Professional Documents
Culture Documents
net/publication/341780898
CITATION READS
1 247
6 authors, including:
Some of the authors of this publication are also working on these related projects:
CIGRE WG C4.45 Technical Brochure: Measuring techniques and characteristics of fast and very fast transient overvoltages in substations and
converter stations View project
Emergent Phenomena Testbed Simulator for Improving SCADA Performance in Power System Security Management View project
All content following this page was uploaded by Božidar Filipović-Grčić on 31 May 2020.
F
Bozidar Filipovic-Grcic, Bruno Jurisic, Samir Keitoue, Ivan Murat,
OO
Dalibor Filipovic-Grcic, and Alan Zupan
PR
3 bushing tap. The focus of the paper is on the analysis of transient overvoltages caused
4 by lightning strikes recorded at the terminals of power transformer. Several recorded
5 overvoltages are analyzed and their amplitudes and frequency spectrum are presented
6 and compared with those referring to standard impulse voltages from IEC standard.
7 Collected data include number, peak and duration of recorded transient overvolt-
ED
8 ages and can be used for the assessment of the transformer insulation condition and
9 estimation of health index. Data recorded by TOMS are also of significant impor-
10 tance since the insulation system of power transformer and other equipment in the
11 substation can be damaged by lightning or switching overvoltages.
CT
14 1 Introduction
E
16 switching operations. Transformer insulation is tested with the standard lighting and
17 switching impulses in high voltage laboratory. However, in the operation various
18 non-standard waveforms stress insulation. Front and tail time of the overvoltages
19 at transformer terminals measured in operation differ from the standard ones, and
20 the waveshapes can be oscillatory contrary to the standard unidirectional double
CO
B. Filipovic-Grcic (B)
Faculty of Electrical Engineering and Computing, University of Zagreb, Zagreb, Croatia
e-mail: bozidar.filipovic-grcic@fer.hr
B. Jurisic · S. Keitoue · I. Murat · D. Filipovic-Grcic
UN
exponentials [1]. Standard lightning impulse and switching impulse test voltages
Author Proof
21
22 have been questioned as they should be based on the actual overvoltages measured
23 in service [2, 3] which can be acquired via appropriate monitoring system.
24 Many investigations have been carried out to study the electric aging of oil-paper
25 insulation, including the research on the accumulative effect of repeated lightning
OF
26 impulses in power transformers [4–11]. The work presented in [11] reveals that the
27 accumulation of repeated lightning impulses may lead to the breakdown of insula-
28 tion and thus threaten the safe operation of power system. Although many studies
29 have been focused on the influence of accumulative effect, the basic mechanism still
30 has not been clearly clarified, hence it’s necessary to study the influence of accumu-
RO
31 lative effect on property of oil impregnated paper (OIP) insulation and explore its
32 mechanisms. In [12], the research shows significant influence of repeated lightning
33 impulses on OIP samples and confirmed the existence of accumulative effect. The
34 repeated application of lightning impulses will weaken performance of OIP insula-
35 tion and could eventually lead to breakdown. During the tests, translucent gelatinous
36
37
38
39
DP
substance was detected and the colour of OIP changed on the sample surface, which
can be attributed to the variation of cellulose paper itself. With the accumulation of
repeated lightning impulses, the dielectric parameters of OIP increase significantly
in the lower frequency range, including relative permittivity, volume conductivity
40 and dielectric loss angle. The generation of polar products and translucent gelati-
41 nous substance on the sample surface and the variation of cellulose fiber itself are
TE
42 the main factors.
43 In [13], statistics of amplitudes and time parameters of the intrusive lightning
44 overvoltages into the substation have been investigated based on the measured
45 data, which were monitored at the HV bushings of the power transformers in a
110 kV air-insulated substation. Measured data indicated bidirectional oscillatory
EC
46
51 front time and tail time obtained in [13], the 50% values of the cumulative frequency
52 are 20.8 and 198 µs, respectively. In real operating conditions, lightning overvoltage
53 at transformer terminals can be an oscillatory waveform, due to multiple reflections
54 at the points where the impedance significantly changed. It is important to know the
55 frequency spectrum of lightning overvoltages at transformer terminals, since in case
CO
61
spectrum is presented and compared with those referring to the standard impulse
Author Proof
66
67 voltages from the IEC standard [15]. Collected data include number, peak and dura-
68 tion of recorded transient overvoltages and can be used as the basis for the assessment
69 of the transformer insulation condition and estimation of health index. Data recorded
70 by TOMS are also of significant importance since the insulation system of power
OF
71 transformer and other equipment in the substation can be damaged by lightning or
72 switching overvoltages.
RO
73
78
79
80
DP
matching circuit is carried out with a coaxial cable. The inbuilt acquisition card is
fast enough to capture transients containing frequency components up to 1 MHz. It
is important to note that the capacitive voltage divider used for the measurements
does not change the shape of the measured signal (the calibration range of frequen-
81 cies extends up to 500 kHz) which allows monitoring of fast front and slow front
82 overvoltages. In Croatia, the TOMS is currently installed in eight power transformer
TE
83 units and one shunt-reactor located in four 220/110 kV substations, one 400/110 kV
84 and one 400/220/110 kV substation. More details about TOMS measuring system,
85 matching circuit and triggering of overvoltage acquisition can be found in [16].
86 TOMS is installed in two substations which are managed by the Croatian trans-
mission system operator. The brief layout of the 110/220 kV substations and trans-
EC
87
102
SUBSTATION 1
Author Proof
110 kV TOMS
220 kV
busbars busbars
245 MVA
220/13.8 kV
OF
AT1 150 MVA G
110 kV 220±12x1.25%/115 kV
7 transmission lines YNa0d5
240/40 mm2 Al/St
150/25 mm2 Al/St
RO
AT2 150 MVA
220±12x1.25%/115 kV, YNa0d5 Fault 2
150 MVA
13.8/110 kV
Fault 3
G AT3
DP
150/150/50 MVA
220±12x1.25%/115/10.5 kV, YNy0d5 220 kV double circuit
transmission line
l=46 km
SUBSTATION 2
220 kV
TE
TOMS 220 kV double circuit
110 kV busbars transmission line
busbars
l=18.2 km
Fault 1
SF6 subsation
9 bays with cables AT1 150 MVA
EC
SUBSTATION 3
and transmission 220±12x1.25%/115 kV
lines YNa0d5 220 kV transmission line
l=13.7 km
220±12x1.25%/115 kV
YNa0d5
Fig. 1 Layout of 110/220 kV substations which are connected with 220 kV transmission lines
CO
104 At the end of 2008, a LLS was established as part of the LINET network, covering a
105 wide area of the Croatian territory. LINET is a modern LLS with a network of more
106 than 125 sensors covering most of Europe. LLS measures the VLF/LF frequency
UN
107 spectrum of electromagnetic waves which lightning strikes emit. The measurement
108 of magnetic flux is carried out through highly sensitive sensors which are arranged
109 across the area with spacing of around 150–250 km. Since the electromagnetic emis-
110 sion of the lightning spreads at the speed of light, it reaches the sensors at different
points in time. Although the difference is in the order of micro-seconds, the rela-
Author Proof
111
112 tively accurate calculation of the original emission location of the lightning strike
113 is possible. The data measured by every single sensor is transmitted to a central
114 server. The exact geographical position for all the lightning strikes measured is
115 calculated and stored in a database. This measurement method is also known as
OF
116 the “time-of-arrival” method. Application of LLS in power system control of Croa-
117 tian transmission system operator enables lightning activity tracking and time-spatial
118 correlation with incidences (faults, automatic re-closures, outages) registered by the
119 relay protection system [16].
RO
120 4 Transient Overvoltages Recorded on Power Transformers
121 Overvoltages in power network can be caused by lightning strikes to overhead trans-
122
123
124
125
DP
mission lines, circuit breaker switching operations and faults. Power transformers
can be exposed to such transient overvoltages during the operation. Transient over-
voltages with steep wave front have an impact on dielectric stresses of the insulation
of the first few winding turns or in the case of the resonance voltage built up locally
126 inside the winding. The number and amplitudes of overvoltages which stress the
127 insulation depend on various parameters such as the lightning strike density in the
TE
128 considered area, since it determines how often the transformer is stressed by light-
129 ning overvoltages. Since the overvoltage amplitudes at transformer terminals are
130 usually unknown, an on-line overvoltage transient recorder is used with the ability
131 to sample, analyze and store transients in real-time. Three cases of faults (Fig. 1)
caused by lightning strikes are analysed in more detail to investigate amplitude and
EC
132
138 Transients recorded by TOMS installed in substations 1 and 2 are shown in Figs. 2
139 and 3. The recorded transients were time-correlated with a lightning strike which
140 was detected by LLS. Lightning strike with current amplitude 115 kA occurred hit
141 tower of 220 kV transmission line connecting substations 2 and 3, at a distance of
142 11.2 km from substation 2 (7 km from substation 3). At the same time, SCADA
UN
143 system detected double phase to ground fault in phases A and C, following the auto-
144 reclosure operation of circuit breakers in the line bays in substations 2 and 3. Although
145 substation 2 is closer to the fault location compared to substation 1, overvoltages in
146 substation 2 are lower due to network topology and reflections of traveling waves,
Fig. 2 Transient
Author Proof
overvoltages recorded in
substation 1
OF
RO
Fig. 3 Transient
overvoltages recorded in
substation 2
DP
TE
EC
RR
147 coming simultaneously from both circuits of transmission line (induced and direct
148 overvoltages) and entering substation 1.
149 It is possible to extract the transient overvoltage waveforms from the recorder data
150 using the high-pass FIR filter. High-pass filter is used to obtain only high-frequency
151 components caused by lightning overvoltages and to remove low-frequency and
CO
Fig. 4 Lightning
Author Proof
overvoltages recorded in
substation 1 (after filtering
out low frequency
components)
OF
RO
Fig. 5 Lightning
overvoltages recorded in
substation 2 (after filtering
out low frequency
components)
DP
TE
EC
recorded in substation 1
CO
UN
of lightning overvoltages
recorded in substation 2
OF
RO
4.2 Case 2—Lightning Strike to 220 kV Transmission Line
157
158
160 Transients recorded by TOMS in substations 1 and 2 are shown in Figs. 8 and 9.
TE
161 Recorded transients were time-correlated with a lightning strike which was
162 detected by LLS. Lightning strike with current amplitude −75.3 kA occurred on
163 the 220 kV transmission line connecting substations 1 and 2 at a distance of 2.9 km
164 from the substation 1. At the same time, SCADA system detected double phase to
165 ground fault in phases A and B, following the auto-reclosure operation of circuit
EC
166 breakers in the line bays in substation 1 and 2. Transient overvoltages after filtering
167 out low frequency components and their frequency spectrum is shown in Figs. 10,
168 11, 12 and 13.
RR
Fig. 8 Transient
overvoltages recorded in
substation 1
CO
UN
Fig. 9 Transient
Author Proof
overvoltages recorded in
substation 2
OF
RO
Fig. 10 Lightning
overvoltages recorded in
substation 1 (after filtering
out low frequency
components)
DP
TE
EC
Fig. 11 Lightning
overvoltages recorded in
substation 2 (after filtering
RR
of lightning overvoltages
recorded in substation 1
OF
RO
Fig. 13 Frequency spectrum
of lightning overvoltages
recorded in substation 2 DP
TE
EC
169
172 Another interesting event recorded by TOMS was caused by multiple lightning strike
CO
173 which occurred on the 220 kV transmission line route, at a distance of 16 km from
174 the substation 1 (30 km from substation 2). Parameters of multiple lightning strikes
175 are given in Table 1. Recorded transients were time-correlated with a lightning flash
176 consisting of seven subsequent lightning strikes which were detected by LLS. Three
177 lightning strikes marked in Table 1 were selected as the ones that probably caused
UN
178 recorded transients. This was done by matching the time difference between the
179 successive lightning strikes detected by LLS with the time difference between the
180 events recorded by TOMS.
Transients recorded by TOMS which were caused by strike no. 2 (current ampli-
Author Proof
181
182 tude −80.2 kA) and corresponding frequency spectrums are shown in Figs. 14, 15,
183 16, 17, 18 and 19. At the same time, SCADA system detected line to ground fault
184 in all phases, following the auto-reclosure operation of circuit breakers in the line
185 bays in substations 1 and 2. Circuit breakers interrupted short-circuit current firstly
OF
186 in substation 2. While short-circuit current was still supplied from substation 1, two
187 successive lightning strikes 13 ms apart (no. 6 and 7 from Table 1) hit transmission
188 line, and overvoltages were recorded by TOMS at power transformer terminals in
189 substation 1 (Figs. 20 and 21). These two lightning overvoltages can be clearly seen
190 in Fig. 20, while at the end of recording (around 25 ms) switching overvoltages occur
RO
191 due to opening of circuit breaker in substation 1.
Fig. 14 Transient
overvoltages recorded in
substation 1
DP
TE
EC
Fig. 15 Transient
overvoltages recorded in
substation 2
RR
CO
UN
Fig. 16 Lightning
Author Proof
overvoltages recorded in
substation 1 (after filtering
out low frequency
components)
OF
RO
Fig. 17 Lightning
overvoltages recorded in
substation 2 (after filtering
out low frequency
components)
DP
TE
EC
recorded in substation 1
CO
UN
of lightning overvoltages
recorded in substation 2
OF
RO
Table 1 Parameters of multiple lightning strikes detected by LLS
Lightning strike
number
1
Time (h:min:s.ms)
01:02:12.257
DP
Lightning current
amplitude (kA)
15.5
Time difference
between subsequent
lightning strikes 1t (ms)
–
2 01:02:12.261 −80.2 4
TE
3 01:02:12.274 −20 13
4 01:02:12.293 −8 19
5 01:02:12.294 −30.3 1
6 01:02:12.306 −19 12
7 01:02:12.319 13
EC
−12.7
Fig. 20 Transient
overvoltages recorded in
substation 1 (continued)
RR
13 ms
CO
UN
of lightning overvoltages
recorded in substation 1
(continued)
OF
RO
5 Discussion and Future Work
192
193
194
DP
The wide variety of lightning stroke characteristics and the modifying effects of
power system components result in a diversity of intrusive lightning voltage wave-
195 shapes that stress transformers. These are not the traditional standard lightning
impulses with a waveshape of 1.2/50 µs which are used according to IEC [15]. There-
TE
196
197 fore, the applicability of the standard lightning impulse voltage to power transformer
198 testing has been questioned, and the overvoltage used in the test on transformers
199 should be as close as possible to the lightning overvoltages measured in service [13].
200 Analysis of measurement results indicated that bidirectional oscillatory over-
EC
201 voltage waveforms caused by lightning strikes appear at terminals of power trans-
202 former. Oscillatory character of recorded overvoltages is caused by multiple reflec-
203 tions of travelling waves in the substations and at the points where the system
204 impedance significantly changed. Considering the lightning strike characteristics,
205 structure of power system and observation conditions, the operation of surge arresters
RR
206 in transformer and line bays, and the winding resonance in the transformer, resulted in
207 the relatively long-time parameters of overvoltages recorded on power transformers.
208 Recorded lightning overvoltages are bidirectional oscillatory with duration of several
209 milliseconds (5–6 ms), which is quite different from standard lightning impulse
210 waveform used for testing of power transformers. Maximum recorded amplitude of
CO
211 overvoltages is 371 kV (Fig. 10) causing the operation of surge arresters installed in
212 line and transformer bay.
213 It is also important to investigate the frequency spectrum of lightning overvoltages
214 at transformer terminals, since in case when dominant frequency of overvoltage is
215 close to the natural frequency of the transformer winding, resonance overvoltages can
UN
216 occur which in some cases may cause transformer failure. FFT analysis of recorded
217 overvoltages showed that dominant frequency components are in range 1–30 kHz.
218 Frequency spectrum of measured overvoltages differs from the frequency spectrum
OF
RO
219 of standard impulse waveforms. Figure 22 shows comparison between spectral densi-
220
221
222
223
DP
ties of measured overvoltage (from case 2, Fig. 10, phase A) and standard lightning
impulse waveform 1.2/50 µs with amplitude 1050 kV. According to the method
described in [17], Frequency Domain Severity Factor (FDSF) can be determined
which is defined as the ratio between the spectral density of the measured over-
224 voltage and the spectral density of the standard lightning waveform used for testing
225 transformers. It considers the frequency content of the overvoltages measured in the
TE
226 substation and compares it to the frequency content of voltage waveforms for which
227 the transformer had been tested. The FDSF factor should be less than 1 to ensure that
228 the stresses arising from a particular event occurring in the system will be adequately
229 covered by dielectric tests performed in the HV laboratory.
Figure 23 shows calculated FSDF factor which was greater than 1 at frequency
EC
230
231 ranges 2.5–6.1 kHz and 9.3–10.4 kHz, meaning that at these frequencies the highest
232 electrical stress on the transformer insulation is expected. It also means that trans-
233 former tests performed with lightning waveforms do not cover adequately low
234 frequency stresses. Therefore, overvoltages measured in a substation can excite a
RR
235
236 30 kHz range. The FDSF approach can thus be used both for design review upon
237 incoming transients and in analysis of failures. When combined with online moni-
238 toring, it can also be used as indicator of increased transient risks for a power
239 transformer.
OF
240 Some investigations presented in [12, 18] show negative accumulative effect of
241 multiple lightning impulses on insulation properties of oil-paper insulation systems.
242 The repeated application of lightning impulses will weaken the insulation perfor-
243 mance of insulation system and can eventually lead to breakdown. Lightning
244 discharges (flashes) that transfer to ground both positive and negative charges are
RO
245 termed bipolar lightning discharges. In case 3 presented in the previous section,
246 bipolar lightning flash consisting of seven subsequent lightning strikes caused tran-
247 sient overvoltages on power transformer terminals. As can be seen from this case,
248 time difference between subsequent lightning strikes varies from 1 to 19 ms. There-
249 fore, in a relatively short time period multiple transient overvoltages of different
250
251
252
253
DP
polarity may occur on transformer terminals. Experimental investigations confirm
that degradation of transformer insulation system increases significantly as time
difference between successive transient overvoltages decrease. Therefore, it is very
important to measure transient overvoltages on transformer terminals and to record
254 such events in order to assess an overall condition of transformer insulation system
255 and to include this effect in estimation of health index. The future investigations will
TE
256 consider:
257 – Automatic grouping of overvoltage types (temporary, switching, lightning) based
258 on correlation with SCADA and LLS.
259 – Statistical analysis of amplitudes, frequency spectrum and FDSF based on a larger
EC
265 time, crest voltage and tail time of equivalent waveforms can be determined and
266 compared to standard impulse waveforms.
267 – Simulation of electromagnetic transients and comparison with measurements.
268 Such analysis can be used for example for validation of high-frequency power
269 transformer models or to study the interaction between power transformers and
CO
270 network.
271 – Development of method for assessment of the transformer insulation degradation
272 caused by transient overvoltages based on measured data from TOMS. This will
273 be used for estimation of power transformer health index.
274 – Use an existing TOMS for measurement of transient currents through station
UN
275 surge arresters caused by lightning or switching overvoltages. This will enable to
276 determine energy stress of station arresters directly from measurements.
6 Conclusions
Author Proof
277
278 In this paper, an on-line transient overvoltage monitoring system for power trans-
279 formers is used for measurement of overvoltages on the transformer bushing tap. The
280 focus of the paper is on the analysis of transient overvoltages caused by lightning
OF
281 strikes recorded at the terminals of power transformer. Three cases of faults caused by
282 lightning strikes to 220 kV double circuit transmission line are presented. Measured
283 overvoltages and faults were correlated with SCADA system and LLS data.
284 Recorded lightning overvoltages are bidirectional oscillatory with duration of
285 several milliseconds, which is quite different from standard lightning impulse wave-
RO
286 forms used for testing of power transformers. FFT analysis of recorded overvoltages
287 showed that dominant frequency components are in range 1–30 kHz. Frequency
288 spectrum of measured overvoltages differs from the frequency spectrum of standard
289 lightning impulse waveforms. Analysis of FSDF factor showed that transformer
290 tests performed with lightning waveforms do not cover adequately low frequency
291
292
293
294
DP
phenomena which are present in recorded overvoltages.
Bipolar lightning flash consisting of seven subsequent lightning strikes caused
transient overvoltages on power transformer terminals. Therefore, in a relatively
short time period multiple transient overvoltages of different polarity may occur
295 on transformer terminals. Experimental investigations confirm that degradation of
transformer insulation system increases significantly as time difference between
TE
296
301 References
RR
302 1. Darveniza M (1988) The generalized integration method for predicting impulse volt-time
303 characteristics for non-standard wave shapes-a theoretical basis. IEEE Trans Electr Insul
304 23(3):373–381
305 2. Okabe S, Takami J (2008) Evaluation of breakdown characteristics of oil-immersed trans-
306 formers under non-standard lightning impulse waveforms—method for converting non-
307 standard lightning impulse waveforms into standard lightning impulse waveforms. IEEE Trans
CO
314
315 transformer liquids under standard lightning impulse voltage. IEEE Trans Dielectr Electr Insul
316 18:285–294
317 6. Kaufhold M, Borner G, Eberhardt M, Speck J (1996) Failure mechanism of the interturn
318 insulation of low voltage electric machines fed by pulse-controlled inverters. IEEE Electr Insul
319 Mag 12(5):9–16
7. Yin W (1997) Failure mechanism of winding insulations in inverter-fed motors. IEEE Electr
Author Proof
320
OF
326 acteristics of oil-paper insulation under steep front impulse voltages. IEEE Trans Power Del
327 9:1926–1935
328 10. Balaji SP, Merin Sheema IP, Krithika G, Usa S (2011) Effect of repeated impulses on transformer
329 insulation. IEEE Trans Dielectr Electr Insul 18:2069–2073
330 11. Okabe S (2006) Voltage-time and voltage-number characteristics of insulation elements with
331 oil-filled transformers in EHV and UHV classes. IEEE Trans Dielectr Electr Insul 13:436–444
RO
332 12. Sima W, Sun P, Yang Q, Yuan T, Lu C, Yang M (2014) Study on the accumulative effect of
333 repeated lightning impulses on insulation characteristics of transformer oil impregnated paper.
334 IEEE Trans Dielectr Electr Insul 21(4)
335 13. Sima W, Lan X, Yang Q, Yuan T (2015) Statistical analysis on measured lightning overvoltage
336 surges in a 110 kV air-insulated substation. IET Sci Measur Technol 9(1):28–36. https://doi.
337 org/10.1049/iet-smt.2013.0235
338
339
340
341
342
DP
14. CIGRE Joint Working Group A2/C4.39, “Electrical transient interaction between transformers
and the power system, Part 2: Case studies”, CIGRE brochure 577B, April 2014
15. IEC 60060-1 (2010) High voltage test techniques, Part I: General definitions and test
requirements
16. Filipović-Grčić B, Franc B, Uglešić I, Pavić I, Keitoue S, Murat I, Ivanković I (2017) Monitoring
343 of transient overvoltages on the power transformers and shunt reactors—field experience in the
344 Croatian power transmission system. Procedia Eng 202:29–42
TE
345 17. CIGRE technical brochure, “Electrical transient interaction between transformers and the
346 power system, Part 1—Expertise”, Joint Working Group A2/C4.39, April 2014
347 18. Sun P, Sima W, Yang Q, Yuan T, Lan X, Lu C (2013) Accumulative effect of oil-paper insulation
348 system under multiple lightning impulse voltage. In: 2013 annual report conference on electrical
349 insulation and dielectric phenomena, Shenzhen, pp 202–205
EC
RR
CO
UN