Backflashover Analysis for 110-kV Lines at Multi-Circuit Overhead Line Towers

Transcription

1 Backflasover Analysis for 11-kV Lines at Multi-Circuit Overead Line Towers M. Kizilcay, C. Neumann Abstract-- An increase of back-flasovers in a 11-kV system as been observed along an overead line route tat consists of multi-circuit transmission of voltage levels 38-kV, 22- kv and 11-kV at te same tower. Te eigt of multi-circuit varies in te range of 55 m 88 m. Te 11-kV doublecircuit line is positioned at te lowest cross-arm of te tower. Influence of te various factors on te back-flasover of te 11- kv insulator strings as been studied by means of EMTP-ATP simulations. Different current waveforms of te ligtning stroke ave been used to represent te first stroke and subsequent strokes. Available flasover analysis metods like leader development metod by Pigini et al and by Motoyama, as well as te voltage-time integration metod by Kind ave been implemented using a simulation language and te performance of flasover models as been compared. Te purpose of tis study was to identify tose at wic back-flasover is more likely to occur tan at oter along te line route. Replacement of one insulator string of a 11- kv duplex insulator by a surge arrester is sown to be a successful mitigation tecnique to reduce te back-flasover rate of tose 11-kV lines. Keywords: flasover, back-flasover, ligtning stroke, ligtning surge, transmission tower, EMTP. I. INTRODUCTION e tripping of a 11-kV double-circuit overead line T as been increased in a certain region at tunderstorms, were relatively tall multi-circuit transmission were installed. Te multi-circuit transmission route consists of 38-kV, 22-kV and 11-kV overead lines at te same tower. Ligtning strokes registered by ligtning flas counters in tis region sowed a maximum stroke current of 9 ka. Te ig-frequency measurement of te tower footing resistance wit a 26-kHz measuring current as revealed tat te resistance value is relatively ig at te tree. A back-flasover analysis sould indicate wic of tat 5.2-km line route are rater prone to back-flasovers of te 11-kV insulation strings depending on different factors like tower footing resistance, tower surge impedance, tower eigt, etc. Tere are various metods publised before to model lines,, ligtning strokes and flasover mecanism over te insulators. Since measurements on real [1] are costly, various simulation models sould be compared wit eac oter to validate te simulation results. A measure to prevent back-flasovers is to replace one insulator string of a duplex line insulator by a surge arrester. Te protective level of te surge arrester for ligtning strokes sould be selected suc tat te surge arrester discarges earlier tan te flasover of te insulator. Furtermore, it is important to equip a series of wit surge arresters witout leaving out a tower in-between. Te transients program EMTP-ATP wit te integrated simulation language MODELS is well suited to analyze ligtning surge penomenon on overead lines as reported several times in publications [3], [4]. II. MODELING METHOD Te modelling metods for te back-flasover analysis applied in tis paper are based upon various publications in tis field [3], [6] [9]. A. Multi-Circuit Towers Te eigt of multi-circuit varies in te range of m. Te tower structure also varies from tower to tower along te 5.2-km route. Te layout of a typical suspension tower is sown in Fig. 1. Te distances are given in meters. Te upper two cross-arms carry at left and rigt side a 22-kV and 38-kV single-circuit line, respectively. A 11-kV double-circuit line is suspended from te lowest cross-arms. Te tower is represented by loss-less Constant-Parameter Distributed Line (CPDL) model [2]. Te propagation velocity of a traveling wave along a tower is taken to be equal to te ligt velocity, [3], [8]. Te tower traveling time is τ t =. c is te tower eigt. Tere are several formulas to calculate te surge impedance of te tower [3], [8-1]. As a basis, te formula given in [1] for waisted tower sape (Fig. 2) and recommended by IEEE and CIGRE [8] is used: Mustafa Kizilcay is wit te University of Siegen, Department of Electrical and Computer Engineering, Siegen, Germany ( kizilcay@ieee.org). Claus Neumann is wit te RWE Transportnetz Strom, electric power utility, Dortmund, Germany ( claus.neumann@rwe.com). Presented at te International Conference on Power Systems Transients (IPST 7) in Lyon, France on June 4-7, 27 Z twaist = 6 ln cot.5 tan 1 R r 1 2+ r 2 + r 3 1 were R = and = (1) 1

2 1 2 Fig. 1. Layout of a typical multicircuit suspension tower Fig. 2. Waisted tower sape to calculate tower surge impedance For a tower of 76.5-m eigt (1) delivers te following value: Zt waist= Ω. It is recommended in Japan [3] to consider frequency-dependent effects for wave propagation along, wen te tower footing impedance is represented by a linear resistance, wic is te case in tis study. Te tower model consisting of CPDL model sections is added by RL parallel circuits at eac section to represent traveling wave attenuation and distortion as sown in Fig. 3. Te RL values are determined as functions of surge impedance Z t, traveling time τ t, distances between cross-arms x 1, x 2, x 3, x 4, and Fig. 3. Tower model wit attenuation factor, α =.89 as additional RL-circuits recommended in [3] by following equations: xi 1 Ri = 2Zt ln (2) α L = 2τ R (3) i t i Te cross-arms are not represented in te tower model. B. Number of Towers Total 19 of a part of a line route sown in Fig. 4 are represented including all overead lines. Direct ligtning strokes to between tower #1 and #12 are analyzed. Fig. 4. Modelled part of te transmission line route wit a junction at tower #1 (GW: ground wire) C. Transmission Lines All overead lines at te same tower are represented by te CPDL model at f = 4 khz. Ground wire is represented like a pase wire, wic is connected to te top of te (see Fig. 1). Data of te conductors are: - 38 kv: 4 conductors/pase, ACSR 265/35 Al/St - 22 kv: 4 conductors/pase, ACSR 265/35 Al/St - 11 kv: 1 conductor/pase, ACSR 265/35 Al/St - ground wire: AY/AW 216/33 (aerial cable) In order to take into account te effect of te AC steadystate voltage of te lines on a ligtning surge, te transmission lines are connected to AC voltage sources via multipase matcing impedance (surge impedance matrix). D. Ligtning Current and Impedance Te ligtning stroke is modeled by a current source and a parallel resistance, wic represents te ligtning-pat impedance. Ligtning-pat impedance is selected as 4 Ω according to [3]. Two different ligtning current waveforms are used to represent a) first stroke and b) te subsequent strokes: a) CIGRE waveform of concave sape wit front time, T f = 3µs and time to alf value, T = 77.5 µs. b) Linear ramp waveform wit T f = 1µs and T = 3.2 µs In fact, according to [8] te front time of te first stroke depends on te peak value of te ligtning current. In tis study T f and T are assumed to be constant. Te maximum rate-of-rise S m of te current as been so adjusted, tat te ratio I T f to S m corresponds to te average values of a first stroke: I = 31 ka, Sm = 26 ka/µs, Tf = 3 µs [8]. Fig. 5 sows bot current waveforms wit a magnitude of 5 ka. E. Flasover Models Flasover models estimate te breakdown of te air between te arcing orns of te line insulators under nonstandard wave forms. 2

3 current (ka) ramp CIGRE time (µs) Fig. 5. Ligtning current waveforms; CIGRE concave waveform, linear ramp function In te literature tere are mainly two metods are known besides te simple flasover estimation by means of a volttime curve of an insulator [7], [8]. Tey are integration metods and Leader development metods. In tis study tree flasover models are applied for comparison purposes. a) Equal-area criterion by Kind [6], [8], [14]; b) Leader development metod by Motoyama [4], [12]; c) Leader development metod by Pigini et al. [8], [13]. Wave deformation due to corona is not considered in te ligtning surge simulations. Te surge propagating on te ground wire can be normally deformed by corona. In tis paper it is assumed tat te ligtning stroke terminates at te tower. 1) Equal-area criterion by Kind Te criterion by Kind requires two parameters, U and F, and it is tested simply by evaluating te following integral numerically: t flo [ ut () U ] dt F (4) were u(t) is te voltage waveform across te insulator. Wen te time integral of te voltage difference (u U ) becomes greater tan te value of F, ten at t = t flo te flasover occurs. In oter words, any impulse voltage waveform can lead to a flasover, if a certain volt-time area will be covered. Te unknown parameters U and F can be obtained from te 5 % sparkover volt-time caracteristic of te insulator [16]. Tis caracteristic is establised as a function of te insulator lengt [7] and alternatively, from te known arcing distance of te 11-kV insulators [15]. Bot curves are very close to eac oter and sown in Fig. 6. Te unknown parameters in (5) are determined according to [16]: U = kv, F =.34 Vs. flasover voltage (kv) time to breakdown (µs) reference [7] reference [15] Fig % flasover volt-time curve of te 11-kV insulator 2) Leader development metod by Motoyama Leader development metods for flasover analysis gives special consideration to pysical aspects associated wit te discarge mecanism [7], [8]. Te flasover model by Motoyama [12] is developed for sort tail ligtning impulse voltages. It is based on experiments for 1m 3m gap lengts. Te leader onset condition for positive polarity is used: T s 1 utdt ( ) Uave 4( kv) D( m) 5( kv) T = > + (5) s were u(t) is te imposed voltage between arcorns and D is te gap lengt in meter. T s is te streamer developing time (= leader onset time). Te leader developing process is defined by following equations: v LAVE ut () K1A E for xlave < D 4 D 2 xlave ( t) = (m/s) (6) ut () K1B E for D 4 xlave < D 2 D 2 xlave ( t) LAVE ( ) ( ) x t = v t dt (m) (7) AVE ql = 2K xlave (C) (8) were q L is te carge accumulated in te leader; x LAVE is te average value of te leader-developing lengt; and v LAVE is te leader-developing velocity. Te constants E, K, K 1A, K 1B are set to 75 kv/m, 41 µc/m, 2.5 m²/(vs) and.42 m²/(vs), respectively. Te breakdown occurs wen x LAVE attains D/2. If te applied voltage u(t) becomes less tan E ( D 2x LAVE ) during a leader-developing process, te leader is considered to stop its development. 3) Leader development metod by Pigini et al. Te flasover condition is estimated by te imposed voltage across te air gap. Te leader onset condition is given as [13] ut () E p D (9) were D is te gap lengt and E p = 67 kv m. Te equivalent leader-developing velocity v l (m/s) is computed according to following equation, wic was evaluated by several measurements [13]: ut () vl = 17 D E p exp(.15 u( t) / D) (1) D ll were l l is te leader lengt in meter; u(t) is te voltage imposed to te air gap. Te leader lengt is obtained by te integral of leader-developing velocity: l l ( ) l = v t dt (11) Te breakdown occurs, wen te leader lengt l l is equal to te gap lengt D. 3

4 4) Representation of te Air Gap Breakdown Te discarge in te air gap can be represented by a timedependent arc resistance, decreasing from 1 kω, to 1Ω in.1 µs and to.1ω in 1 s, based on [14]. III. BACK-FLASHOVER PERFORMANCE ESTIMATION In order to estimate rougly wic on te route from tower #1 to #12 (Fig. 4) are endangered by backflasovers across 11-kV insulators, a systematic analysis is performed. Following two ligtning current waveforms are injected to eac tower in question. CIGRE waveform, I = 2 9 ka; 3 µs / 77.5 µs Linear ramp function, I = 2 9 ka; 1µs / 3.2 µs. Te current amplitude as been increased in 5 ka steps from 2 ka up to 9 ka and back-flasover across te 11- kv insulators as been examined simultaneously by te tree flasover models. Te simulation results are summarized in figures 7 and 8 for te two current waveforms. In tose diagrams te minimum ligtning peak current is sown tat causes a back-flasover at te 11-kV insulator. Following observations are made regarding back-flasovers: ligtning peak current (ka) Generally #3, #4, #5, #8, #9 are more likely to produce a back-flasover tan te oter. Te back-flasover at te depends on te current waveform as expected Kind Pigini et al Motoyama mean value Fig. 7. Minimum ligtning peak currents of CIGRE waveform (3/77.5 µs) causing back-flasover at te 11-kV insulators ligtning peak current (ka) Kind Pigini et al Motoyama mean value Fig. 8. Minimum ligtning peak currents of linear ramp type (1/3.2 µs) causing back-flasover at te 11-kV insulators tower surge impedance (Om) Te flasover models perform differently depending on te ligtning current waveform. Te flasover model by Motoyama produces conservative results in Fig. 8 compared to te oter two models, i.e. te flasovers occur at iger ligtning currents. A similar beavior can observed by te metod of Pigini in te case of steep linear ramp function. Te equal-area criterion by Kind performs well in bot cases. Tere is a clear inverse correlation between te flasover tendency and te tower footing resistance, and a rater weak correlation between te flasover tendency and tower surge impedance can be observed in Fig tower surge impedance tower footing resistance Fig. 9. Tower surge impedances and measured footing resistances eigt (m) tower eigt eigt of te 11-kV crossarm Fig. 1. Heigt of and crossarms of te 11-kV line Anoter factor influencing te flasover performance of te 11-kV insulators is te eigt of te tower and crossarm of te 11-kV line as sown in Fig. 1. Taking te probability distribution relation for ligtning crest current magnitudes according to IEEE [9] 1 pi ( > I) = (12) 2.6 I ka into consideration, it can be said tat at te mostly endangered #3 and #8 wit an average peak value of I = 36 ka, 4 % of ligtning strokes would cause a back-flasover across 11-kV insulators. IV. MITIGATION OF BACK-FLASHOVERS BY LINE SURGE ARRESTERS Line surge arresters parallel to te pase insulators of 11 kv circuits prevent back-flasovers at tose [18] tower footing resistance (Om) 4

5 Towers #3, #5 and #8 are selected as endangered by back-flasovers of te 11-kV lines and are equipped wit line surge arresters, wic replace one insulator string of te duplex insulator of te double-circuit 11-kV line. Te model referring to [17] of te selected surge arrester wit rated voltage of 156 kv and its nonlinear voltage-current caracteristic are sown in figures 11 and 12, respectively. It can be easily cecked by te Kind equal-area criterion tat no flasover can occur across te insulator string parallel to te surge arrester, because te voltage across te insulator will be limited by surge arresters below U in (4). 5 L =.37 µh; R = Ω; C = 65.1 pf Fig. 11. Surge arrester model across te 11-kV pase insulators. discarge current (A) time (µs) pase pase A cct.2 pase pase C pase A cct.1 pase B cct.1 C cct.1 B cct.2 cct.2 Fig. 13. Discarge current of te six line surge arresters at tower #3 (no flasover at adjacent is assumed) T1 pase C T2 pase C T4 pase C T5 pase C time (µs) voltage (MV) Fig. 14. Voltages between pase c and te tower at #1, #2, #4 and #5 of te 11-kV line due to discarging of line surge arresters at tower #3. No flasover at adjacent is assumed. Voltage (kv) voltage (MV) T2: pase B T4: pase B T4: pase C time (µs) Fig. 12. Current (ka) Voltage-current caracteristic of te surge arrester Te simulations of ligtning strokes wit I = 2 ka; 3 µs / 77.5 µs at te #3 and #8 confirmed also tat no breakdown can occur across parallel insulators according to te oter two flasover models by Pigini et al and Motoyama. Due to discarging of te surge arresters (see Fig. 13) te voltage of te 11-kV pase conductors temporarily increases significantly. Fig. 14 sows voltages of pase c at te #1, #2, #3, #4 and #5, wen a ligtning stroke wit I = 1 ka; 3 µs / 77.5 µs its te top of te tower #3. Te operating 5-Hz voltage of pase c is at moment of te ligtning stroke equal to te negative peak value (9 kv). Depending on te amplitude of te discarge current of surge arresters, a flasover may take place at oter, wic are not equipped wit surge arresters. In tis respect two cases ave been studied: ligtning stroke to #3 and #8, wic are equipped wit line surge arresters for 11 kv. Adjacent do not contain any line surge arresters. Te CIGRE current waveform wit 3/77.5 µs as ligtning stroke is used by increasing te amplitude in 5 ka steps. Te flasover condition is cecked by te Kind equal-area criterion. At tower #3, wen I > 95 ka and at tower #8, wen I > 9 ka, a flasover is expected at te adjacent Fig. 15. Waveforms of voltages across 11-kV pase insulators wit flasovers at #2 and #4 (no line surge arresters are installed at tose ) Waveforms of te voltage across flased-over insulators are sown in Fig. 15 for te case of ligtning stroke to tower #3 wit 11 ka. At tower #2 te pase b and at tower #4 te pases b and c attain flasover. An important question is, ow well te surge arresters will perform in terms of energy absorption. A ligtning stroke wit I = 2 ka; 3 µs / 77.5 µs is applied as worst-case to te top of #3 and #8. It is assumed tat no line arresters are installed at adjacent. Consequently, flasover takes place in all pases of te 11-kV double-circuit line at adjacent. Maximum energy absorption computed is 34 kj, wic is uncritical. V. CONCLUSION A systematic flasover analysis as been performed for a 11-kV double-circuit overead line, wic is a part of a multi-circuit transmission route. Two different ligtning stroke current waveforms ave been applied. Te backflasover performance is estimated by means of tree different flasover models. Te at wic back-flasover is more likely tan at oters are identified in order to take countermeasures like replacement of one 11-kV insulation string by a surge arrester at tose. 5

6 Multi-circuit tower system is modeled wit te grapical preprocessor ATPDraw and te simulations are performed using EMTP-ATP. It as been sown tat line surge arresters can be successfully utilized to prevent back-flasovers across 11-kV pase insulators at endangered. For ligtning stroke current amplitudes greater tan 9 ka, flasover may occur at te adjacent due to discarge current of operated surge arresters, wen te pase conductors at tose are not equipped wit surge arresters. Energy absorption of te selected 11-kV line arresters remains uncritical. VI. REFERENCES [1] Yamada, T.; Mocizuki, A.; Sawada, J.; Zaima, E.; Kawamura, T.; Ametani, A.; Isii, M.; Kato, S.: Experimental Evaluation of a UHV Tower Model for Ligtning Surge Analysis, IEEE Trans. on Power Delivery, Vol. 1, No. 1, pp , Jan [2] Canadian/American EMTP User Group: ATP Rule Book, distributed by te European EMTP-ATP Users Group Association, 25. [3] Ametani, A.; Kawamura, T.: A Metod of a Ligtning Surge Analysis Recommended in Japan Using EMTP, IEEE Trans. on Power Delivery, Vol. 2, No. 2, pp , April 25. [4] Mozumi, T.; Baba, Y.; Isii, M.; Nagaoka, N.; Ametani, A.: Numerical Electromagnetic Field Analysis of Arcorn Voltages During a Back- Flasover on a 5-kV Twin-Circuit Line, IEEE Trans. on Power Delivery, Vol. 18, No. 1, pp , Jan. 23. [5] Dommel, H. W.: EMTP Teory Book, Bonneville Power Administration, conversion into electronic format by Can/Am EMTP User Group in [6] Scmitt, H.; Winter, W.: Simulation of Ligtning Overvoltages in Electrical Power Systems, Proceedings IPST 21 (International Conference on Power System Transients), Rio de Janerio, June 24-28, 21. [7] IEEE Fast Front Transients Task Force, Modeling and Analysis of System Transients Working Group: Modeling Guidelines for Fast Front Transients, IEEE Trans. on Power Delivery, Vol. 11, No. 1, pp , Jan [8] CIGRE WG 33-1: Guide to Procedures for Estimating te Ligtning Performance of Transmission Lines, Tecnical Brocure, October [9] IEEE Working Group on Ligtning Performance of Transmission Lines: A Simplified Metod for Estimating Ligtning Performance of Transmission Lines, IEEE Trans. on Power App. & Systems, Vol. PAS- 14, No. 4, pp , April [1] Cisolm, W. A.; Cow, Y. L.; Srivastava, K. D.: Travel Time of Transmission Towers, IEEE Trans. on Power App. and Systems, Vol. PAS-14, No. 1, S , Oktober [11] IEEE Working Group on Estimating te Ligtning Performance of Transmission Lines: IEEE Working Group Report Estimating Ligtning Performance of Transmission Lines II Updates to Analytical Models, IEEE Trans. on Power Delivery, Vol. 8, No. 3, pp , July [12] Motoyama, H.: Experimental study and analysis of breakdown caracteristics of long air gaps wit sort tail ligtning impulse, IEEE Trans. on Power Delivery, Vol. 11, No. 2, pp , April [13] Pigini, A.; Rizzi, G.; Garbagnati, E.; Porrino, A.; Baldo, G.; Pesavento, G.: Performance of large air gaps under ligtning overvoltages: Experimental study and analysis of accuracy of predetermination metods, IEEE Trans. on Power Delivery, Vol. 4, No. 2, pp , April [14] Fernandes, M.; Correia de Barros, M. T.; Ameida, M.E.: Statistical Study of te Ligtning Overvoltages at a Gas Insulated Station Transformer, Proceedings IPST 1995 (International Conference on Power System Transients), Lisbon, 3-7 September [15] Koettniz, H.; Winkler, G.; Weßnigk, K.-D.: Fundamentals of Electrical Operational Penomena in Electrical Power Systems, (original title in German: Grundlagen elektriscer Betriebsvorgänge in Elektroenergiesystemen), Deutscer Verlag für Grundstoffindustrie, Leipzig, [16] CIGRE WG 33.2: Guidelines for representation of network elements wen calculating transients, CIGRE Tecnical Brocure, No. 39, 199. [17] IEEE Working Group Surge Protective Devices Committee: Modeling of Metal Oxide Surge Arresters, IEEE Trans. on Power Delivery, Vol. 7, No. 1, pp , Januar [18] Tarasiewicz, E. J.; Rimmer, F.; Morced, A. S.: Transmission Line Arrester Energy, Cost, and Risk of Failure Analysis for Partially Sielded Transmission Lines, IEEE Trans. on Power Delivery, Vol. 15, No. 3, pp , July 2. VII. BIOGRAPHIES Mustafa Kizilcay was born in Bursa, Turkey in He received te B.Sc. degree from Middle East Tecnical University of Ankara in 1979, Dipl.-Ing. degree and P.D. degree from University of Hanover, Germany in 1985 and From 1991 until 1994, e was as System Analyst wit Lameyer International in Frankfurt, Germany e as been professor for Power Systems at Osnabrueck University of Applied Sciences, Germany. Since 24 e is wit te University of Siegen, Germany, olding te cair for electrical power systems as full professor. Dr. Kizilcay is winner of te literature prize of Power Engineering Society of German Electro-engineers Association (ETG-VDE) in His researc fields are power system analysis, digital simulation of power system transients and dynamics, insulation-coordination and protection. He is a member of IEEE, CIGRE, VDE and VDI in Germany. Claus Neumann studied electrical engineering at te Tecnical University of Aacen. After finising wit te Dipl.-Ing. degree in 1972 e was engaged as a testing engineer in te ig voltage and insulating material laboratory of a switcgear manufacturer. In 1979 e joined te department for HV equipment of RWE in Essen, Germany. Among oters e worked on non-standardised stresses of switcgear in te network, on testing tecnique, monitoring and diagnostics and on maintenance strategies. In 1992 e became ead of tis department. In 1992 e also received is Dr.- Ing. degree from te Tecnical University of Darmstadt. In 21 e became Honourable Professor at te Tecnical University of Darmstadt were e gives a lecture in HV switcgear and substations. Since 25 is in carge of te Operational Asset Management at RWE Transportnetz Strom, Dortmund, Germany. His main fields of activity are fundamental design and layout of HV apparatus and asset management including maintenance and renovation strategy, monitoring and diagnostics, environmental impact of SF6 tecnology etc.. He is among oters a member of CIGRE Study Committee C4, Power System Performance, of CIGRE SC D1 Working Group 3, Gas-insulated Systems. 6