Sunteți pe pagina 1din 6

Paper accepted for presentation at 2003 IEEE Bologna Power Tech Conference, June 23th-26th, Bologna, Italy

Line surge arrester selection to improve lightning performance of transmission lines


J.A. Tarchini and W. Gimenez
used as simulation tools to estimate the lightning performance of the studied line. It provides additional results concerning the application of surge arresters to improve such a performance, and illustrates the influence of tower footing resistance on arrester energy stress. II. INCIDENCE OF LIGHTNING IN OVERHEAD LINES The double 132 kV transmission line considered in this paper crosses a desert flat region near Mendoza city, where it is rather difficult to obtain a good tower footing resistance. In addition, this region has a high lightning activity with a ground flash density of 4,3 strokes/km/year. This line, which represents an important link between EDEMSA utility and CTM power plant, has very poor lightning performance (especially in summer) of the order of 15 outages/100 km line/year. According to data collected between 1997 and 2001, lightning is the cause of 68% of the interruptions of the line. In order to assess the incidence of direct lightning strikes over the line, expressed by the number of strokes each 100 km of line per year, the Leader Progression Model (LPM) [5,6] was used to determinate the total exposure area of the line, as described in what follows.
1.70 22.30

Abstract-- Line arresters are considered as an effective mean to improve the lightning performance of transmission lines, particularly in areas with high soil resistivity and lightning ground flash density. This paper describes a case study concerning a 132 kV transmission line of EDEMSA (Empresa Distribuidora de Electricidad de Mendoza, Argentina), and investigates the most adequate solution concerning surge arrester location and type. The different variables, that have an influence on energy stress calculations, are evaluated to select the arrester characteristics. EMTP96 models are used to estimate the critical currents that exceeds arresters rated energy stress during direct lightning return-strokes for different tower footing resistance values. The analysis is carried out with reference to arresters of different classes in order to select the kind of protection to adopt along the line in order to minimising the arrester outage due to energy stress. Index TermsLine arresters, lightning transmission lines, EMTP, energy stress. performance,

I. INTRODUCTION ightning is one of the main sources of non-programmed interruptions in transmission lines. As known, lightning strokes to transmission lines may cause failures essentially via two phenomena: shielding failure (SF) and backflashover (BF) [1,2,3]. Both concur to the formation of the overall number of flashovers of the transmission line. In shielding studies, the protection angles of ground wires can be set considering the acceptable rate of flashover due to direct impacts over the phase conductors. However, the total flashover rate is mainly influenced by the backflashover rate, which strongly depends on the value of grounding resistance of the ground wires at each tower footing. The lower the grounding resistance, the lower the number of expected interruptions due to backflashover. In areas with high soil resistivity, however, there is a technical limit which prevents the footing resistance improvement, and, in turn, the decrease of the flashover rate of the line. For these cases, such as the 132 kV line studied in this paper, the use of line surge arresters represents a proper solution to improve the Lightning Performance of the line. This paper follows a previous study [4] in which the Leader Progression Model and EMTP96 were
J. A. Tarchini is with the Department of Electrical Engineering, University of Bologna, 40136 Bologna, Italy (e-mail: jtarchini@edemsa.com). W. Gimenez is with the GISEP. National Technological University, 3000 Santa Fe, Argentina (wgimenez@frsf.utn.edu.ar)

19.50 16.50 13.50

Fig. 1: 132 kV line configuration

A. Strikes to the line Each line tower is 22.3 m high and has a shielding angle between the ground wire at the top, and the upper phase conductors of 31,3 degrees. The line geometry is shown in Fig. 1. To calculate the total number of strokes to the line, the

0-7803-7967-5/03/$17.00 2003 IEEE

main parameter to consider is the Lateral Distance (LD), which is the maximum distance a lightning stroke of a given current amplitude can hit the structure (or ground wire) or the phase conductors. In this paper, the LD value is determined by means of the LPM that allows its calculation at each individual tower and span. The lateral distance variation for the structure under study, calculated with a software program provided by CESI [7] based on the LPM, is shown in Fig. 2.
Lateral Distance Variation
100 90 80 70 60

TABLE 1: CALCULATED PARAMETERS FOR THE CONSIDERED STATISTICAL


CURRENT DISTRIBUTION

AREA EXPOSURE TOWER AREA EXPOSURE SPAN NUMBER STROKES TOWER NUMBER STROKES SPAN STROKES/100 km LINE

Aeqt Aeqs Nst Nss Nl

8.77E-03 3.20E-02 3.80E-02 1.26E-01 50.48

III. LIGHTNING EFFECTS IN TRANSMISSION LINES A. Shielding Analysis Also for the shielding failure area for each structure and span use was made of the LPM. In this case, for each tower and span, the distance from which lightning strikes the phase conductors, eluding the shield effect of the ground wire is calculated. For the considered line, SFAT is the shielding failure area of each tower, and SFAS is the shielding failure area of each span. The total shielding failure area expected (SFA), considering 100 km of line is:
SFA = 100 * ( SFA = 100 * (
N s +1 1

LD [m]

50 40 30 20 10 0 0 20 40 60 80 100 120 140

DOUBLE LINE H=22 m

Current [kA]

Fig. 2: Lateral Distance of the structure as a function of the lightning current.

SFAT + SFAS )[ km
1

Ns

/ 100 km ]
Ns 1 iC

(4)

For each tower we can calculate the relevant total equivalent exposure area (Aeqt) by taking into account the concept of LD and assessing a lognormal distribution for the lightning current amplitude [1] as follows,

N s +1 1

SFW
I min

Ic

( I ) p( I )dI + 2 S

I min

SFW ( I ) p( I )dI

Aeqt = p( I ) LD 2 ( I )dI p(I) = z=


2 1 e (z /2) 2I ln(I/I med )

Considering the standard CIGRE peak current probability distribution as previously, and the regional ground density, the Shielding Failure Rate is,

(1)

SFR = N g SFA [ strikes / 100 km / year ] Results are shown in Table 2.

(5)

Where p(I) is the density probability distribution of current amplitude, is the logarithmic standard deviation and Imed the median value of current amplitude. For each span, with a span length (S) of 300 m, the total area of exposure is calculated also from LD as:

TABLE 2: SHIELDING FAILURE RATE OF THE CONSIDERED LINE FOR THE CURRENT PROBABILITY DISTRIBUTION ADOPTED. PARAMETER SHIELDING FAILURE AREA SFA 0.054 SHIELDING FAILURE RATE SFR 0.232

B. Backflashover calculation An EMTP96 model was developed to estimate the lightning critical current (Ic) from which backflashover occurs. The critical current depends strongly on the footing resistance of ground wires at each tower. The backflashover rate of the line can be determined as a function of such a critical current [1,2,8]:
BFR = 0,6N l p(I)dI
Ic

Aeqs = 2 S p( I ) LD( I )dI

(2)

Considering the ground flash density of the region Ng, the number of strokes per year for each tower is N st = N g * Aeqt . The number of strokes per year for each span is N ss = N g * Aeqs The number of strokes for 100 km of line per year Nl is therefore:
Nl =
N s +1 1

(6)

The lightning current distribution earlier mentioned is used also for the backflashover analysis, as it will be illustrated later. 1) Footing resistance values The terrain near the line has a rocky nature, which makes it difficult to achieve good grounding. EDEMSA personnel measured the footing resistance of each tower with an

st

N
1

Ns

ss

(3)

Where Ns is the number of spans in 100 km of line. In Table 1 we show the results for the considered line, assuming the CIGRE statistical distribution [1] for lightning current peak amplitude.

important dispersion in results. The highest dc value was 117 Ohm. The results are shown in Fig. 3 as a percentage of the total measured values. For simulation models soil ionization effects were not considered.
45% 40% 35%

30% 25% 20% 15% 10% 5% 0%


HM <1 0O HM HM HM HM <1 5O <4 5O <3 0O <6 0O <1 17 O HM

TOWER FOOTING

wire, and thus the largest overvoltage stress. In the line model used for simulations, seven poles of the line were considered. Case 1: No surge arresters Case 2: Surge arresters each every two poles (odd poles for example): line performance is expected to be worst than for Case 3, but better than for Case 1. Its implementation needs a lower investment respect of the other case. Case 3: Surge arresters at each pole: this configuration is expected to give the best protection and line performance for each tower footing resistance considered. The surge arrester model proposed by the IEEE WG 3.4.11 [10] was used in the simulations (see Fig. 5). The arresters have a characteristic of 98 kV of MCOV and a 10 kA discharge of 300 kV. The arrester energy stress and failure rate will be dealt with in section 5.
R0 L0 L1

DISTRIBUTION

RT

RT

RT

RT

RT 45 <

15 <

10 <

30 <

Fig. 3: Distribution of tower footing resistance.

2) Simulation Model The line model developed in EMTP96 is based on a constant parameter transmission line model at a frequency of 100 kHz, with a ground of resistivity =500 Ohmm. The model is shown in Fig. 4. In order to model the flashover across the air insulation, a TACS [9] routine was used to control logic switches applied to the volt-time curve values of the air gap.
C i0 R1 i1

60 <

RT

Fig. 5: Frequency dependent model of surge arrester. From [10]

4) Simulation Results The lightning performance of the line was first determined for different arrester locations (Cases 1, 2 and 3) and tower footing resistance values, assumed constant for each pole. The backflashover rate of the line for Ng = 1 [stroke/km2/year] is shown in Fig. 6, as a function of the arrester location and the value of footing resistance.
7

Z c

7 Z c

R t

R t

R t

R t

R t

R t

R t

Fig. 4: Line model for EMTP96 simulation


BFR (Flashover/100 km/year)

6 5 4 3 2 1 0 10 15
CASE Nr. 2

Each tower was considered as a transmission line, with a surge impedance (Zt) calculated as [2]:
2H T Z t = 60 ln( 2 ) - 1 = 147 ,1 rT
(7)

Where HT is the tower height (HT =22,3 m) and rT is the tower base radius (rT =2,0 m). The CFO of line insulation is 550 kV, and its volt-time curve variation was determined as [2]

710 V = ( 400 + 0.75 )W t

(8)

30
CASE Nr. 3

45

60

CASE Nr. 1

Footing Resistance [Ohm]

Where V is the flashover strength (kV); t is the time to flashover (s) and W is the gap or insulator length. 3) Line surge arresters The use of surge arresters can provide different protection levels to the line, depending on their location. In the simulations carried out in this paper, surge arresters were applied only at the bottom phase conductors. These conductors have the worst coupling factor with the ground

Fig. 6: BFR for different tower footing resistance values and arrester location (for Ng =1 stroke/km2/year)

We now consider the distribution of the tower footing resistance shown in Fig. 3 and investigate how to locate the arresters along the line in order to reduce the backflashover rate up to a target value. For accomplishing that, we assume the line as divided in six sections of different length, corresponding to the six footing resistance values of Fig. 6, having the same value of tower footing resistance for each pole. We then evaluate the BFR for each line section and for

Arrester Discharge Energy [kJ]

various cases of arrester location (1, 2 and 3), eventually inferring a solution that satisfies our target, if any. It is worth mentioning that we use the same arrester configuration for each pole belonging to the same section of line. In case of more than one solution, the most convenient from the technical-economical point of view can be selected. Considering the tower footing resistance distribution of Fig. 6, the overall calculated backflashover rate of the line without surge arresters (Case 1) is 13.6 [flashovers/100 km/year]. If we fix the target backflashover rate for the whole line as 50% of this value, our calculations show that we can get such a value with the solution shown in Table 3. In particular we get a total BFR for the line of 6.1 [flashovers/100 km/year].
TABLE 3: SOLUTION THAT ALLOWS FOR THE TWO ASSUMED CURRENT DISTRIBUTIONS A REDUCTION OF 50% OF THE BFR TOWER Case of Arrester BFR RESISTANCE Location Rt<15 Ohm 1 3.6 15<Rt<30 Ohm 2 6.0 30<Rt<45 Ohm 3 4.9 45<Rt<60 Ohm 3 7.1 Rt>60 Ohm 3 9.6

1) Influence of stroke current The voltage drop of the tower footing resistance, resulting from a lightning flash to the tower top, will increase with the stroke current. When voltage across the arrester increases, more current flows through it. This increase in stress (in terms of voltage and current through the arrester) will result in more energy.
800 700 600 500 400 300 200 100 0 0 10 Ohm 45 Ohm 50 15 Ohm 60 Ohm 100 30 Ohm 150 200 Stroke Current [kA]

Fig. 7: Arrester energy for different stroke peak current. Strokes to tower top

IV. ARRESTER SELECTION A. Arrester energy stress calculation Traditionally line arresters have been applied to protect unshielded transmission lines. In terms of reliability, line arresters have to be dimensioned or selected in such a manner that they would rarely fail when stressed. For the case presented in this study we considered for arrester energy calculations the effect of lightning strokes hitting the phase conductors (shielding failure effect) or the ground wire (backflashover effect). In both cases, to select the proper arrester class as a function of its energy stress, voltage across the arrester, lightning peak current and time to half-width of the impinging waveshapes are considered in arrester failure calculations. The energy dissipated through the arrester depends on different quantities and a sensitivity analysis was carried out considering the following parameters: Voltage: it depends strongly on ground footing resistance at each pole. For a fixed current, and time-to-half values, voltage varies as a function of the tower footing. Current: the critical lightning peak current from which energy dissipated within the arrester is larger than the rated one specified for each arrester class, was determined by means of EMTP96 simulations. The current variation for different tower footing resistance is calculated in order to find the energy stress limit. Time to half value: Energy increases with it. A sensitivity analysis has been carried out, in order to select the proper arrester, considering the different parameters that affects its energy calculation.

As shown in Fig. 7, arrester energy depends on the lightning stroke amplitude and footing resistance value. For the line under study, when tower footing resistance values are larger than 30 Ohm, flashovers that appear on the conductors without arresters absorb most of the energy, and arrester energy stress depends essentially on the stroke current amplitude. Depending on the line configuration, backflashover of line insulation in other phases without arresters provides a parallel path to ground through which a portion of the stroke is diverted from the arrester. Backflashover critical current decreases with the increase of tower footing resistance value. This situation is shown in Fig. 8.
100.0

Surge Arrester Current [kA]

10.0

1.0

0.1 0 10 Ohm 45 Ohm 50 15 Ohm 60 Ohm 30 Ohm 100 150 Stroke Current [kA] 200

Fig. 8: Arrester Current vs. Stoke peak current for different tower footing resistance values

For different tower footing resistance values, up to a certain critical lightning peak current where flashover appears in other phases (which depends on footing resistance), surge

arrester current increases with tower footing resistance increasing. From this lightning peak current value, arresters placed in towers with lower footing resistance are more stressed than arresters in towers with larger footing resistance. 2) Influence of footing resistance Arrester discharge current and energy depends on the tower footing resistance value. Fig. 9 shows the critical lightning peak current, namely the current which allows to reach the rated arrester discharge energy (550 kJ) across the SA, for each footing resistance considered, and increasing the time to half value of lightning waveshape [Th].
180 170 160 CRITICAL CURRENT [kA] 150 140 130 120 110 100 90 80 70 50 100 150 200 Th [us] 10 OHM 15 OHM 30 OHM 45 OHM 60 OHM

3) Influence of time to half value Energy discharged across the arrester is the integration over the time of instantaneous power stressing it. Increasing the duration of lightning stress increases the arrester energy absorption. Fig. 12 shows the effect of the time to half value of the stroke current for different current peak values and fixed tower footing resistance.
1.4E+06 Arrester Discharge Energy [J] 1.2E+06 1.0E+06 8.0E+05 6.0E+05 4.0E+05 2.0E+05 0.0E+00 0 50 100 150 200

150 kA

80 kA

100 kA

Time to half value [s]

Fig. 12: Arrester discharge Energy for 10 Ohm of tower footing resistance

V. APPLICATION CRITERIA To predict the arrester risk of failure P(Af) [11], energy duties are determined statistically. The arrester risk depends on the probability density function of energy stress p(Ws) (which is expressed in terms of random variables of lightning) and the cumulative distribution function of energy capability P(Wc) [12].
P( Af ) = P( Wc ) p( Ws )dWs
0

Fig. 9: Stroke Critical current for different time to half value

Fig. 10 and 11 show the arrester current and the discharged energy across it, respectively, for different values of lightning peak current and time to half value. In general, tower footing resistance increasing leads to the increase of arrester current and energy duty.
800 700 Surge Ar rester Current [A] 600 500 400 300 200 100 0 0 20 10 kA Stroke Current 20 kA Stroke Current 40 60 80 Footing Resistance [Ohm]

(9)

The probability of arrester energy capability P(Wc) can be approximated [11] by a Weibull cumulative distribution, where the rated energy capability supplied by the manufacturer (WR) is assumed to have a zero probability of failure.
P(Wc) = f ( Z=
( + 1 )5 WC ) = 1 0 .5 4 WR Z

(10)

WC WR 2.5 0.375

Fig. 10: Arrester current for different footing resistance values


600

500 400 300 200 100 0 0 20 40 60

80

40 kA

80 kA

Footing Resistance [Ohm]

Fig. 11: Arrester discharge energy for different tower footing resistance values

WC is the energy capability for a probability of failure of P(Wc). To represent the cumulative distribution, and estimate the probability of arrester failure, we considered a rated energy capability WR of 548 kJ. Thus, for 0%, 1%, 5%, 20%, 50% and 70% probability of failure, the energy capability WC is 550.0 kJ, 900.5 kJ, 1036.3 kJ, 1203.3 kJ, 1370.0 kJ and 1466.0 kJ respectively. The next step is to calculate the density function of energy stress probability. The probability that a surge arrester could be damaged due to a direct lightning stroke to a pole depends on the probability density function of peak value of lightning current (P(I)) and the probability density function of time to half-value of current waveshape (P(Th)). In this paper we use

Arrester Discharge Energy [kJ]

the standard CIGRE distribution [3] for lightning current peak amplitude and time to half-value statistics, with a median value of 77.5 s and a standard deviation of 0.577. The arrester energy stress probability is defined by (11) [14,15]. Results are shown in Table 4 for different tower footing resistance values.
P( WA ) = f ( I )dI f ( Th )dTh = [ 1 F ( I )] f ( Th )dTh (11)
0 Ic 0

The lightning performance of the considered HV transmission line can be improved by adequate arrester installation strategy, as proposed in this paper. Arrester energy rate was selected considering the different random variables of lightning and the cumulative distribution function of energy capability. VII. ACKNOWLEDGMENT This work is part of a PhD thesis developed by the author at the Department of Electrical Engineering of the University of Bologna, under the supervision of Prof. C. A. Nucci. The authors gratefully acknowledge CESI, in special Dr. M. Bernardi for her contributions in the set up of Leader Progression Model, and Dr. Bill Chisholm for his assistance and advise concerning line modeling. VIII. REFERENCES
[1] [2] Working Group 33.01 CIGRE - Guide to procedures for estimating the lightning performance of transmission Lines, October 1991. IEEE Working Group on Lightning Performance on Transmision Lines Report Estimating Lightning Performance of Transmision Lines II Updates to Analytical Models IEEE Transactions on PWD, Vol 8, N.3, July 1993. IEEE Working Group on Lightning Performance on Transmision Lines A Simplified method for estimating the Lightning Performance of Transmision Lines, April 1985. J. Tarchini Improvement of lightning performance of transmissions lines by use of line surge arresters 26th IPST, Krakow, September 2002 L. Dellera, E. Garbagnati - Lightning stroke simulation by means of the leader progression model (Part I) IEEE Transactions on PD. Vol.5, No.4. November 1990. L.Dellera, E.Garbagnati - Lightning stroke simulation by means of the leader progression model (Part II) IEEE Transactions on PD. Vol. 5, No. 4. November 1990. CESI Int. Technical Report SR-953/1 Nov. 1986 C.A. Nucci, F. De la Rosa, V.A. Rakov Lightning and its impact on power Systems Cigr SC33 International Conference, Zagreb 1998. Dommel EMTP96 Reference Manual (Theory Book) BPA, August 1986. IEEE Working Group 3.4.11 (Application of surge protective devices Subcommittee. Surge Protective devices committee) Modelling of metal oxide surge arresters. A.R. Hileman Insulation Coordination for Power Systems Marcel Decker, 1999. Standard reference: IEC 60071-2 - "Insulation co-ordination Part 2: Application guide", 1996 M.L. Martinz, L.C. Zanetta A testing method to evaluate the energy withstanding capability of metal oxide surge arresters CIGRE SC33 Colloquium, Toronto 1997 S. Sadovic, R. Joulie et al Use of line surge arresters for the improvement of the lightning performance of 63 kV and 90 kV shielded and unshielded transmission lines IEEE Transaction on PD, Vol.12, July 1997. M. Babuder, S. Sadovic et al Lightning Perfomance Improvement of 123 kV Transmission line by use of line surge arresters High Voltage Engineering Symposium, IEE 1999.

TABLE 4: CUMULATIVE PROBABILITY DISTRIBUTION P(WA) OF ARRESTER ENERGY [KJ] FOR DIFFERENT TOWER FOOTING RESISTANCE VALUES 10 % 5% 2% 1% 0.5 % 0.1 % 10 Ohm 56.9 75.2 296.9 396.9 446.8 714.2 15 Ohm 100.9 133.4 334.6 434.6 484.6 757.9 30 Ohm 144.8 181.7 336.0 436.0 486.0 726.9 45 Ohm 152.8 187.6 303.8 403.8 453.8 668.7 60 Ohm 151.1 183.4 275.1 375.1 425.1 628.5

To calculate the arrester probability of failure P(Af), estimate the arrester failure rate and the mean time between failures along the line, the number of direct flashes to the line must be considered. According to LPM simulations, the line under study collects 50.48 [strokes/100 km line/year]. Considering 2 arresters in the bottom phases for each tower, we can convert the number of SA failures each 100 km of line per year into a failure rate per 1000 arresters. The probability of failure P(Af) for each tower footing resistance value, the arrester failure rate (number of arrester failures each 1000 SA per year) and mean time between failures (years with at least one arrester failure for each 1000 arresters installed), is shown in Table 5.
TABLE 5: PROBABILITY OF FAILURE, ARRESTER FAILURE RATE AND MEAN TIME BETWEEN FAILURE

[3] [4] [5] [6] [7] [8] [9] [10]

From these estimations, we can conclude that for the line under study, the number of arrester failures show a reasonable
RT Ohm 10 15 30 45 60 Probability of arrester failure

P(Af)
% 0.0242% 0.0547% 0.0479% 0.0692% 0.2540%

Annual number of SA failure Nfail Fail/100km/ye ar 0.0028 0.0064 0.0056 0.0081 0.0298

Arrester Failure Rate SAFR AF/1000 arresters/year 0.00427 0.00963 0.00844 0.0122 0.0448

MTBF Year/1000 Arresters 234.3 103.8 118.5 81.9 22.3

[11] [12] [13] [14]

value, considering the line performance and power quality reliability improvement by means of the arrester application. The reliability of the design could be improved by increasing the energy capability of the surge arrester. VI. CONCLUSIONS The Leader Progression Model has been successfully used to determine the strike incidence to the line and the Shielding Failure Rate from the calculation of Shielding Failure Width. Backflashover effects are simulated by means of EMTP96 to infer the critical currents for each line arrester configuration and tower footing resistance.

[15]

S-ar putea să vă placă și