on the empirical equation (1)
that describes the test voltage, U
t
, that the insulation is
subjected to, under lightning impulse voltage with an
overshoot magnitude, .

) (
) (
) (
f mp
mp e f mp t
k U
U U k U U
+ =
+ =
(1)
In (1)
U
mp
is the peak voltage of the base curve;
U
e
, is the peak voltage of the original noise
free recorded curve, and
, is the overshoot magnitude.
The test voltage factor is a frequency dependent
function given by (2) where f is the frequency in MHz.
2
) (
2 , 2 1
1
f
k
f
+
= (2)
3.2. A.3 Procedure for manual calculation from
graphical waveforms
This procedure is an implementation of (1), and it is
used for manual calculation of the impulse parameters
from waveforms in graphical formats.
1. Draw a base curve U
m(t)
manually through the
recorded curve U(t) so as to remo ve oscillations on
the front and peak.
2. Find the peak value of U
m(t)
, U
mp
.
3. Find the peak value U
e
of the recorded curve U
(t)
.
4. Calculate the overshoot amplitude, = U
e
- U
mp
.
5. Calculate the duration, t, of the overshoot by
finding the difference in time values at the two
crossing points of U
(t)
and U
m(t)
curves on both side
of the maximum peak of U
(t)
, and calculate the
overshoot frequency f
0
= 1/2t.
6. Calculate the value of test voltage factor k
(f)
from
equation (2).
7. Calculate the value of the test voltage, U
t
using
equation (1).
8. Determine the time parameters from the base curve
using as the peak voltage U
t
.
9. Report the value of the test voltage, front time,
time to half value, and overshoot magnitude.
4 USED CURVES
Six different curves were sent to the participants for
evaluation. Two pictures were sent for each curve. One
picture showing the complete curve, and one picture
showing in detail the front of the curve.
The curves are presented in Fig.1 to Fig.6. The
curves are all cases from the present IEC 61083-2 [1].
The reason that these curves were selected was that they
were used in the recent comparison for automatic
evaluation [2].

Fig. 1 Curve 1: Case 6, analytical noisy impulse.

Fig. 2 Curve 2: Case 8, analytical noise impulse with long
duration overshoot.

Fig. 3 Curve 3: Case 9, analytical noisy impulse with short
duration overshoot.

Fig. 4 Curve 4: Case 11, measured impulse.

Fig. 5 Curve 4: Case 11, front of the measured impulse.

Fig. 6 Curve 5: Case 13, measured impulse with long duration
overshoot.

Fig. 7 Curve 6: Case 14, measured impulse with short duration
overshoot.
5 RESULTS
The results are presented without any comments in
the following paragraphs. The results are commented in
chapter 6 of this paper.
In total results were obtained from 14 participants.
Results are also available from the values in
IEC 61083-2 (TDG)[1] and from the automatic
evaluation [2].
To be able to determine the test voltage value first
both the oscillation frequency and the oscillation
amplitude need to be determined. That is why these
results are presented before the test voltage value.
5.1. Oscillation frequency
The obtained values in this comparison are presented
in kHz. This parameter is not available from other
studies and was in this comparison used to be able to
understand the reasons for possible differences in test
voltage value.
Tab. 1 Reported oscillation frequency in kHz
Curve 1 Curve 2 Curve 3 Curve 4 Curve 5 Curve 6
Average 2500 403 566 2606 539 742
Minimum 0 257 337 662 250 455
Maximum 15000 517 931 4444 1250 2083
5.2. Oscillation amplitude
The obtained values in this comparison are presented
in %. This parameter is not available from other studies
and was in this comparison used to be able to
understand the reasons for possible differences in test
voltage value.
Tab. 2 Reported oscillation amplitudes in %
Curve 1 Curve 2 Curve 3 Curve 4 Curve 5 Curve 6
Average 0,09 4,27 3,92 1,63 6,94 7,85
Minimum 0,00 2,60 1,00 0,00 4,30 4,80
Maximum 1,20 8,10 6,10 9,40 10,31 10,31
5.3. Test Voltage Value
The obtained values in this comparison (manual
mean) and values from the automatic evaluation (digital
mean) are presented in kV in Tab. 3.
Tab. 3 Reported test voltage values in kV
Curve 1 Curve 2 Curve 3 Curve 4 Curve 5 Curve 6
Digital
mean
1048 1034 991 953 1042 1027
Manual
mean
1054,7
(+0,6%)
1038,8
(+0,5%)
1009,2
(+1,8%)
941,1
(-1,3%)
1032,6
(-0,9%)
1023,6
(-0,3%)
Standard
deviation
in %
2,31 2,60 2,49 4,58 3,02 2,49
Manual
minimum
990,0
(-5,5%)
985,2
(-4,7%)
968,0
(-2,3%)
820,0
(-14%)
956,2
(-8,2%)
973,2
(-5,2%)
Manual
maximum
1095
(+4,5%)
1089
(+5,3%)
1067
(+7,7%)
989
(+3,8%)
1063
(+2,0%)
1054
(+2,6%)
5.4. Front time
The obtained values in this comparison (manual
mean) and values from the automatic evaluation (digital
mean) are presented in s in Tab. 4.
Tab. 4 Reported front time values in s
Curve 1 Curve 2 Curve 3 Curve 4 Curve 5 Curve 6
Digital
mean
0,84 1,6 1,2 1,36 3,36 2,17
Manual
mean
0,91
(+8,8%)
1,55
(-2,9%)
1,18
(-1,4%)
1,26
(-7,3%)
3,15
(-6,3%)
2,22
(+2,2%)
Standard
deviation
in %
24,09 13,99 13,94 24,96 13,15 13,40
Manual
minimum
0,8
(-4,8%)
1,162
(-27%)
1,02
(-15%)
0,949
(-30%)
2,34
(-30%)
1,83
(-16%)
Manual
maximum
1,6
(+90%)
2,09
(+31%)
1,56
(+23%)
2,1
(+54%)
3,74
(+11%)
3
(+38%)
5.5. Time to half-value
The obtained values in this comparison (manual
mean) and values from the automatic evaluation (digital
mean) are presented in s in Tab. 5.
Tab. 5 Reported time to half-value in s
Curve 1 Curve 2 Curve 3 Curve 4 Curve 5 Curve 6
Digital
mean
60,2 47,4 48,3 87,6 61,4 41,9
Manual
mean
59,81
(-0,64%)
47,01
(-0,83%)
47,63
(-1,4%)
82,71
(-5,6%)
60,92
(-0,78%)
41,47
(-1,0%)
Standard
deviation
in %
9,41 3,38 4,43 9,21 4,98 5,24
Manual
minimum
44,7
(-15%)
44,67
(-5,8%)
44
(-8,9%)
64,8
(-26%)
55
(-10%)
37
(-12%)
Manual
maximum
71
(+18%)
50,758
(+7,1%)
50,75
(+5,1%)
94
(+7,3%)
66,4
(+8,1%)
46
(+9,8%)
6 ANALYSIS OF THE RESULTS
6.1. Oscillation frequency
It seems to be very difficult to determine the
oscillation frequency using manual evaluation, see
Tab. 1. There is no agreement in the obtained values
between the different participants.
For instance for curve No.5, six participants reported
oscillation frequencies in the 500 kHz range, where the
other participants reported values far from this value.
For curve No. 4, four participants reported a value of
around 2 MHz and the other participants either half this
value or twice this value.
This disagreement in oscillation frequency has direct
effect on the test voltage value, since the oscillation
frequency is used to be able to determine the k-factor.
The oscillation frequencies reported for the curves 5
and 6 from participants who have NOT been performing
manual evaluation before are in general more in
agreement with each other than from the experienced
participants. In general the participants experienced in
manual evaluation do NOT agree with each other
considering the oscillation frequency of the curves 5 and
6. For the other curves experience does not play an
important role.
None of the participants reported two oscillation
frequencies when the curve had both overshoot
(strongly damped low frequency oscillation) and
oscillations (weakly damped oscillations).
It is not clear what caused the really high difference
in reported results. Maybe it has to do with the fact that
people have difficulties calculating the oscillation
frequencies (f=1/T) using a time plot.
6.2. Oscillation amplitude
It does seem to be difficult to achieve general
agreement considering the oscillation amplitude.
In general (with the exception of curve 1 and 2)
about 50% of the participants reported values within 1%
of each other. The other participants reported values up
to half or twice this value. For the curves 1 and 2 most
participants did agree. This could arise because it was
not defined in this draft how to calculate the oscillation
amplitude in % (ratio to the extreme value of the test
voltage value).
6.3. Test voltage value
The average of all obtained values is within 1,8% of
the average value obtained by the digital evaluation.
The reported values do however have a much larger
spread than the results from a digital evaluation. The
reported values are differing from -12,8% up to + 5,7%
from the average reported value (manual evaluation).
The standard deviation in results varies from 2,3% to
4,6%. This means that the measurement uncertainty in
manual evaluation is in the range 2,5 and 3% (assuming
a normal distribution), but can in individual cases be
much larger. In case a measuring system had an overall
measurement uncertainty of 2,7% using a digital system
(measurement uncertainty 1,0%) the measurement
uncertainty would be in average 4,0% using manual
evaluation, which is higher than accepted.
The manual evaluation of the test voltage as a single
system for measuring can give measurement
uncertainties much higher than the allowed
measurement uncertainty, depending on the test
engineer and luck. Since these factors can neither be
controlled nor calibrated the method is not acceptable as
replacement for an automatic digital evaluation.
Therefore the conclusion of the obtained results is
that manual evaluation can only be used as a rough
guide for the evaluation of test parameters.
6.4. Front time
The differences in the mean values of the digital
evaluation and the manual evaluation vary from -1,4 to
8,8%.
The standard deviations in the results vary from 13%
to 25%. The reported values lie in the range of -30% up
to more than +50% of the average value.
This means that the measurement uncertainty in
manual evaluation is around 14% (assuming a normal
distribution), which is larger than the maximum allowed
10%.
Also here the same conclusion needs to be made:
Manual evaluation can only be used as a rough guide
for the evaluation of test parameters.
6.5. Time to half-value
The differences in the digital evaluation and the
manual evaluation vary from -0,6 to -5,6%.
The standard deviations in the results vary from 3,4
to 9,4%. The reported values lie in the range of -26%
and 18% of the average value. This means that the
measurement uncertainty in manual evaluation is around
7% (assuming a normal distribution).
Manual evaluation of the time to half-value can be
acceptable if the other contributions are not causing the
overall measurement uncertainty to be larger than the
maximum 10%, but unfortunately also here the
conclusion that manual evaluation can only be used as a
rough guide for the evaluation of test parameters needs
to be made
6.6. General comments
For some, yet not understood, reason one cannot say
that the complexity of the waveform plays an important
role in the obtained measurement uncertainties.
Front time
The measurement uncertainty was the largest
for the standard curve form, curve No. 6.
Test voltage value
The measurement uncertainty was the largest
for the curve with oscillations on the peak but,
the measurement uncertainty was lower for the
curve which was much more difficult to
analyse, namely the one with oscillations and
overshoot.
Time to half-value
The measurement uncertainty was the largest
for the curve with oscillations on the front and
practically no overshoot.
(Note: This is odd, one would expect the
measure ment uncertainty to be the largest for
cases with overshoot and oscillations around
the peak.)
7 CONCLUSIONS
The measurement uncertainty using manual
evaluation is as follows:
Test voltage value: 3%
Front time: 14%
Time to half-value: 7%
(Note that this means that the overall measurement
uncertainty of the measuring system is higher than the
values mentioned above)
It seems to be difficult to establish the
oscillation/overshoot frequency manually. The values in
oscillation frequency varied a lot, see Tab. 1. None of
the participants reported two oscillation frequencies in
an overshoot/oscillation with multiple frequencies
occurred. They either took the lowest oscillation
frequency or the highest. Therefore it is usually not
possible to extract reliable parameters for an overshoot
with a multiple frequency.
Apparently a better guidance in the standard is
needed on how to do establish oscillation frequency.
Anyhow, the enormous differences in oscillation
frequency did not lead to enormous differences in test
voltage value, this is maybe thanks to the k-factor or
just luck.
For the evaluation of T
1
and T
2
it is not defined
which peak value to use, is this U
e
, U
t
or U
mp
?
It is defined how to calculate in kV, but it is not
defined how to calculate in %. Should this be
calculated in relation to U
e
, U
t
or U
mp
?
It should be explained in the standard how to deal
with off-set using manual evaluation.
The most important conclusion of the work
performed is manual evaluation can only be used as a
rough guide for the evaluation of test parameters.
8 RECOMMENDATIONS GIVEN TO
WG D1.33
It is recommended to clarify the manual evaluation
procedure on some points as described in the
conclusions.
It is recommended to add at least a note in the
standard about the influence towards the measurement
uncertainty in case manual evaluation is used. It is even
better to explain to reader of the standard that manual
evaluation can only be used as a rough guide for the
evaluation of test parameters.
9 WHAT HAPPENED WITH THE
RESULTS OF THE WORK
The relevant committees have changed the text in
the draft for the revision of IEC 60060-1. For instance
the overshoot magnitude, is the difference in peak
values between the recorded curve and the base curve.
The relative overshoot magnitude, is the ratio of the
overshoot magnitude to the extreme value, usually
expressed as a percentage.
10 ACKNOWLEDGEMENTS
All participants who took the time to read the draft
for the revision of IEC 60060-1 and evaluated the
curves are thanked. WG D1.33 is thanked for the time
they took to discuss the results. Further Juris Rungis and
Jari Hllstrm are thanked for their valuable
contributions.
11 REFERENCES
[1] IEC 601083-2 Digital recorders for measurement in high-voltage
impulse tests- part 2: Evaluation of software used for the
determination of the parameters of impulse waveforms. First
edition 1996-07
[2] Applicability of different implementations of the k-factor
filtering schemes for the revision of IEC 60060-1 and 2, Jari
Hllstrm et al, paper ISH 2005
[3] IEC 60060-1Draft for revision 2005

You might also like