You are on page 1of 8

21, rue d’Artois, F-75008 PARIS D1_204_2012 CIGRE 2012

http : //www.cigre.org

Study of Lightning Impulse Test Waveforms for UHV Equipment

S. OKABE, T. TSUBOI, G. UETA, Y. YAMAGATA, and E. ZAIMA


Tokyo Electric Power Company
Japan

SUMMARY

The lightning impulse (LI) voltage test waveform for electrical power equipment is specified in IEC
60060-1 “High-voltage test techniques”. IEC TC42 made revisions to the IEC 60060-1 standard in
2010, including the introduction of a method to convert an overshoot waveform to a test voltage curve
using the K-factor (test voltage) function. At present, a study on the standard for UHV-class electrical
power equipment is underway.
In this context, the present paper reports two subjects concerning the issues and technical development
related to LI withstand voltage test waveforms mainly for UHV-class equipment as follows: One is a
study on the test circuit technology investigating LI waveforms actually generated and solving the
issue of exceeding the standard value through analysis and experiment (Sections 2 and 3). The other is
a study on waveform evaluation technology (K-factor function method) for the purpose of determining
an appropriate waveform (Sections 4 and 5). The main contents are as follows:
[I. Test circuit technology to generate LI waveforms]
(1) Assuming a UHV-class transformer or GIS, the equipment capacitance to be tested is about
5,000 pF to 15,000 pF or larger. According to the actual results and calculation, in order to
obtain a waveform with an overshoot rate (β’) of 10% or less, the front time (T1) becomes 2.2 μs
or more and even considerably exceeds 3.0 μs in some cases.
(2) According to the experimental results with large-sized models, the T1 extension for less than 3.6
μs has only a minor influence on the insulation characteristics. From a converse perspective, the
required dielectric strength can be verified using a waveform with T1 extended.
[II. Test waveform evaluation technology (K-factor function method)]
(3) The base curve extraction method newly proposed can evaluate the β’ more appropriately.
(4) According to the experimental results using large-sized models, the K-factor evaluation method
itself is effective also for UHV-class equipment.
(5) In the K-factor evaluation, a smaller β’ is desirable, the realistic upper limit of which is 10%.
(6) Since the K-factor value is inclined to be relatively smaller if the insulation distance is longer,
the existing K-factor function may have to be reviewed for UHV-class equipment.
The authors would use these study results alongside of data from other countries to contribute to
establishing the IEC standards for UHV-class equipment through the IEC TC42 WG19 “Adaptation of
TC 42 standards to UHV test requirements”.

KEYWORDS

Lightning impulse withstand voltage test - Ultra high voltage - Test voltage function - IEC 60060-1
okabe.s@tepco.co.jp
1. Introduction
The lightning impulse withstand voltage (LIWV) test waveform for electrical power equipment is
specified in IEC 60060-1 “High-voltage test techniques” [1]. However, with the increase in voltage
and equipment capacity to meet the increase in electricity demand and reduce power transmission
losses, the application of the standard waveform specified in the previous standard has been hampered
because it eventually becomes an overshoot waveform or the front time (T1) is extended [2]. To deal
with such issues, IEC TC42 revised the IEC 60060-1 standard in 2010, including the introduction of a
method to convert an overshoot waveform to the test voltage curve using the K-factor (test voltage)
function and clarification of the standard values, such as the overshoot rate (β’). At present, a study on
the standard for UHV-class equipment is underway.
In the present report, assuming the LIWV test facilities for UHV-class equipment as shown in Figure 1,
the issues related to the test circuit technology used to generate LI waveforms are summarized
(Section 2). The breakdown voltage characteristics with respect to the T1 with large-sized models are
exhibited and the extendibility of the standard value
for the T1 is stated (Section 3). Also subsequently Impulse Voltage
presented in relation to the test waveform evaluation generator
Voltage divider
technology (K-factor evaluation method) is an
improved idea for the base curve extraction method
that can evaluate the β’ more appropriately (Section
4). The experimental results of the K-factor values
with large-sized models are displayed and the idea of
reviewing the K-factor function suitable for UHV- Test object
class equipment is proposed (Section 5). Based on
these proposals, the K-factor evaluation method is
applied to actual test waveforms to calculate each
parameter of the converted waveforms on a trial Figure 1. Photo of the test facility for UHV-
basis (Section 5). class GIS.

2. Actual Status of the Test of UHV-class Equipment


2.1 Actual LIWV test results for UHV-class equipment
An investigation was conducted into LIWV test waveforms now available for UHV-class equipment
[2]. The results of this investigation included not only the actual test results but also those calculated
through the circuit simulation used for the waveform adjustment for the LIWV test facilities based on
the latest design data and are plotted later in Figure 3 in Section 2.2.
According to the test results for a UHV-class transformer, if the β’ is adjusted to 10% or less, the T1
far exceeds its upper limit of 1.56 μs in the existing standard. As for GIS, a test that meets the present
standard can be conducted for single GCB; however, it is difficult to conduct such a test for GIS
assembly. Consequently, in actual fact, the LIWV test for UHV-class equipment frequently cannot be
conducted under the test conditions specified in the existing standard, and must be conducted in
accordance with the explanatory note or such for the standard for individual equipment.

2.2 Trial calculation of the testable equipment capacitance


Figure 2 illustrates a simplified equivalent circuit for the LI voltage test. In the test for actual large-
sized equipment, the capacitance of the test equipment Ct Impulse generator Rs Ls
is large, which is likely to cause a long front time T1.
Consequently, in order for T1 to be within the standard
value, T1 must be shortened by, for example, reducing the Wiring
series resistance of Rs. This eventually causes a relatively Cs Rp Ct
large residual inductance Ls of, for example, wiring with
Test
respect to the resistance, and a waveform with overshoot is equipment
generated due to the influence of this inductance.
Accordingly, T1 and β’ are in a contradictory relationship Rs: Series resistor Cs: Power source
Rp: Parallel resistor Ct: Test equipment
to each other, making it important to understand the Ls: Residual inductance
relationship between T1 and β’ to discuss whether or not a Figure 2. Image of the equivalent circuit
waveform that meets the standard is generated. of LIWV test.

1
Consequently, its relationship was investigated using an equivalent circuit in Figure 2.
Assuming the LIWV test for UHV-class equipment, the value of each element was determined as
follows: The impulse generator capacitance Cs was fixed at 25,000 pF and the Ls was set to 150 μH.
For the test equipment capacitance Ct, the parameters were varied widely, assuming various types of
apparatuses, and the parallel resistance Rp was adjusted so that the time to half value (T2) may be
about 50 μs. Under such conditions, the Rs was varied to obtain the relationship between T1 and β’.
The values T1 and T2 are those before the K-factor process.
The calculation results are added in Figure Other small
GCB GIS
3. As seen in the value of Ct, the smaller apparatus 3000 pF 5000 pF 8000 pF Transformer
its value, the smaller those of T1 and β’ as Ct=2000 pF 4000 pF 6000 pF 10000 pF 12000 pF Ls=150 μH
a matter of course. According to the 30 Transformer of JPN
comparison with the criteria (0.84 μs ≤ T1 25
Transformer of CHN

Overshoot rate, β’ (%)


GCB of JPN
≤ 1.56 μs, β’ ≤ 10%) in the present GCB of CHN
standard [1], it is found difficult to 20
GIB of JPN
generate waveforms that meet the standard 15 Assembly GIS of JPN
0.84 1.56 2.4 3.0 3.6
if Ct exceeds about 2,300 pF when Ls is (-30%) 1.2 (+30%) (+100%) (+150%) (+200%)
150 μH. Consequently, for UHV-class 10 5µs, <10%

equipment, particularly for the full GIS 5


8µs, <some %
assembly and transformers with a large
capacitance, it is actually difficult to 0
generate waveforms that meet the standard. 0.5 1 1.5 2 2.5 3 3.5 4
Front time, T1 (μs)
Under such circumstances, it will be an Figure 3. Relation between the front time T1 and the
idea to relax the criterion for T1 and β’ in overshoot rate β’ of voltage test waveforms.
the test standard for UHV-class equipment.
Here, the values of Ct that meet the standard are compared using Figure 3 on condition that the
criterion for each value is relaxed. However, this study is only from the perspective of circuit constants
apart from the equivalence of the breakdown mechanism. First, where β’ is determined to be 20% or
less, Ct that meets the standard can be up to about 3,000 pF, which differs only slightly from the
original value of 2,300 pF. Conversely, where T1 is determined to be 2.4 μs (1.2 μs + 100%) or less, Ct
that meets the standard can be significantly increased up to 6,000 pF. If the T1 is further extended by
150% or 200% to 3.0 μs or 3.6 μs respectively, Ct that meets the standard can be further increased to
10,000 pF or more, which will cover almost all apparatuses. Consequently, it is considered effective to
extend T1 to increase the kinds of apparatuses that meet the standard for UHV-class equipment.

3. Evaluation on the Influence of Front Impulse


Impulse Test equipment
Time Extension voltage
voltage
generator
generator
In order to study the possibility of the T1 Bushing

extension, as referred to in Section 2.2, a Voltage


Voltage divider
divider
breakdown experiment was conducted for Acrylics
waveforms with a long T1. Figure 4 diagrams window

experimental setup assuming an oil-immersed Test electrode

transformer. Figure 5 summarizes the change in


Circulating pump
the 50% breakdown voltage (BDV), where the Micro filter
T1 was varied from about 1.2 μs to 4.8 μs (with
the β’ of 0%) using the largest possible model
assuming UHV-class equipment. Figure 6 shows
the plots of the applied voltage and the 1000mm
150
40
breakdown time as well as the applied voltage 150
waveforms for an oil-immersed transformer. Test
Test equipment
equipment Disk electrode
1m diameter
For all the GIS, transformers, and air insulation
tested, the longer the T1, the lower the 50% Figure 4. Photo of the experimental setup for a
BDV was inclined to be. Based on this result, large-sized oil gap model, assuming a UHV-class
the insulation performance is considered to be oil-immersed transformer.

2
more severely verified if the T1 is extended for the withstand voltage test. The change in the 50% BDV
with a T1 up to 3.6 μs was roughly within 6%, or the width of the tolerance ±3% of the test voltage.
According to this result, it is considered more suitable for the verification of insulation performance to
allow the T1 extension rather than to increase the β’. However, for the oil-immersed transformer, the
50% BDV declined to 94.1% when the T1 was 4.6 μs, showing a breakdown condition different from
that under other conditions as indicated in Figure 6. Also for the air insulation, the 50% BDV declined
to 94.2% when the T1 was 5.0 μs. Therefore, it is considered necessary to set the T1 to about 3.6 μs or
less, otherwise the T1 extension may lead to excessively severe test conditions.
1800 140%
Air insulation (positive) 1.2μs 2.4μs 3.5μs 4.6μs
1.2μs 2.4μs 3.5μs 4.6μs
50% breakdown voltage (kV).

50%BDV / (50%BDV at 1.2μs) .


1500 130% 1000
Left axis 900 1.4
GIS
1200 Oil-immersed transformer 120%

Test voltage waveform.


800 1.2

Breakdown voltage (kV) .


700

(Standardized)
1
900 110% 600
GIS 0.8
500
600 100% 400 0.6
300
0.4
300 Air insulation (positive) 90% 200
Right axis 0.2
100
0 80% 0 0
0.0 1.0 2.0 3.0 4.0 5.0 6.0 0 5 10 15 20 25 30
Front time (μs) Time (µs)

Figure 5. Changes in the breakdown voltage with Figure 6. The breakdown voltage and time with
respect to the front time T1. respect to the T1 for oil-immersed transformer.

4. Base Curve Extraction Method


4.1 Issues of the existing method and a proposal of a new base curve extraction method
The present IEC standard [1] specifies a method of fitting the base curve using a double exponential
function in a series of procedures from the evaluation of the β’ to the conversion of waveforms using
the K-factor function. In this method, the part of the 20% value or less at the wavefront and the 40%
value or less at the wavetail is removed from the recorded curve to extract the core part of the recorded
curve to conduct fitting. In this paper, this method is called Method I, where a double exponential
function is used as the formula to extract the base curve for the purposes of simplicity and general
versatility, which is considered significant as a starting point of standardization [3].
However, in the case of the overshoot waveform, the fitting result sometimes deviates upward from
the centerline of the waveform at the wavefront area due to the influence of the overshoot part and
conversely downward at the latter part of the wavetail area, which leads to an error in the residual
curve separation. It results in irrational insulation property and the β’ calculation, which could
eventually degrade the reliability of the insulation performance. Consequently, another fitting method
(defined as “Method II”) was devised, allowing a more rational base curve to be extracted from the
perspective of the equivalent circuit of the LI voltage test circuit and the insulation property by means
of the reduced overshoot influence. Method II is a method of fitting for the data of which the 80%-or-
more portion is eliminated from that used for Method I.

4.2 Comparison between the existing method and a new base curve extraction method
These two methods were applied to actual overshoot waveforms as well as to those simulated using a
mathematical formula and their advantages and disadvantages were evaluated. Figure 7 represents an
example of the application to an overshoot waveform (β’ = 17.8%). In Method I, it is found that, since
the waveform including the oscillation part is fitted, the part from around the crest value to the first
half of the wavetail is fitted to be relatively high and the β’ is consequently calculated as relatively low.
Conversely, in Method II, since the influence of the oscillation part is eliminated, the β’ is calculated
to be close to its correct value.
Figure 8 summarizes the deviation rate in the β’ from the correct value, where the β’ is calculated by
applying each method to various waveforms with the β’ and the frequency of the overshoot part varied.
In Method II, the β’ calculated was relatively close to the correct value under all conditions because it

3
was not influenced by the overshoot part. Consequently, the newly devised Method II is considered
rational as a means of calculating the β’ of several to several tens of percentage points actually
evaluated for the LI test waveform.
Subsequently, the K-factor function was applied to the residual curves extracted using both methods
and the shape parameters of the test voltage curve (Ut: value of the test voltage, T1, T2) were compared.
As a result, the shape parameters differed very little regardless of the base curve extraction method,
hence the application of the new base curve extraction method was deemed to improve the evaluation
results of the β’ with little influence on the test voltage curve parameters.

1 150 kHz, Method I 150 kHz, Method II


Recorded curve 250 kHz, Method I 250 kHz, Method II
0.8 450 kHz, Method I 450 kHz, Method II
Method I β’ of each method
0.6 Method I: 14.2%
35%
Voltage (pu)

Method II Method II: 18.1%


0.4 Method I
Base curve
Recorded curve 25%

Relative error
0.2
Base curve
0 Residual curve
15%
Method I
-0.2
Residual curve Method II
-0.4 5% Method II
0 5 10 15 20
Time (μs) -5%
0 10 20
Figure 7. Result of fitting by the two methods in Overshoot rate (%)
the case of β’ = 17.8%. Figure 8. Relative error from correct overshoot
rate for waveforms with β’ of 4 to 25%.

5. K-factor Values with Large-sized Models


5.1 Experimental results on the K-factor value with GIS
Figure 9 is a photo of the test equipment for a UHV-
[SF6 gas gap testing bus]
class GIS (central conductor simulating the shielding GIS tank inner diameter: Φ340mm
part). The test conditions, including for the Test electrodes: Φ240mm × 100mm × 4
experimental data already available, are summarized in
Table 1. “UHV model test” means the present test
using the large-sized model assuming UHV-class
equipment whereas “European Project [4]” means the
test from which the existing K-factor function is
derived. “Practical apparatus model test” means the
experimental data already available, based on the
measurement of the breakdown voltage using
Figure 9. Photo of the test equipment for a
overshoot waveforms.
large-sized SF6 gas gap model assuming
As an example, the experimental results for the large
UHV-class GIS.
sized-model with the superimposed frequency: fs =
250 kHz are shown in Figure 10. The fs of the overshoot waveform with the T1 of 1.2 μs was about
250 kHz. In Figure 10, the peak values of applied voltages with respect to the corresponding
breakdown time are plotted. In addition, the voltage waveforms with no occurrence of breakdown are
standardized and displayed. According to the plots in Figure 10, the higher the β’, the higher the 50%
BDV. Where the 50% BDV of the overshoot waveform exceeds that of the waveform with the β’ of
0% like in this case, the waveform conversion using the K-factor function is to be effective.
Figure 11 summarizes the experimental results of the K-factor values and the existing K-factor
function. According to Figure 11, the K-factor values when the β’ is 10% for the large-sized model
almost coincide with the existing K-factor function, showing results equivalent to those in the
European Project and for a practical apparatus model. Meanwhile, the test results for β’s of 20% and
30% deviated from existing data. Also, in the breakdown time plots in Figure 10, where the β’ is high,
the ratio of breakdown occurrence at the wavefront is high, showing a different breakdown condition.
Based on these results, for test waveforms with the β’ of about 10%, evaluation using the existing K-
factor function is also considered useful for UHV-class GIS. However, in the K-factor evaluation for

4
waveforms with the β’ of 20% or more, it may be impossible to maintain equivalence with the
standard waveform in terms of breakdown mechanism.

Table 1. Experimental conditions for GIS data.


Project
[Present] UHV model test Practical apparatus model
European Project
(Large-sized model) (Non-standard waveform test)
Time frame 2009 – 1997 – 1999 1998 –
[1] Actual GIS (Coaxial [1] Plane-plane electrode [1] Hemisphere – plane
cylinder) (homo) electrode
Experimental conditions

Flat area: 30 mm


Gap len gth: 2.0 mm (a) Rod electrode: 20 mm
Test electrodes Gap length: 9 mm
geometry [2] Rod – plane electrode
(non-homo) (b) Rod electrode: 15 mm
Gap length: 12 mm
Rod electrode: 21 mm
Gap length: 6.35 mm

Electric field  = 0.59  = 1 (Homo)  = 0.60


utilization factor: No data (Non-homo)  = 0.45
SF6 Gas pressure 0.5 MPa-abs 0.4 MPa-abs 0.5 MPa-abs
1-Generator 2-Generator 1-Generator
Waveform
Overshoot LI Superimposed LI Non-standard LI
Waveform

Polarity & magnitude 1000 kV 100 kV 250 kV


Oscillation freq. 0.1, 0.3, 0.5 MHz 0.2, 2.0, 5.0 MHz 0.5, 0.9, 2.2 MHz
Overshoot rate 0, 10, 20, 30% 0, 5, 10, 20% 0 – 25%

0%OS 10%OS 20%OS 30%OS : Measurement


1.4 results of K-factor function, Ref. [1]
UHV model OS_10% (Present)
0%OS 10%OS 20%OS 30%OS 10% overshoot UHV model OS_20% (Present)
1400 1.4 1.2 UHV model OS_30% (Present)
Breakdown voltage (kV)

European Project
1200 1.2 1
Test voltage waveform

Practical apparatus, η=0.45


K-factor k(f)

1000 1 Practical apparatus, η=0.60


(Standardized)

0.8
800 0.8 Practical apparatus model
0.6
600 0.6
0.4 k-factor curve
400 0.4
0.2 : Results of E.P.
200 0.2
0 0 0
0 5 10 15 20 25 30 -0.2
Time (µs)
0.01 0.1 1 10 100
Figure 10. Characteristics of the breakdown
f (MHz)
voltage and time with respect to the β’ in
Figure 11. Measurement results of K-factor values
case of fs = 250 kHz.
for GIS.
5.2 Experimental results on the K-factor value with an oil-immersed transformer
Figure 12 indicates the experimental results
1.4
Measurement results of K-factor function, Ref. [1]
of the K-factor values for oil-immersed OS_10% (Present)
large-sized model OS_20% (Present)
transformer. According to Figure 12, all the 1.2
OS_30% (Present)
results in the large-sized model were slightly 1 European Project
K-factor k(f)

Practical apparatus, Oil path


lower than the K-factor function values; 0.8 Practical apparatus, Turn to turn
however, the trend of change with respect to 0.6 Practical apparatus model
fs was the same. The results were the same
0.4
as those in the European Project, which is k-factor curve
the basis of the existing K-factor function. 0.2
Based on these results, the K-factor 0 : Results of E.P.
evaluation method itself is considered useful -0.2
also for a UHV-class transformer. It will 0.01 0.1 1 10 100
also be an idea to review the K-factor f (MHz)
function to take lower values for large-sized Figure 12. Measurement results of K-factor for oil-
equipment. immersed transformers.
5
5.3 Experimental results on the K-factor value with air insulation
Figure 13 indicates the experimental results
1.2 (-2400kV negative) K-factor function, Ref. [1]
of the K-factor values in air insulation. As OS_10% (-, Present)
OS_20% (-, Present)
seen in Figure 13, the measurement results 1 OS_30% (-, Present)
OS_10% (+, Present)
(in positive polarity) for the large-sized 0.8 OS_20% (+, Present)

K-factor k(f)
model assuming actual equipment were E.P. Homogeneous
0.6 E.P. Non-homogeneous
lower than the K-factor function values. Long air gap_Homo
Long air gap_Non-homo
This is considered attributable to the 0.4
k-factor curve
relatively long breakdown time, ranging 0.2
from 10 μs to 30 μs, mainly due to the long
0
insulation distance. However, the trend of Measurement results of bushing
change with respect to fs was similar to that -0.2 (+1800kV positive)
model
of the existing K-factor function. 0.01 0.1 1 10 100
Where the evaluation using the existing K- f (MHz)
factor function is applied to UHV-class air Figure 13. Measurement results of K-factor for air
insulation, the verification test of the insulation.
insulation characteristics is inclined to be optimistic. However, for air insulation of, for example,
bushings, switching impulse overvoltage and temporary overvoltage under pollution conditions are
dominant and air insulation is not really an issue in LIWV test. Consequently, the characteristics of
SF6 and oil for internal insulation must be emphasized to determine the K-factor function.

5.4 Study of a new K-factor curve


Given the long insulation distance in UHV-class equipment, the formative time lag is also extended,
hence it is considered natural for the K-factor value to decline. With the decline of the K-factor value
for an oil-immersed transformer also taken into account, a relatively low K-factor function may
possibly be introduced especially for UHV-class equipment. Since insufficient data are considered to
be available to determine the function individually for each kind of apparatus, the assumption here is
to determine a single common function, the idea of which is exhibited in Figure 14.
The proposed function (Idea 2) indicated in Figure 14 is a curve obtained on the basis of the
experimental results of the K-factor value with large-sized models of GIS and an oil-immersed
transformer using the least mean square method with the frequency exponent fixed at 2. As described
in Section 5.3, to emphasize the internal insulation as one to be tested, the proposed function was
derived only based on the data of GIS and 1.2
an oil-immersed transformer after GIS with ’10% Idea 1 (Existing
Ideafunction):
1
eliminating the results of air insulation. 1 k=1 / (1+2.2 
Idea
f 2)2-A
SF6, β'≈10%
According to comparison with the existing Idea 2 (Proposed function):
0.8 Transformer Oil, β'≈10%
f2)β'≈10%
k=1 / (1+7.5Air,
K-factor function (Idea 1) from the with ’10%
k-factor

perspective of reliability in the insulation 0.6


test, the test conditions became severe for Air insulation with
0.4
GIS but closer to those appropriate for a ’10%, positive
transformer and air insulation equipment 0.2
from the existing optimistic ones. The
0
deviation of the K-factor values of the
0.01 0.1 1 10
experimental results from the K-factor Frequency (MHz)
function also existed when the existing K- Figure 14. Changes in the breakdown voltage with
factor function was derived. The respect to the front time.
difference between the proposed function
and the K-factor values of the experimental results is at the same level as the difference in the
European project between the existing K-factor function and the experimental results. Consequently,
the proposed curve has become more suitable overall for UHV-class equipment.

5.5 Example applications of the K-factor function including the proposed method
Figure 15 summarizes the application results of the existing K-factor function (Idea 1) and the new K-
factor function (Idea 2) studied in Section 5.4 to actual waveforms with the β’ of about 10% and where
fs = 150 kHz and 400 kHz. Here, Method II in Section 4 is used to extract the base curve. Since the K-

6
factor value in Idea 2 is slightly smaller in the case of the low frequency fs = 150 kHz, the crest value
Ut of the test voltage curve calculated is slightly Recorded curve
2500
lower in Idea 2. As the K-factor value itself is near Ue=2300 kV Base curve
Test voltage curve, Idea 1
1, Ut does not differ significantly from the crest 2000
Test voltage curve, Idea 2

value Ue of the recorded curve. Conversely, where

Voltage (kV)
400 kHz
the fs = 400 kHz, the K-factor value differs by 1500
Idea and the Ut values differ to some extent. K Ut (kV) Ut/Ue
In large-sized equipment, where the excess of the 1000 150 Idea 1 0.95 2295 0.998
150 kHz kHz Idea 2 0.86 2285 0.993
β’ and T1 from the respective standard values is an
500 400 Idea 1 0.74 2258 0.982
issue, there will potentially be few cases where Ut kHz Idea 2 0.45 2215 0.963
practically differs excessively, because fs is 0
assumed to be about 100 kHz to 200 kHz. In 0 4 8 12 16 20
practical use, the existing function might be Time (μs)
applied as-is to eliminate the confusion to use Figure 15. Evaluation results of the test voltage
different K-factor functions for the UHV-class and curve by the application of the existing or
a lower voltage class. proposed K-factor function.

6. Conclusion
In conjunction with the review IEC 60060-1 “High-voltage test techniques” for UHV-class electrical
power equipment, the issues and technical development related to LIWV test waveforms were
reported in the present paper. The main contents are as follows:
[I. Test circuit technology to generate LI waveforms]
(1) Assuming a UHV-class transformer or the full GIS assembly, the capacitance of the equipment
to be tested is about 5,000 pF to 15,000 pF or larger. According to the actual results and trial
calculation, in order to obtain a waveform with the β’ of 10% or less using existing test
equipment, the front time T1 becomes 2.2 μs or more and even considerably exceeds 3.0 μs in
some cases.
(2) Where the β’ is set to 10% or less, it becomes difficult for the T1 to meet the present standard for
large substation equipment, the T1 extension for less than 3.6 μs has only a minor influence on
the insulation characteristics. From a converse perspective, the required dielectric strength can
be verified using a waveform with T1 extended.
[II. Test waveform evaluation technology (K-factor function method)]
(3) The newly proposed base curve extraction method can evaluate the β’ more appropriately than
the existing method.
(4) According to the experimental results using large-sized models, the K-factor evaluation method
itself is effective also for UHV-class equipment.
(5) For a waveform with an excessive β’, the breakdown condition differs from that of the standard
waveform in some cases. Consequently, in the evaluation using the K-factor function, a smaller
β’ is desirable, the realistic upper limit of which is 10%.
(6) Since the K-factor value is inclined to be smaller with longer insulation distance, the existing K-
factor function may have to be reviewed for the K-factor evaluation for UHV-class equipment.
The authors would use these study results alongside of data from other countries to contribute to
establishing the IEC standards for UHV-class equipment through the IEC TC42 WG19 “Adaptation of
TC 42 standards to UHV test requirements”.

BIBLIOGRAPHY

[1] IEC 60060-1, 2010, “High-voltage test techniques Part 1: General definitions and test
requirements”.
[2] Y. Yamagata, et al., “Utility's experience on design and testing for UHV equipment in Japan”,
(Intern. Sympos. on Standards for Ultra High Voltage Transmission, Session 3, No. 3.6, 2009.)
[3] J. Hällström, et al., “Applicability of different implementations of K-factor filtering schemes for
the revision of IEC 60060-1 and -2” (14th Intern. Sympos. High Voltage, Paper No.B-32, 2005)
[4] S. Berlijn, et al., “Digital Measurement of Parameters Using for Lightning Impulse Tests for
High Voltage Equipment” (Final Report, Contract PL-951210-SMT4-CT96-2132, 1999)

You might also like