- Email: [email protected]
Lightning Performance of Transmission Line with and without Surge Arresters: Comparison between a Monte Carlo method and field experience
Electric Power Systems Research 149 (2017) 169–177
Contents lists available at ScienceDirect
Electric Power Systems Research journal homepage: www.elsevier.com/locate/epsr
Lightning Performance of Transmission Line with and without Surge Arresters: Comparison between a Monte Carlo method and field experience Sandro de Castro Assis a,b,∗ , Wallace do Couto Boaventura a , José Osvaldo Saldanha Paulino a , Rubens Leopoldo Markiewicz b a b
Department of Electrical Engineering, Federal University of Minas Gerais—UFMG, Belo Horizonte, MG, Brazil Energetic Company of Minas Gerais—CEMIG, Belo Horizonte, MG, Brazil
a r t i c l e
i n f o
Article history: Received 4 December 2016 Received in revised form 18 March 2017 Accepted 12 April 2017 Keywords: Lightning performance Transmission lines Field experience ATP Monte Carlo method Zinc oxide surge arrester
a b s t r a c t This paper presents a study regarding the lightning flashover rates of transmission lines comparing results from calculation procedures and field experience. The proposed procedure is based on Monte Carlo method and allows the estimation of the lightning performance of lines equipped or not with surge arresters. The procedure has been implemented in Delphi and linked to Alternative Transients Program—ATP. The calculation is done using simple engineering models for each component. Nevertheless, the overall modeling led to reliable results, concerning the comparison with the reported field data. With this respect, the comparative study of lightning performance of transmission lines, the paper presents results for three lines: a line without surge arrester, in Tennessee—USA, for which the performance was estimated using both IEEE FLASH and the proposed procedure, and two transmission lines equipped with surge arresters, in Minas Gerais—Brazil, calculated solely using the proposed methodology. These three lines have been monitored for more than 7 years, without significant layout/design changes in this period, favoring the statistical relevance of the reported data. © 2017 Elsevier B.V. All rights reserved.
1. Introduction The lightning performance of transmission lines is very important for electric utilities around the world, since lightning is a major cause of outages of overhead lines [1]. In transmission systems up to 138 kV, in Brazil, lightning is especially damaging, even in regions with both average lightning density and soil resistivity, since most tripouts are caused by backflashover [2–4]. The available methodologies and input data for calculating the lightning performance of transmission lines share some common ground. These data usually consider the characterization of return stroke current (peak, front time, time to half value and waveform), lightning striking point on the line, as well as the proper modeling of the electrical system and its response to surge propagation. The lightning performance of transmission lines is widely discussed in literature, with papers presenting either results for specific lines or numerical comparisons of lightning performance, which are cal-
culated using different methodologies or including variation on the component modeling [5–11]. Despite this comprehensive discussion, there are few studies presenting real transmission line performance data, facilitating the validation of proposed methodologies. In fact, there are not many papers comparing calculated values with results obtained from field experience. This paper aims to introduce and detail a procedure for calculating the lightning flashover rate of transmission lines that accounts for the use of surge arresters. The calculation results are compared with lightning performances observed in the field for two transmission lines with a nominal voltage of 138 kV, both owned by CEMIG, and a transmission line of 161 kV owned by TVA. Section 2 presents the adopted modeling for each component. The backflashover rate calculation procedure is detailed in Section 3, while the study cases are shown in Section 4. Finally, the conclusions are stated in Section 5. 2. Component modeling for lightning overvoltage calculations
∗ Corresponding author. E-mail addresses: [email protected], [email protected] (S. de Castro Assis), [email protected] (W. do Couto Boaventura), [email protected] (J.O.S. Paulino), [email protected] (R.L. Markiewicz). http://dx.doi.org/10.1016/j.epsr.2017.04.012 0378-7796/© 2017 Elsevier B.V. All rights reserved.
Several documents have been published regarding modeling guidelines for power systems components aimed to lightning overvoltage simulations [2–4,6,12–14]. The goal of following sub-
170
S. de Castro Assis et al. / Electric Power Systems Research 149 (2017) 169–177
Fig. 1. Tower-footing arrangement consisting of four counterpoise-wires (radius 2.59 mm) of length L, from 20 m to 90 m, buried 0.5 m deep in soil.
sections is to provide a summary of the available models and general guidelines for modeling such components, while indicating the modeling adopted in this work. 2.1. Overhead transmission line Lightning overvoltages are fast transients, whose magnitudes are influenced by traveling waves on the adjacent spans. Then, the line should be modeled to consider 3 or 4 spans at each side of the point of impact. Each span is properly represented by a multi-phase untransposed distributed parameter line section. The electrical parameters of these line sections may be modeled considering either frequency-dependent or constant parameters, calculated at 500 kHz [14]. Constant parameter modeling is chosen for this work. Corona effect is not simulated. To avoid reflections that would affect the simulated overvoltages around the point of impact, the line terminations at each side is represented by means of a long-enough line section, adopted in this work, or by inserting an impedance matrix at each termination, matching the line surge impedance.
IC = k ˛L + ˇ ,
(2)
where L is the electrode length in meters, k is a correction factor that depends on the electrode arrangement, and it is equal to 1 for the horizontal electrode and ␣ and  are coefficients dependent on the soil resistivity and lightning current waveform. Considering the parameters for the first stroke, ␣ and  are calculated by Eqs. (3) and (4) [25]:
2.2. Transmission line towers
˛1st = 0.002 + exp −1.5500.162 ,
The transmission lines towers may be modeled as a vertical transmission lines with distributed parameters, characterized by a surge impedance associated with an electromagnetic wave travel time. There are several studies seeking to characterize the models for towers [3,15–21]. The value for the tower surge impedance is calculated according to Eq. (1) [21] and the electromagnetic wave velocity of propagation in towers is considered to be equal to 80% of the velocity of light (240 m/s). The following equation, derived considering the tower as an equivalent cylinder, gives the surge impedance value:
√ h Z = 60 ln 2 2 −1 , r
ning currents. In these cases, the tower-footing electrodes generally consist of the counterpoise arrangement indicated in Fig. 1. Considering the high frequency involved in lightning analysis, the behavior of a grounding system may be described by its impulse impedance ZP , which is given by the ratio between the peak values of the grounding potential rise (GPR) and the injected current (ZP = VP /IP ). As an alternative, that is adopted in this work, the ground system is modeled as a lumped resistance with value equal to impulse impedance, but calculated in a different way. The impulse impedance value is obtained by multiplying the low frequency resistance (RLF ) by the impulse coefficient (IC ). The impulse coefficient is calculated using Eq. (2), as described in Ref. [25]:
(1)
where h is the tower height in meters, r is the tower base radius in meters and Z is the surge impedance in . 2.3. Tower grounding impedance The tower grounding impedance is very important for determining the occurrence of backflashover [22–24]. For transmission lines with rated voltage up to 230 kV, backflashover is the main cause of the outages. Tower grounding is generally modeled by a lumped resistance whose value is equal to the one obtained either from low frequency measurements (upon which a correcting factor may be used) or calculations. In large grounding systems, as those used in high voltage transmission lines installed in high resistivity soil, which may be composed by long counterpoises, soil ionization may be disregarded, as it occurs only when injecting very high light-
ˇ1st = −0.5 + exp −0.0004600.83 ,
(3) (4)
where 0 is the soil resistivity at 100 Hz. Thus, the impulse grounding impedance ZP is a function of the soil resistivity, the current waveform, primarily its front-time, and counterpoises length. Therefore, a long counterpoise cable may present a low resistance value for industrial frequency currents, but a high value for lightning currents.
2.4. Surge arrester model Only gapless surge arresters are considered and a typical nonlinear resistance (V–I characteristic curve for ZnO arrester) has been adopted. Additionally, a lumped inductance of 1 H/m representing the earth arrester lead was also included.
2.5. Strength of insulation The insulation strength depends on the waveform of the applied voltage. Considering lightning, a flashover across the insulator string may be evaluated using the following approaches: • Voltage–time curves • Integration methods • Physicals models
S. de Castro Assis et al. / Electric Power Systems Research 149 (2017) 169–177
Details of these models can be found in Refs. [15,21]. In this work, it has been adopted the volt–time curve approach. Thus, the insulator string voltage strength is determined by Eq. (5) [4,15]: UImpulse Flashover =
400 +
710 t 0.75
171
Table 1 Statistical parameters of first negative stroke current—Morro do Cachimbo Station [27,28].
Approximation by lognormal Distribution
w,
(5)
where UImpulseFlashover is the flashover strength in kV, w is the insulator string length in meters and t is the time to flashover in s, (0.5 s–16 s).
Parameter
Geometric mean
log
Ip (kA) Tan-G (kA/s)
45 19.4
0.45 0.29
2.6. Return stroke waveform The first return stroke current, assumed of negative polarity, is represented by a triangular waveform with the time to half value for all strokes being 65 s. Alternatively, it could be used a double-exponential waveform, Heidler model or CIGRE waveform. Although approximate, the triangular waveform is easy to implement and allows obtaining more representative results for overvoltages across insulator strings than, for example, obtained using a double exponential waveform. This is because the doubleexponential waveform has a maximum di/dt occurring near time zero, unlike the actual lightning current waveform, which has a maximum di/dt occurring near the first peak. Thus, the triangular waveform enables a sufficiently severe effort, although not being able to reproduce, like the double exponential wave, the wave front concave. The lightning discharges are not of equal severity. The statistical variation for the peak and front time of the lightning strokes has been assumed to follow a log-normal distribution. For the study cases in this work, used two different distribution for first stroke parameters are used, according to the location where transmission lines are installed. Thus, for the transmission line Sequoyah–Concord 161 kV, owned by TVA, described in Ref. [26], the lightning parameters used are based on the measurements performed by Berger in Monte San Salvatore, as recommended by Refs. [3,4]. Both Monte San Salvatore and TL Sequoyah–Concord are located in North hemisphere, presenting similar latitudes. The cumulative probability of exceeding stroke current I is given, approximately, by Eq. (6), while the probability for the stroke maximum di/dt is given, approximately, by Eq. (7) [15]: PI =
1
1+
I 2.6 ,
(6)
where PI is probability of exceeding stroke current I and I is stroke current in kA. PdI =
1
1+
dI
,
6 1 dt 19.4
(9)
where PdI is the probability for a specified value of di/dt to be exceeded and di/dt is the maximum rise time in kA/s. The curves for the probability of exceeding stroke current proposed in this work and those obtained from both Morro do Cachimbo Station and Monte San Salvatore data are shown in Fig. 2. Similar curves for the probability of exceeding di/dt are shown in Fig. 3. 3. Backflashover rate calculation procedure
31
where PI is probability of exceeding stroke current I and I is stroke current in kA, PdI =
Fig. 2. Probability of exceeding stroke current from Morro do Cachimbo Station and Monte San Salvatore. Full data measurements (median and standard deviation) versus simple Eqs. (6) and (8).
1
1+
dI
,
1 4 dt 24
The backflashover rate calculation is carried out using a probabilistic approach. In this respect, a probabilistic analysis for lighting performance of transmission line should include, at least:
(7)
where PdI is the probability for a specified value of di/dt to be exceeded and di/dt is the maximum rise time in kA/s. Conversely, for the transmission lines owned by CEMIG, the lightning parameters from Morro do Cachimbo Station measurements are more representative. Both Morro do Cachimbo Station and the CEMIG lines are located in the State of Minas Gerais, Brazil. The parameters for first stroke and maximum front steepness (TanG) are summarized in Table 1 [27,28]. The log normal data from Morro do Cachimbo Station measurements may be approximated, quite accurately, through simple equations. Eq. (8) is proposed for calculating the cumulative probability of exceeding stroke current I, while (9) is proposed for the probability of maximum di/dt. PI =
1
1+
I 3.9 , 45
(8)
Fig. 3. Probability of exceeding di/dt of the stroke front current from Morro do Cachimbo Station and Monte San Salvatore. Full data measurements (median and standard deviation) versus simple Eqs. (7) and (9).
172
S. de Castro Assis et al. / Electric Power Systems Research 149 (2017) 169–177
Fig. 4. General diagram block for the proposed methodology.
• Distribution of crest current magnitude; • Distribution of steepness of the current (di/dt); • Lightning striking point (at tower or midspan). Monte Carlo method was used for generating the data for the lightning stroke parameters (peak current and steepness) and location of the lightning striking point (tower or midspan). Following the criterion implemented in IEEE Flash program [4,29], the voltages obtained by strokes in towers correspond to 60% of the lightning flash collection of an overhead line (Ns ) while 40% is due to strokes in the spans. Upon these definitions, transient calculations are performed using the Alternative Transients Program—ATP, ® which is called from a routine developed in Delphi . The values for the current peak and steepness do not have a direct correlation, since the available data is not robust enough to establish this dependency. Therefore, each draw for either crest or steepness depends on its own probability of occurrence. Flashover occurrence for each analyzed transmission line is determined by comparing the resulting impulse voltage waveform with the withstand voltage for the insulator string, determined by the corresponding voltage-time curve, as in (5). The backflashover failure rate is calculated considering (i) the lightning flash collection rate, considering the overhead line location and (ii) a transmission line characteristic section, taking into account the geometry of the typical tower, the distribution of the grounding impedances and average span length.
In the proposed methodology, the lightning flash collection rate of an overhead line is determined using the Electrogeometric Model (EGM) and the concept of striking distance, besides the ground flash density along the transmission lines route. Nowadays, IEEE recommends the following striking distance equation [4]: S = 10I 0.65 ,
(10)
where I is the stroke current in kA and S is the strike distance in meters. Fig. 4 shows the proposed methodology diagram block. In the diagram: Ncases is number simulations to be carried out for each line section, Ni is the number of current simulation, NTL is the number of line sections considered, NTLi is the line section number for the current simulation, NBFO is the total number of backflashovers events resulting from the simulations for each line section, BFR is the backflashover rate for NTLi , BFO is a backflashover event and Prob trip is the NBFO /Ncases calculated ratio. 4. Study cases—analysis for real cases The results obtained using the procedure depicted in Section 3 are compared with lightning performances observed in the field for
S. de Castro Assis et al. / Electric Power Systems Research 149 (2017) 169–177
173
Table 2 Main data form 161 kV Sequoyah–Concord OHTL. Sequoyah–Concord OHTL Nominal voltage Length Average span Conductor Ground wire Insulator string (number of standard insulator) Dry arc distance Typical tower Ground flash density—Ng (flashes/km2 /year)
161 kV 29.6 km 319 m 2034.5 ACSR 3 No. 6 Al 11 1.606 m See Fig. 5 5.99
Table 3 Lightning performance calculated and observed in field.
Backflashover (flashover/100 km year)
Zp = 0.8 RBF Field experience
IEEE Flash
Proposed procedure
5.34 2.31
2.58
two transmission lines with nominal voltage of 138 kV, owned by CEMIG, a major power utility in Brazil, and a transmission line of 161 kV, owned by TVA. The data for the TVA line (transmission line design and observed lightning performance) are reported in Ref. [26]. Additionally, the calculated result using the proposed methodology is compared to the result calculated using IEEE Flash [29], for the case of the TVA line, since it is not equipped with surge arrester. 4.1. Transmission line without surge arresters The lightning performance of 161 kV Sequoyah–Concord transmission line was presented by Whitehead in Ref. [26]. Table 2 shows the main OHTL data. This OHTL is located in an area with a keraunic level (Td ) of 55 thunderstorm days per year. The ground flash density (Ng) for this area is 5.99 flashes/km2 /year, estimated as recommended by Ref. [4]. According to Ref. [26], the observed lightning performance for this line is 2.31 flashover/100 km year. Fig. 5 shows the typical tower for Sequoyah–Concord OHTL, while Fig. 6 shows its distribution for tower footing resistance, as showed in [26]. From an average resistivity of 600–700 m, reported for the region where the line is installed, and assuming grounding system through the steel grillages, an impulse coefficient of 0.8 was estimated, in accordance with Eq. (2) [25]. Thus, the calculations were performed considering the grounding impedances equal to 80% of the values for ground resistances at low frequency. The analysis considered 4 sections for the transmission line modeling. For each section, 6000 cases were simulated considering incidence of lightning in the tower and in the mid-span. For each lightning event, values for peak and steepness of the current were randomly draw. As mentioned before, the peak and steepness values for this line were chosen considering the probability curves from Monte San Salvatore measurements, in accordance with Eqs. (6) and (7). Table 3 shows the results for lightning performance calculated using the proposed procedure and the program IEEE Flash for the Sequoyah–Concord OHTL. 4.2. Transmission line with surge arresters This subsection presents original lightning performance data for CEMIG’s transmission lines. These lines are equipped with zinc oxide surge arresters (ZnO) on some towers. Table 4 presents the main data for CEMIG’s transmission lines. Using available data from the Lightning Location System (LLS) for the state of Minas Gerais, Brazil, the ground flash density along
Fig. 5. Typical tower—161 kV Sequoyah–Concord OHTL.
Fig. 6. Distribution of footing resistance for 161 kV Sequoyah–Concord OHTL.
the route of each OHTL was determined. The Table 5 shows the Ng variation for each 10% along the OHTL length. Considering the soil apparent resistivity for each structure site, shown in Fig. 7, and the geometry design for each grounding system, a specific value for impulsive coefficient was calculated for each tower. As shown in Fig. 8, the impulsive coefficient obtained for these towers are, in some cases, higher than the unit. This occurs for grounding systems with grounding electrodes built with long length counterpoises and moderate soil resistivity. Fig. 9 shows the distribution of grounding impedance along the TL Itutinga–Três Corac¸ões 2–138 kV and TL Itutinga–Minduri–138 kV. The typical towers for 138 kV Itutinga–Três Corac¸ões 2 and 138 kV Itutinga–Minduri transmission lines are shown in Fig. 10.
174
S. de Castro Assis et al. / Electric Power Systems Research 149 (2017) 169–177
Table 4 Main data form 138 kV CEMIG’s OHTL.
Nominal voltage Length Average span Conductor Ground wire Insulator string (number of standard units) Dry arc distance Typical tower Ground flash density—Ng (flashes/km2 /year)
TL Itutinga–Três Corac¸ões 2—138 kV
TL Itutinga– Minduri— 138 kV
138 kV 86.9 km 452 m Linnet Petrel and 5/16 HS 9
138 kV 46.9 km 455 m Linnet Petrel and 5/16 HS 9
1.314 m Fig. 10(a) Table 5
1.314 m Fig. 10(b) Table 5
Table 5 Ground flash density along the route of CEMIG’s transmission lines. (%) TL from substation Itutinga
Ground flash density (flashes/km2 /year)
TL Itutinga–Três Corac¸ões 2—138 kV
TL Itutinga– Minduri—138 kV
0 10 20 30 40 50 60 70 80 90 100
2.2902 3.6201 3.2534 2.5751 1.6136 2.6901 1.8127 2.6914 1.9549 2.0971 1.6440
2.2902 3.3649 5.0345 3.0842 3.1416 3.2556 5.1253 2.3510 3.0316 3.3724 4.3087
Average
2.3857
3.4873 Fig. 8. Variation of impulsive coefficients of grounding systems along the overhead transmission lines. (a) TL Itutinga–Três Corac¸ões 2—138 kV; (b) TL Itutinga–Minduri—138 kV.
Fig. 7. Apparent resistivity along the overhead transmission lines. Fig. 9. Grounding impulse impedance along the overhead transmission lines.
The 138 kV Itutinga–Três Corac¸ões TL comprises 48 towers with surge arresters in all three phases and 143 towers without surge arresters. The 138 kV Itutinga–Minduri TL has 10 towers without surge arresters, 47 towers with surge arresters in one phase, 45 towers with surge arresters in two phases and one tower with surge arresters in all phases. The Fig. 11 shows the distribution of the arresters along the two transmission lines, and Fig. 12 shows the V–I characteristic curve for the arresters. Table 6 lists the observed outage caused by lightning for a period of 9 years. It is important to remind the statistical nature in phenomena involving lightning and, consequently, the importance of the average observed over the time, as any calculation procedure aims at the average value.
Table 6 Number of outages of both CEMIG’s transmission lines. Year
138 kV Itutinga–Três Corac¸ões 2 TL Outages
138 kV Itutinga– Minduri TL Outages
2007 2008 2009 2010 2011 2012 2013 2014 2015
1 2 3 1 2 3 2 0 3
1 2 4 4 4 5 0 0 6
Average
1.89
2.89
S. de Castro Assis et al. / Electric Power Systems Research 149 (2017) 169–177
175
Fig. 10. Typical towers for TLs: (a) 138 kV Itutinga–Três Corac¸ões 2; (b) 138 kV Itutinga–Minduri.
Fig. 12. V–I characteristic curve of surge arresters for CEMIG’s transmission lines 138 kV.
Fig. 11. Distribution of arrester along the transmission lines: (a) 138 kV Itutinga–Três Corac¸ões 2 TL; (b) 138 kV Itutinga–Minduri TL.
For 138 kV Itutinga–Três Corac¸ões TL modeling, it was considered 8 sections and for 138 kV Itutinga–Minduri TL 17 sections were considered. These lines sections considered the average span, the grounding impedance and typical towers of each transmission lines with average height. A total of 6000 cases were simulated for each section, considering incidence of lightning at the tower and at the mid-span. For each lightning event, values for peak and steepness of current were randomly draw. Differently from the TVA line, the peak and steepness values for these lines were chosen considering the probabilities curves from Morro do Cachimbo Station measurements, in accordance with Eqs. (8) and (9). These distributions are used since both the lines and the Morro do Cachimbo Station are located in the State of Minas Gerais, Brazil. For each simulation run, the flashover occurrence is evaluated from voltage-time curve
176
S. de Castro Assis et al. / Electric Power Systems Research 149 (2017) 169–177
Table 7 Number of lightning outages of both CEMIG’s transmission lines. 138 kV Itutinga–Três Corac¸ões 2 TL
138 kV Itutinga–Minduri TL
Average flashover rate (lightning outages from field data)
1.89
2.89
Calculated flashover rate (lightning outages)
2.07
in calculating overvoltage and uses a stochastic approach, is effective in evaluating the lightning performance of transmission lines. Additionally, the original field data and results presented in this paper may be of value for other researchers seeking evaluation and validation for models and calculation methods. Acknowledgments
3.20
comparison. The several simulated cases accounted for the variation of Ng, grounding impedance and arrester location, along the lines. Table 7 summarizes the obtained results. 4.3. Result analysis The results presented in Tables 3 and 7 demonstrate the effectiveness of the proposed procedure in evaluating the lightning performance of transmission lines, equipped or not with surge arresters. The largest difference between calculated and field date is less than 12%. As shown in Table 3, the lightning performance observed in field for 161 kV TVA’s transmission line showed better agreement with the value calculated using the developed procedure than that obtained with the IEEE Flash program. This difference may be attributed to approximations adopted in IEEE Flash, while the proposed methodology uses a more refined approach. In this respect, IEEE Flash adopts a fixed front-time of 2 s for the lightning current, evaluates the flashover occurrence considering only two points of the insulator string volt-time curve (2 and 6 s) and uses a flat lightning incidence probability for the whole line. 5. Conclusions The work proposes a methodology for calculating lightning performance of transmission lines, which is based on Monte Carlo method for drawing lightning current parameters and striking point and uses Electrogeometric Model for evaluating the lightning flash collection. Concerning the lightning current parameters, distributions based on Monte San Salvatore and Morro do Cachimbo measurements were used. The transient voltages due to backflashover is calculated using ATP, where transmission line sections are modeled using simple engineering models. The adopted modeling that takes into account the impulsive behavior of the tower grounding system, through impulsive ground impedance concept, the volt-time curve for insulator string when evaluating flashover occurrence and the presence of surge arresters, modeled using V–I characteristic curve. Additionally, the proposed procedures use specific values for each tower, regarding the lightning incidence probability and impulsive impedance. Newly available field data of lightning performance for two CEMIG’s 138 kV transmission lines, for an observation period of 9 years, are presented and discussed. These data, along with reported data for a 161 kV TVA line, are used for validation purposes. The results obtained with the proposed procedure showed good agreement when compared to real performance data observed in the field for both TVA’s and CEMIG’s transmission lines. In the case of CEMIG’s transmission lines, the developed procedure accounted for the surge arresters installed in several sections of these transmission lines. For all study cases in this work, the major discrepancy between the calculated and observed in field lightning average outages was less than 12%. Therefore, the methodology presented in this work, which uses simple engineering models for the various components involved
The authors gratefully acknowledge CEMIG, CNPq, FAPEMIG, CAPES and PPGEE/UFMG. References [1] Report, A. C, Report of Joint AIEE-EEI Subject Committee on Line Outages [includes discussion], Trans. Ame. Inst. Electr. Eng. Part III: Power Appar. Syst. 71 (1) (1952) 1. [2] A simplified method for estimating lightning performance of transmission lines, IEEE Trans. Power Appar. Syst. PAS-104 (4) (1985) 918–932. [3] CIGRE, W., 33.01, Guide to Procedures for Estimating the Lightning Performance of Transmission Lines, CIGRE Brochure 1991, 63. [4] IEEE guide for improving the lightning performance of transmission lines, IEEE Std. 1243–1997 (1997) 1–44. [5] L.A. Martinez, F. Castro-Aranda, Modeling overhead transmission lines for line arrester studies, 10–10 June 2004, in: Power Engineering Society General Meeting, 2004. IEEE, vol 1, 2004, pp. 1125–1130. [6] A. Ametani, T. Kawamura, A method of a lightning surge analysis recommended in Japan using EMTP, IEEE Trans. Power Deliv. 20 (2) (2005) 867–875. [7] J.A. Martinez, F. Castro-Aranda, Lightning performance analysis of overhead transmission lines using the EMTP, IEEE Trans. Power Deliv. 20 (3) (2005) 2200–2210. [8] J.A. Martinez, F. Castro-Aranda, Lightning flashover rate of an overhead transmission line protected by surge arresters, 24–28 June 2007, in: Power Engineering Society General Meeting, 2007. IEEE, 2007, 2007, pp. 1–6. [9] F.M. Gatta, A. Geri, S. Lauria, M. Maccioni, A. Santarpia, An ATP-EMTP Monte Carlo procedure for backflashover rate evaluation: a comparison with the CIGRE method, Electr. Power Syst. Res. 113 (2014) 134–140. [10] P. Sarajcev, Monte Carlo method for estimating backflashover rates on high voltage transmission lines, Electr. Power Syst. Res. 119 (2015) 247–257. [11] P. Sarajcev, J. Vasilj, D. Jakus, Method for estimating backflashover rates on HV transmission lines based on EMTP-ATP and curve of limiting parameters, Int. J. Electr. Power Energy Syst. 78 (2016) 127–137. [12] CIGRE, W., 33.02, Guidelines for representation of network elements when calculating transients. CIGRE Technical Brochure 1990, 39. [13] Modeling guidelines for fast front transients, IEEE Trans. Power Deliv. 11 (1) (1996) 493–506. [14] CIGRE, W., C4. 301, Use of surge arresters for lightning protection of transmission lines. CIGRE Technical Brochure 2010, 440. [15] EPRI, A, Transmission Line Reference Book—345 kV and Above, Electric Power Research Institute, 1982. [16] M. Ishii, T. Kawamura, T. Kouno, E. Ohsaki, K. Shiokawa, K. Murotani, T. Higuchi, Multistory transmission tower model for lightning surge analysis, IEEE Trans. Power Deliv. 6 (3) (1991) 1327–1335. [17] A. Ametani, Y. Kasai, J. Sawada, A. Mochizuki, T. Yamada, Frequency-dependent impedance of vertical conductors and a multiconductor tower model, IEE Proc. Gen. Transm. Distrib. 141 (4) (1994) 339–345. [18] T. Hara, O. Yamamoto, Modelling of a transmission tower for lightning-surge analysis, IEE Proc. Gen. Transm. Distrib. 143 (3) (1996) 283–289. [19] J.A. Gutierrez R, P. Moreno, J.L. Naredo, J.L. Bermudez, M. Paolone, C.A. Nucci, F. Rachidi, Nonuniform transmission tower model for lightning transient studies, IEEE Trans. Power Deliv. 19 (2) (2004) 490–496. [20] A.D. Conti, S. Visacro, A. Soares, M.A.O. Schroeder, Revision, extension, and validation of Jordan’s formula to calculate the surge impedance of vertical conductors, IEEE Trans. Electromagn. Compat. 48 (3) (2006) 530–536. [21] J.A. Martinez-Velasco, Power system transients, in: Parameter Determination, CRC, Boca Raton, FL, USA, 2010. [22] R. Shariatinasab, J. Gholinezhad, K. Sheshyekani, M.R. Alemi, The effect of wide band modeling of tower-footing grounding system on the lightning performance of transmission lines: a probabilistic evaluation, Electr. Power Syst. Res. 141 (2016) 1–10. [23] J. Wu, J. He, B. Zhang, R. Zeng, Influence of grounding impedance model on lightning protection analysis of transmission system, Electr. Power Syst. Res. 139 (2016) 133–138. [24] A.B. Lima, J.O.S. Paulino, W.C. Boaventura, I.J.S. Lopes, M.F. Guimarães, W.A. Chisholm, B. Jamali, F. Bologna, Transient grounding impedance and transient resistivity measurements using a very short current lead, Electr. Power Syst. Res. 118 (2015) 69–75. [25] R. Alipio, S. Visacro, Impulse efficiency of grounding electrodes: effect of frequency-dependent soil parameters, IEEE Trans. Power Deliv. 29 (2) (2014) 716–723.
S. de Castro Assis et al. / Electric Power Systems Research 149 (2017) 169–177 [26] J.T. Whitehead, Lightning performance of TVA’S 500-kV 161-kV transmission lines, IEEE Trans. Power Appar. Syst. PAS-102 (3) (1983) 752–768. [27] S. Visacro, A. Soares, M.A.O. Schroeder, L.C.L. Cherchiglia, V.J. de Sousa, Statistical analysis of lightning current parameters: measurements at Morro do Cachimbo Station, J. Geophys. Res.: Atmos. 109 (D1) (2004).
177
[28] S. Visacro, C.R. Mesquita, A. De Conti, F.H. Silveira, Updated statistics of lightning currents measured at Morro do Cachimbo Station, Atmos. Res. 117 (2012) 55–63. [29] Pes, I. IEEE PES Lightning Performance of Overhead Lines Working Group: 15.09.08. http://ewh.ieee.org/soc/pes/lpdl/ (Accessed 11 November 2016).
Contents lists available at ScienceDirect
Electric Power Systems Research journal homepage: www.elsevier.com/locate/epsr
Lightning Performance of Transmission Line with and without Surge Arresters: Comparison between a Monte Carlo method and field experience Sandro de Castro Assis a,b,∗ , Wallace do Couto Boaventura a , José Osvaldo Saldanha Paulino a , Rubens Leopoldo Markiewicz b a b
Department of Electrical Engineering, Federal University of Minas Gerais—UFMG, Belo Horizonte, MG, Brazil Energetic Company of Minas Gerais—CEMIG, Belo Horizonte, MG, Brazil
a r t i c l e
i n f o
Article history: Received 4 December 2016 Received in revised form 18 March 2017 Accepted 12 April 2017 Keywords: Lightning performance Transmission lines Field experience ATP Monte Carlo method Zinc oxide surge arrester
a b s t r a c t This paper presents a study regarding the lightning flashover rates of transmission lines comparing results from calculation procedures and field experience. The proposed procedure is based on Monte Carlo method and allows the estimation of the lightning performance of lines equipped or not with surge arresters. The procedure has been implemented in Delphi and linked to Alternative Transients Program—ATP. The calculation is done using simple engineering models for each component. Nevertheless, the overall modeling led to reliable results, concerning the comparison with the reported field data. With this respect, the comparative study of lightning performance of transmission lines, the paper presents results for three lines: a line without surge arrester, in Tennessee—USA, for which the performance was estimated using both IEEE FLASH and the proposed procedure, and two transmission lines equipped with surge arresters, in Minas Gerais—Brazil, calculated solely using the proposed methodology. These three lines have been monitored for more than 7 years, without significant layout/design changes in this period, favoring the statistical relevance of the reported data. © 2017 Elsevier B.V. All rights reserved.
1. Introduction The lightning performance of transmission lines is very important for electric utilities around the world, since lightning is a major cause of outages of overhead lines [1]. In transmission systems up to 138 kV, in Brazil, lightning is especially damaging, even in regions with both average lightning density and soil resistivity, since most tripouts are caused by backflashover [2–4]. The available methodologies and input data for calculating the lightning performance of transmission lines share some common ground. These data usually consider the characterization of return stroke current (peak, front time, time to half value and waveform), lightning striking point on the line, as well as the proper modeling of the electrical system and its response to surge propagation. The lightning performance of transmission lines is widely discussed in literature, with papers presenting either results for specific lines or numerical comparisons of lightning performance, which are cal-
culated using different methodologies or including variation on the component modeling [5–11]. Despite this comprehensive discussion, there are few studies presenting real transmission line performance data, facilitating the validation of proposed methodologies. In fact, there are not many papers comparing calculated values with results obtained from field experience. This paper aims to introduce and detail a procedure for calculating the lightning flashover rate of transmission lines that accounts for the use of surge arresters. The calculation results are compared with lightning performances observed in the field for two transmission lines with a nominal voltage of 138 kV, both owned by CEMIG, and a transmission line of 161 kV owned by TVA. Section 2 presents the adopted modeling for each component. The backflashover rate calculation procedure is detailed in Section 3, while the study cases are shown in Section 4. Finally, the conclusions are stated in Section 5. 2. Component modeling for lightning overvoltage calculations
∗ Corresponding author. E-mail addresses: [email protected], [email protected] (S. de Castro Assis), [email protected] (W. do Couto Boaventura), [email protected] (J.O.S. Paulino), [email protected] (R.L. Markiewicz). http://dx.doi.org/10.1016/j.epsr.2017.04.012 0378-7796/© 2017 Elsevier B.V. All rights reserved.
Several documents have been published regarding modeling guidelines for power systems components aimed to lightning overvoltage simulations [2–4,6,12–14]. The goal of following sub-
170
S. de Castro Assis et al. / Electric Power Systems Research 149 (2017) 169–177
Fig. 1. Tower-footing arrangement consisting of four counterpoise-wires (radius 2.59 mm) of length L, from 20 m to 90 m, buried 0.5 m deep in soil.
sections is to provide a summary of the available models and general guidelines for modeling such components, while indicating the modeling adopted in this work. 2.1. Overhead transmission line Lightning overvoltages are fast transients, whose magnitudes are influenced by traveling waves on the adjacent spans. Then, the line should be modeled to consider 3 or 4 spans at each side of the point of impact. Each span is properly represented by a multi-phase untransposed distributed parameter line section. The electrical parameters of these line sections may be modeled considering either frequency-dependent or constant parameters, calculated at 500 kHz [14]. Constant parameter modeling is chosen for this work. Corona effect is not simulated. To avoid reflections that would affect the simulated overvoltages around the point of impact, the line terminations at each side is represented by means of a long-enough line section, adopted in this work, or by inserting an impedance matrix at each termination, matching the line surge impedance.
IC = k ˛L + ˇ ,
(2)
where L is the electrode length in meters, k is a correction factor that depends on the electrode arrangement, and it is equal to 1 for the horizontal electrode and ␣ and  are coefficients dependent on the soil resistivity and lightning current waveform. Considering the parameters for the first stroke, ␣ and  are calculated by Eqs. (3) and (4) [25]:
2.2. Transmission line towers
˛1st = 0.002 + exp −1.5500.162 ,
The transmission lines towers may be modeled as a vertical transmission lines with distributed parameters, characterized by a surge impedance associated with an electromagnetic wave travel time. There are several studies seeking to characterize the models for towers [3,15–21]. The value for the tower surge impedance is calculated according to Eq. (1) [21] and the electromagnetic wave velocity of propagation in towers is considered to be equal to 80% of the velocity of light (240 m/s). The following equation, derived considering the tower as an equivalent cylinder, gives the surge impedance value:
√ h Z = 60 ln 2 2 −1 , r
ning currents. In these cases, the tower-footing electrodes generally consist of the counterpoise arrangement indicated in Fig. 1. Considering the high frequency involved in lightning analysis, the behavior of a grounding system may be described by its impulse impedance ZP , which is given by the ratio between the peak values of the grounding potential rise (GPR) and the injected current (ZP = VP /IP ). As an alternative, that is adopted in this work, the ground system is modeled as a lumped resistance with value equal to impulse impedance, but calculated in a different way. The impulse impedance value is obtained by multiplying the low frequency resistance (RLF ) by the impulse coefficient (IC ). The impulse coefficient is calculated using Eq. (2), as described in Ref. [25]:
(1)
where h is the tower height in meters, r is the tower base radius in meters and Z is the surge impedance in . 2.3. Tower grounding impedance The tower grounding impedance is very important for determining the occurrence of backflashover [22–24]. For transmission lines with rated voltage up to 230 kV, backflashover is the main cause of the outages. Tower grounding is generally modeled by a lumped resistance whose value is equal to the one obtained either from low frequency measurements (upon which a correcting factor may be used) or calculations. In large grounding systems, as those used in high voltage transmission lines installed in high resistivity soil, which may be composed by long counterpoises, soil ionization may be disregarded, as it occurs only when injecting very high light-
ˇ1st = −0.5 + exp −0.0004600.83 ,
(3) (4)
where 0 is the soil resistivity at 100 Hz. Thus, the impulse grounding impedance ZP is a function of the soil resistivity, the current waveform, primarily its front-time, and counterpoises length. Therefore, a long counterpoise cable may present a low resistance value for industrial frequency currents, but a high value for lightning currents.
2.4. Surge arrester model Only gapless surge arresters are considered and a typical nonlinear resistance (V–I characteristic curve for ZnO arrester) has been adopted. Additionally, a lumped inductance of 1 H/m representing the earth arrester lead was also included.
2.5. Strength of insulation The insulation strength depends on the waveform of the applied voltage. Considering lightning, a flashover across the insulator string may be evaluated using the following approaches: • Voltage–time curves • Integration methods • Physicals models
S. de Castro Assis et al. / Electric Power Systems Research 149 (2017) 169–177
Details of these models can be found in Refs. [15,21]. In this work, it has been adopted the volt–time curve approach. Thus, the insulator string voltage strength is determined by Eq. (5) [4,15]: UImpulse Flashover =
400 +
710 t 0.75
171
Table 1 Statistical parameters of first negative stroke current—Morro do Cachimbo Station [27,28].
Approximation by lognormal Distribution
w,
(5)
where UImpulseFlashover is the flashover strength in kV, w is the insulator string length in meters and t is the time to flashover in s, (0.5 s–16 s).
Parameter
Geometric mean
log
Ip (kA) Tan-G (kA/s)
45 19.4
0.45 0.29
2.6. Return stroke waveform The first return stroke current, assumed of negative polarity, is represented by a triangular waveform with the time to half value for all strokes being 65 s. Alternatively, it could be used a double-exponential waveform, Heidler model or CIGRE waveform. Although approximate, the triangular waveform is easy to implement and allows obtaining more representative results for overvoltages across insulator strings than, for example, obtained using a double exponential waveform. This is because the doubleexponential waveform has a maximum di/dt occurring near time zero, unlike the actual lightning current waveform, which has a maximum di/dt occurring near the first peak. Thus, the triangular waveform enables a sufficiently severe effort, although not being able to reproduce, like the double exponential wave, the wave front concave. The lightning discharges are not of equal severity. The statistical variation for the peak and front time of the lightning strokes has been assumed to follow a log-normal distribution. For the study cases in this work, used two different distribution for first stroke parameters are used, according to the location where transmission lines are installed. Thus, for the transmission line Sequoyah–Concord 161 kV, owned by TVA, described in Ref. [26], the lightning parameters used are based on the measurements performed by Berger in Monte San Salvatore, as recommended by Refs. [3,4]. Both Monte San Salvatore and TL Sequoyah–Concord are located in North hemisphere, presenting similar latitudes. The cumulative probability of exceeding stroke current I is given, approximately, by Eq. (6), while the probability for the stroke maximum di/dt is given, approximately, by Eq. (7) [15]: PI =
1
1+
I 2.6 ,
(6)
where PI is probability of exceeding stroke current I and I is stroke current in kA. PdI =
1
1+
dI
,
6 1 dt 19.4
(9)
where PdI is the probability for a specified value of di/dt to be exceeded and di/dt is the maximum rise time in kA/s. The curves for the probability of exceeding stroke current proposed in this work and those obtained from both Morro do Cachimbo Station and Monte San Salvatore data are shown in Fig. 2. Similar curves for the probability of exceeding di/dt are shown in Fig. 3. 3. Backflashover rate calculation procedure
31
where PI is probability of exceeding stroke current I and I is stroke current in kA, PdI =
Fig. 2. Probability of exceeding stroke current from Morro do Cachimbo Station and Monte San Salvatore. Full data measurements (median and standard deviation) versus simple Eqs. (6) and (8).
1
1+
dI
,
1 4 dt 24
The backflashover rate calculation is carried out using a probabilistic approach. In this respect, a probabilistic analysis for lighting performance of transmission line should include, at least:
(7)
where PdI is the probability for a specified value of di/dt to be exceeded and di/dt is the maximum rise time in kA/s. Conversely, for the transmission lines owned by CEMIG, the lightning parameters from Morro do Cachimbo Station measurements are more representative. Both Morro do Cachimbo Station and the CEMIG lines are located in the State of Minas Gerais, Brazil. The parameters for first stroke and maximum front steepness (TanG) are summarized in Table 1 [27,28]. The log normal data from Morro do Cachimbo Station measurements may be approximated, quite accurately, through simple equations. Eq. (8) is proposed for calculating the cumulative probability of exceeding stroke current I, while (9) is proposed for the probability of maximum di/dt. PI =
1
1+
I 3.9 , 45
(8)
Fig. 3. Probability of exceeding di/dt of the stroke front current from Morro do Cachimbo Station and Monte San Salvatore. Full data measurements (median and standard deviation) versus simple Eqs. (7) and (9).
172
S. de Castro Assis et al. / Electric Power Systems Research 149 (2017) 169–177
Fig. 4. General diagram block for the proposed methodology.
• Distribution of crest current magnitude; • Distribution of steepness of the current (di/dt); • Lightning striking point (at tower or midspan). Monte Carlo method was used for generating the data for the lightning stroke parameters (peak current and steepness) and location of the lightning striking point (tower or midspan). Following the criterion implemented in IEEE Flash program [4,29], the voltages obtained by strokes in towers correspond to 60% of the lightning flash collection of an overhead line (Ns ) while 40% is due to strokes in the spans. Upon these definitions, transient calculations are performed using the Alternative Transients Program—ATP, ® which is called from a routine developed in Delphi . The values for the current peak and steepness do not have a direct correlation, since the available data is not robust enough to establish this dependency. Therefore, each draw for either crest or steepness depends on its own probability of occurrence. Flashover occurrence for each analyzed transmission line is determined by comparing the resulting impulse voltage waveform with the withstand voltage for the insulator string, determined by the corresponding voltage-time curve, as in (5). The backflashover failure rate is calculated considering (i) the lightning flash collection rate, considering the overhead line location and (ii) a transmission line characteristic section, taking into account the geometry of the typical tower, the distribution of the grounding impedances and average span length.
In the proposed methodology, the lightning flash collection rate of an overhead line is determined using the Electrogeometric Model (EGM) and the concept of striking distance, besides the ground flash density along the transmission lines route. Nowadays, IEEE recommends the following striking distance equation [4]: S = 10I 0.65 ,
(10)
where I is the stroke current in kA and S is the strike distance in meters. Fig. 4 shows the proposed methodology diagram block. In the diagram: Ncases is number simulations to be carried out for each line section, Ni is the number of current simulation, NTL is the number of line sections considered, NTLi is the line section number for the current simulation, NBFO is the total number of backflashovers events resulting from the simulations for each line section, BFR is the backflashover rate for NTLi , BFO is a backflashover event and Prob trip is the NBFO /Ncases calculated ratio. 4. Study cases—analysis for real cases The results obtained using the procedure depicted in Section 3 are compared with lightning performances observed in the field for
S. de Castro Assis et al. / Electric Power Systems Research 149 (2017) 169–177
173
Table 2 Main data form 161 kV Sequoyah–Concord OHTL. Sequoyah–Concord OHTL Nominal voltage Length Average span Conductor Ground wire Insulator string (number of standard insulator) Dry arc distance Typical tower Ground flash density—Ng (flashes/km2 /year)
161 kV 29.6 km 319 m 2034.5 ACSR 3 No. 6 Al 11 1.606 m See Fig. 5 5.99
Table 3 Lightning performance calculated and observed in field.
Backflashover (flashover/100 km year)
Zp = 0.8 RBF Field experience
IEEE Flash
Proposed procedure
5.34 2.31
2.58
two transmission lines with nominal voltage of 138 kV, owned by CEMIG, a major power utility in Brazil, and a transmission line of 161 kV, owned by TVA. The data for the TVA line (transmission line design and observed lightning performance) are reported in Ref. [26]. Additionally, the calculated result using the proposed methodology is compared to the result calculated using IEEE Flash [29], for the case of the TVA line, since it is not equipped with surge arrester. 4.1. Transmission line without surge arresters The lightning performance of 161 kV Sequoyah–Concord transmission line was presented by Whitehead in Ref. [26]. Table 2 shows the main OHTL data. This OHTL is located in an area with a keraunic level (Td ) of 55 thunderstorm days per year. The ground flash density (Ng) for this area is 5.99 flashes/km2 /year, estimated as recommended by Ref. [4]. According to Ref. [26], the observed lightning performance for this line is 2.31 flashover/100 km year. Fig. 5 shows the typical tower for Sequoyah–Concord OHTL, while Fig. 6 shows its distribution for tower footing resistance, as showed in [26]. From an average resistivity of 600–700 m, reported for the region where the line is installed, and assuming grounding system through the steel grillages, an impulse coefficient of 0.8 was estimated, in accordance with Eq. (2) [25]. Thus, the calculations were performed considering the grounding impedances equal to 80% of the values for ground resistances at low frequency. The analysis considered 4 sections for the transmission line modeling. For each section, 6000 cases were simulated considering incidence of lightning in the tower and in the mid-span. For each lightning event, values for peak and steepness of the current were randomly draw. As mentioned before, the peak and steepness values for this line were chosen considering the probability curves from Monte San Salvatore measurements, in accordance with Eqs. (6) and (7). Table 3 shows the results for lightning performance calculated using the proposed procedure and the program IEEE Flash for the Sequoyah–Concord OHTL. 4.2. Transmission line with surge arresters This subsection presents original lightning performance data for CEMIG’s transmission lines. These lines are equipped with zinc oxide surge arresters (ZnO) on some towers. Table 4 presents the main data for CEMIG’s transmission lines. Using available data from the Lightning Location System (LLS) for the state of Minas Gerais, Brazil, the ground flash density along
Fig. 5. Typical tower—161 kV Sequoyah–Concord OHTL.
Fig. 6. Distribution of footing resistance for 161 kV Sequoyah–Concord OHTL.
the route of each OHTL was determined. The Table 5 shows the Ng variation for each 10% along the OHTL length. Considering the soil apparent resistivity for each structure site, shown in Fig. 7, and the geometry design for each grounding system, a specific value for impulsive coefficient was calculated for each tower. As shown in Fig. 8, the impulsive coefficient obtained for these towers are, in some cases, higher than the unit. This occurs for grounding systems with grounding electrodes built with long length counterpoises and moderate soil resistivity. Fig. 9 shows the distribution of grounding impedance along the TL Itutinga–Três Corac¸ões 2–138 kV and TL Itutinga–Minduri–138 kV. The typical towers for 138 kV Itutinga–Três Corac¸ões 2 and 138 kV Itutinga–Minduri transmission lines are shown in Fig. 10.
174
S. de Castro Assis et al. / Electric Power Systems Research 149 (2017) 169–177
Table 4 Main data form 138 kV CEMIG’s OHTL.
Nominal voltage Length Average span Conductor Ground wire Insulator string (number of standard units) Dry arc distance Typical tower Ground flash density—Ng (flashes/km2 /year)
TL Itutinga–Três Corac¸ões 2—138 kV
TL Itutinga– Minduri— 138 kV
138 kV 86.9 km 452 m Linnet Petrel and 5/16 HS 9
138 kV 46.9 km 455 m Linnet Petrel and 5/16 HS 9
1.314 m Fig. 10(a) Table 5
1.314 m Fig. 10(b) Table 5
Table 5 Ground flash density along the route of CEMIG’s transmission lines. (%) TL from substation Itutinga
Ground flash density (flashes/km2 /year)
TL Itutinga–Três Corac¸ões 2—138 kV
TL Itutinga– Minduri—138 kV
0 10 20 30 40 50 60 70 80 90 100
2.2902 3.6201 3.2534 2.5751 1.6136 2.6901 1.8127 2.6914 1.9549 2.0971 1.6440
2.2902 3.3649 5.0345 3.0842 3.1416 3.2556 5.1253 2.3510 3.0316 3.3724 4.3087
Average
2.3857
3.4873 Fig. 8. Variation of impulsive coefficients of grounding systems along the overhead transmission lines. (a) TL Itutinga–Três Corac¸ões 2—138 kV; (b) TL Itutinga–Minduri—138 kV.
Fig. 7. Apparent resistivity along the overhead transmission lines. Fig. 9. Grounding impulse impedance along the overhead transmission lines.
The 138 kV Itutinga–Três Corac¸ões TL comprises 48 towers with surge arresters in all three phases and 143 towers without surge arresters. The 138 kV Itutinga–Minduri TL has 10 towers without surge arresters, 47 towers with surge arresters in one phase, 45 towers with surge arresters in two phases and one tower with surge arresters in all phases. The Fig. 11 shows the distribution of the arresters along the two transmission lines, and Fig. 12 shows the V–I characteristic curve for the arresters. Table 6 lists the observed outage caused by lightning for a period of 9 years. It is important to remind the statistical nature in phenomena involving lightning and, consequently, the importance of the average observed over the time, as any calculation procedure aims at the average value.
Table 6 Number of outages of both CEMIG’s transmission lines. Year
138 kV Itutinga–Três Corac¸ões 2 TL Outages
138 kV Itutinga– Minduri TL Outages
2007 2008 2009 2010 2011 2012 2013 2014 2015
1 2 3 1 2 3 2 0 3
1 2 4 4 4 5 0 0 6
Average
1.89
2.89
S. de Castro Assis et al. / Electric Power Systems Research 149 (2017) 169–177
175
Fig. 10. Typical towers for TLs: (a) 138 kV Itutinga–Três Corac¸ões 2; (b) 138 kV Itutinga–Minduri.
Fig. 12. V–I characteristic curve of surge arresters for CEMIG’s transmission lines 138 kV.
Fig. 11. Distribution of arrester along the transmission lines: (a) 138 kV Itutinga–Três Corac¸ões 2 TL; (b) 138 kV Itutinga–Minduri TL.
For 138 kV Itutinga–Três Corac¸ões TL modeling, it was considered 8 sections and for 138 kV Itutinga–Minduri TL 17 sections were considered. These lines sections considered the average span, the grounding impedance and typical towers of each transmission lines with average height. A total of 6000 cases were simulated for each section, considering incidence of lightning at the tower and at the mid-span. For each lightning event, values for peak and steepness of current were randomly draw. Differently from the TVA line, the peak and steepness values for these lines were chosen considering the probabilities curves from Morro do Cachimbo Station measurements, in accordance with Eqs. (8) and (9). These distributions are used since both the lines and the Morro do Cachimbo Station are located in the State of Minas Gerais, Brazil. For each simulation run, the flashover occurrence is evaluated from voltage-time curve
176
S. de Castro Assis et al. / Electric Power Systems Research 149 (2017) 169–177
Table 7 Number of lightning outages of both CEMIG’s transmission lines. 138 kV Itutinga–Três Corac¸ões 2 TL
138 kV Itutinga–Minduri TL
Average flashover rate (lightning outages from field data)
1.89
2.89
Calculated flashover rate (lightning outages)
2.07
in calculating overvoltage and uses a stochastic approach, is effective in evaluating the lightning performance of transmission lines. Additionally, the original field data and results presented in this paper may be of value for other researchers seeking evaluation and validation for models and calculation methods. Acknowledgments
3.20
comparison. The several simulated cases accounted for the variation of Ng, grounding impedance and arrester location, along the lines. Table 7 summarizes the obtained results. 4.3. Result analysis The results presented in Tables 3 and 7 demonstrate the effectiveness of the proposed procedure in evaluating the lightning performance of transmission lines, equipped or not with surge arresters. The largest difference between calculated and field date is less than 12%. As shown in Table 3, the lightning performance observed in field for 161 kV TVA’s transmission line showed better agreement with the value calculated using the developed procedure than that obtained with the IEEE Flash program. This difference may be attributed to approximations adopted in IEEE Flash, while the proposed methodology uses a more refined approach. In this respect, IEEE Flash adopts a fixed front-time of 2 s for the lightning current, evaluates the flashover occurrence considering only two points of the insulator string volt-time curve (2 and 6 s) and uses a flat lightning incidence probability for the whole line. 5. Conclusions The work proposes a methodology for calculating lightning performance of transmission lines, which is based on Monte Carlo method for drawing lightning current parameters and striking point and uses Electrogeometric Model for evaluating the lightning flash collection. Concerning the lightning current parameters, distributions based on Monte San Salvatore and Morro do Cachimbo measurements were used. The transient voltages due to backflashover is calculated using ATP, where transmission line sections are modeled using simple engineering models. The adopted modeling that takes into account the impulsive behavior of the tower grounding system, through impulsive ground impedance concept, the volt-time curve for insulator string when evaluating flashover occurrence and the presence of surge arresters, modeled using V–I characteristic curve. Additionally, the proposed procedures use specific values for each tower, regarding the lightning incidence probability and impulsive impedance. Newly available field data of lightning performance for two CEMIG’s 138 kV transmission lines, for an observation period of 9 years, are presented and discussed. These data, along with reported data for a 161 kV TVA line, are used for validation purposes. The results obtained with the proposed procedure showed good agreement when compared to real performance data observed in the field for both TVA’s and CEMIG’s transmission lines. In the case of CEMIG’s transmission lines, the developed procedure accounted for the surge arresters installed in several sections of these transmission lines. For all study cases in this work, the major discrepancy between the calculated and observed in field lightning average outages was less than 12%. Therefore, the methodology presented in this work, which uses simple engineering models for the various components involved
The authors gratefully acknowledge CEMIG, CNPq, FAPEMIG, CAPES and PPGEE/UFMG. References [1] Report, A. C, Report of Joint AIEE-EEI Subject Committee on Line Outages [includes discussion], Trans. Ame. Inst. Electr. Eng. Part III: Power Appar. Syst. 71 (1) (1952) 1. [2] A simplified method for estimating lightning performance of transmission lines, IEEE Trans. Power Appar. Syst. PAS-104 (4) (1985) 918–932. [3] CIGRE, W., 33.01, Guide to Procedures for Estimating the Lightning Performance of Transmission Lines, CIGRE Brochure 1991, 63. [4] IEEE guide for improving the lightning performance of transmission lines, IEEE Std. 1243–1997 (1997) 1–44. [5] L.A. Martinez, F. Castro-Aranda, Modeling overhead transmission lines for line arrester studies, 10–10 June 2004, in: Power Engineering Society General Meeting, 2004. IEEE, vol 1, 2004, pp. 1125–1130. [6] A. Ametani, T. Kawamura, A method of a lightning surge analysis recommended in Japan using EMTP, IEEE Trans. Power Deliv. 20 (2) (2005) 867–875. [7] J.A. Martinez, F. Castro-Aranda, Lightning performance analysis of overhead transmission lines using the EMTP, IEEE Trans. Power Deliv. 20 (3) (2005) 2200–2210. [8] J.A. Martinez, F. Castro-Aranda, Lightning flashover rate of an overhead transmission line protected by surge arresters, 24–28 June 2007, in: Power Engineering Society General Meeting, 2007. IEEE, 2007, 2007, pp. 1–6. [9] F.M. Gatta, A. Geri, S. Lauria, M. Maccioni, A. Santarpia, An ATP-EMTP Monte Carlo procedure for backflashover rate evaluation: a comparison with the CIGRE method, Electr. Power Syst. Res. 113 (2014) 134–140. [10] P. Sarajcev, Monte Carlo method for estimating backflashover rates on high voltage transmission lines, Electr. Power Syst. Res. 119 (2015) 247–257. [11] P. Sarajcev, J. Vasilj, D. Jakus, Method for estimating backflashover rates on HV transmission lines based on EMTP-ATP and curve of limiting parameters, Int. J. Electr. Power Energy Syst. 78 (2016) 127–137. [12] CIGRE, W., 33.02, Guidelines for representation of network elements when calculating transients. CIGRE Technical Brochure 1990, 39. [13] Modeling guidelines for fast front transients, IEEE Trans. Power Deliv. 11 (1) (1996) 493–506. [14] CIGRE, W., C4. 301, Use of surge arresters for lightning protection of transmission lines. CIGRE Technical Brochure 2010, 440. [15] EPRI, A, Transmission Line Reference Book—345 kV and Above, Electric Power Research Institute, 1982. [16] M. Ishii, T. Kawamura, T. Kouno, E. Ohsaki, K. Shiokawa, K. Murotani, T. Higuchi, Multistory transmission tower model for lightning surge analysis, IEEE Trans. Power Deliv. 6 (3) (1991) 1327–1335. [17] A. Ametani, Y. Kasai, J. Sawada, A. Mochizuki, T. Yamada, Frequency-dependent impedance of vertical conductors and a multiconductor tower model, IEE Proc. Gen. Transm. Distrib. 141 (4) (1994) 339–345. [18] T. Hara, O. Yamamoto, Modelling of a transmission tower for lightning-surge analysis, IEE Proc. Gen. Transm. Distrib. 143 (3) (1996) 283–289. [19] J.A. Gutierrez R, P. Moreno, J.L. Naredo, J.L. Bermudez, M. Paolone, C.A. Nucci, F. Rachidi, Nonuniform transmission tower model for lightning transient studies, IEEE Trans. Power Deliv. 19 (2) (2004) 490–496. [20] A.D. Conti, S. Visacro, A. Soares, M.A.O. Schroeder, Revision, extension, and validation of Jordan’s formula to calculate the surge impedance of vertical conductors, IEEE Trans. Electromagn. Compat. 48 (3) (2006) 530–536. [21] J.A. Martinez-Velasco, Power system transients, in: Parameter Determination, CRC, Boca Raton, FL, USA, 2010. [22] R. Shariatinasab, J. Gholinezhad, K. Sheshyekani, M.R. Alemi, The effect of wide band modeling of tower-footing grounding system on the lightning performance of transmission lines: a probabilistic evaluation, Electr. Power Syst. Res. 141 (2016) 1–10. [23] J. Wu, J. He, B. Zhang, R. Zeng, Influence of grounding impedance model on lightning protection analysis of transmission system, Electr. Power Syst. Res. 139 (2016) 133–138. [24] A.B. Lima, J.O.S. Paulino, W.C. Boaventura, I.J.S. Lopes, M.F. Guimarães, W.A. Chisholm, B. Jamali, F. Bologna, Transient grounding impedance and transient resistivity measurements using a very short current lead, Electr. Power Syst. Res. 118 (2015) 69–75. [25] R. Alipio, S. Visacro, Impulse efficiency of grounding electrodes: effect of frequency-dependent soil parameters, IEEE Trans. Power Deliv. 29 (2) (2014) 716–723.
S. de Castro Assis et al. / Electric Power Systems Research 149 (2017) 169–177 [26] J.T. Whitehead, Lightning performance of TVA’S 500-kV 161-kV transmission lines, IEEE Trans. Power Appar. Syst. PAS-102 (3) (1983) 752–768. [27] S. Visacro, A. Soares, M.A.O. Schroeder, L.C.L. Cherchiglia, V.J. de Sousa, Statistical analysis of lightning current parameters: measurements at Morro do Cachimbo Station, J. Geophys. Res.: Atmos. 109 (D1) (2004).
177
[28] S. Visacro, C.R. Mesquita, A. De Conti, F.H. Silveira, Updated statistics of lightning currents measured at Morro do Cachimbo Station, Atmos. Res. 117 (2012) 55–63. [29] Pes, I. IEEE PES Lightning Performance of Overhead Lines Working Group: 15.09.08. http://ewh.ieee.org/soc/pes/lpdl/ (Accessed 11 November 2016).