Energy Stress of Surge Arresters Due to Temporary Overvoltages

Transcription

1 Eergy Stress of Surge Arresters Due to Temporary Overvoltages B. Filipović-Grčić, I. Uglešić, V. Milardić, A. Xemard, A. Guerrier Abstract-- The paper presets a method for selectig the rated voltage of a metal oxide surge arresters (MOSA) based o the calculatio of eergy stresses. The electrical behaviour of gapless MOSA uder temporary overvoltages (TOV) is importat whe selectig the rated voltage of MOSA. The study of the appropriate MOSA model is coducted for the evaluatio of eergy stresses due to TOV. The resistive part of the leakage curret was derived from the voltage-curret (U-I) characteristic for AC voltage that was gaied durig the laboratory tests. The MOSA model was implemeted i the study of eergy stresses of statio arresters that are istalled at both termials of a compact upgraded lie. By implemetig the proposed method MOSA with relatively low protectio level ca be selected without beig overstressed by TOV ad thus the overvoltage protectio of compact lie ca be improved. II. LABORATORY MEASUREMENTS I order to determie the behaviour of MOSA uder TOV laboratory measuremets were coducted o MOSA whose cotiuous operatig ad rated voltages are U c = kv ad U r =5 kv, respectively. A laboratory circuit used for the measuremet of the U-I characteristic uder AC voltage is show i Fig.. Keywords: TOV, MOSA, eergy stress, laboratory measuremets, compact lie, EMTP-RV, rated voltage selectio. I. INTRODUCTION So far the voltage ratig of statio MOSAs has bee usually chose higher tha the amplitude of the TOV which ca occur i the system. This approach leads to the use of MOSAs with high protectio levels, especially if the system is ot solidly grouded, which do ot reduce the level of slow-frot overvoltages (SFO) ad fast-frot overvoltages as efficietly as MOSAs with lower protectio level would do. SFO due to eergizatio or reclosig of the lie may be reduced by usig various techiques like poit-o-the-wave switchig or circuit breakers (CBs) equipped with pre-isertio resistors, but SFO due to faults o compact or upgraded lies might get critical. Obviously they caot be reduced by the techiques listed above ad the use of statio MOSAs with a low protectio level at the termials of the lie might be the oly optio [], []. This low protectio level requires the use of MOSAs whose rated voltage might be below the level of TOV ad, cosequetly, the MOSA eergy stress has to be determied accurately. Therefore, i the first part of the paper laboratory measuremets were performed i order to obtai MOSA model for calculatio of eergy stress due to TOV. B. Filipović-Grčić, I. Uglešić ad V. Milardić are with are with Faculty of Electrical Egieerig ad Computig, Uiversity of Zagreb, Croatia ( of correspodig author: bozidar.filipovic-grcic@fer.hr). A. Xemard is with EDF R&D, Paris, Frace (alai.xemard@edf.fr). A. Guerrier is with RTE, Paris, Frace (araud.guerrier@rte-frace.com). Paper submitted to the Iteratioal Coferece o Power Systems Trasiets (IPST) i Delft, the Netherlads Jue -7, Fig.. Experimetal circuit for the measuremet of leakage curret A capacitive divider was used for measuremet of voltage applied o MOSA, ad the total leakage curret sigal was obtaied from a shut resistace R shut. The voltage sigals were recorded usig a digital oscilloscope ad the data was saved i a computer. The total leakage curret cosists of a resistive ad a capacitive part. A method for calculatio of the resistive part of leakage curret was developed i MATLAB software. A method is based o compesatio techique []. The applied voltage cotais harmoic compoets: where u = U ( ωt + Φ ) si () the rak of the harmoics (= correspods to the fudametal compoet); U voltage peak value of the -th harmoic compoet; Ф phase agle of the -th harmoic compoet. The capacitive curret i c cotais the fudametal ad higher harmoics compoets: ic = ic + ic () The compoets i c arise because of the voltage harmoics. The curret i r is also composed of fudametal ad higher harmoics compoets:

2 Compoet characteristics whereas ir = ir + ir () r ' r r i = i + i () ' i r arises due to the oliearity of MOSA U-I harmoics. The magitude of i i r arises due to presece of voltage r voltages U ad their phase agles trasform, the compoet depeds o the peak value of Φ. By applyig Fourier i r is derived from i r ad the ' remaiig compoet i r is used for modellig the U-I characteristics of MOSA. At U c the curret through MOSA cosists primarily of capacitive curret ad a small resistive compoet of o-siusoidal curret. Measured total leakage curret is show i Fig. ad calculated resistive curret i Fig. [V] x - Total leakage curret x - Curret [A] Raisig the voltage icreases the resistive compoet of the curret ad eergy losses. For AC voltage, the curret waveshape chages sigificatly aroud the kee of the U-I characteristics, where starts the breakdow regio. Measured U-I characteristics i per-uit of the U r for AC voltage (5 Hz) are show i Fig Temperature= C AC resistive curret AC capacitive curret AC leakage curret.. Curret (ma) Fig.. Measured U-I characteristics After the voltage has exceeded the kee of the U-I characteristic, the capacitive compoet becomes egligible small compared to the resistive oe. For example, if rated voltage U r is applied o MOSA, the total leakage curret ad its resistive ad capacitive compoets are show i Fig. 5, Fig. ad Fig. 7 respectively. 8 x Total leakage curret x Fig.. Waveshapes of voltage U c ad total leakage curret (peak value I=55.5 µa) x Resistive curret x.5 [V] Curret [A] [V] Curret [A] Fig. 5. Waveshapes of voltage U r ad total leakage curret (peak value I=. ma) 8 x Resistive curret x Fig.. Waveshapes of voltage U c ad resistive curret (peak value I R=.5 µa) Resistive curret I R is ot symmetrical about its peak value (Fig.) ad there is a offset at the voltage zero crossig. Therefore for this measurig poit the peak of capacitive curret is higher tha the peak of total leakage curret. The resistive compoet creates eergy losses ad icreases the temperature of MOSA, compared to surroudig temperature. [V] Fig.. Waveshapes of voltage U r ad resistive curret (peak value I R=. ma) - - Curret [A]

3 [V] 8 x - Capacitive curret x Fig. 7. Waveshapes of voltage U r ad capacitive curret (peak value I C=.7 ma) III. MODELLING OF MOSA MOSA model used for TOV studies i EMTP-RV [], [5] cosists of oliear resistace defied by average U-I R characteristic (Fig 8.) i parallel with capacitace. Average U-I R characteristic was obtaied by joiig the peak values of the various U-I R loops at differet voltage levels (Fig 9.). MOSA capacitace of 5. pf was determied from the measuremets with RLC-Meter istrumet. - Curret [A] itegratio i MATLAB: E = k= u i k k + uk ik t () MOSA eergy was calculated i EMTP-RV by usig expressio () with time step t= µs. Compariso of measuremet ad calculatio for applied voltage U r is show i Fig.. Eergy waveshapes show steps because the eergy icreases more rapidly at the istat whe resistive curret ad voltage reach the peak value (Fig.). Eergy slowly icreases i the area where the resistive part of curret is low. Eergy (J) Measured eergy Calculated eergy Fig.. Compariso of measured ad calculated eergy for applied voltage U r Differece betwee the results of measuremet ad calculatio occurs due to the hysteretic behaviour of MOSA (Fig.). Fig. 8. MOSA model for calculatio of eergy stress due to TOV Temperature= C.. Curret (ma) Fig. 9. Average U-I R characteristic of MOSA MOSA eergy was calculated by usig the followig expressio: t E = u( t) i ( t) dt (5) R The eergy was determied from measured voltage ad resistive part of curret by usig trapezoidal umerical Resistive curret (A) x (V) x Fig.. Measuremet of the dyamic U-I R curve for the applied voltage U r The MOSA model implemeted i EMTP uses the static U-I R characteristic, which is show i Fig. The differece betwee the calculated ad the measured resistive compoet of the curret is show i Fig. ad. The hysteretic behaviour of U-I R characteristics at differet voltages is show i Fig 5. The ifluece of hysteretic behaviour o calculatio of eergy stress is higher i a low curret regio of the dyamic U-I R characteristic ad it decreases as the applied voltage across the

4 MOSA icreases. Resistive curret (A) x (V) 8 x Fig.. Static U-I R characteristic i EMTP for applied voltage U r Curret (A) x Time(s) Fig.. Compariso of measured ad calculated I R for applied voltage U r IR(mA) Measuremet Calculatio Harmoic order Fig.. Compariso of measured ad calculated harmoic spectrum of I R Resistive curret (ma) U=.97 p.u. U=.995 p.u. U=. p.u Fig. 5. Measured U-I R characteristics at differet voltages Compariso of measuremet ad calculatio results after t= ms for U/U r >.79 p.u. is show i Fig. Results show good agreemet. Eergy (J).. Calculatio Measuremet Fig.. Compariso of measuremet ad calculatio for U/U r >.79 p.u. The calculatio of the resistive curret from the total leakage curret should be doe whe modellig MOSA for the calculatio of eergy at low values of TOV (low curret regio). Large errors i calculatio of the eergy stress ca be obtaied if the total leakage curret is used for the modellig of the MOSA istead of the resistive compoet []. IV. CALCULATION OF TOV ON COMPACT LINES High voltage trasmissio etworks ca operate with directly grouded eutral of power trasformers or the etwork may have a isolated eutral. The coectio of phase coductor with the groud causes a sigle-phase short circuit, whereby the phase voltage icreases i the "healthy" phases of the etwork. Guidace for the determiatio of TOV amplitudes is give i aex A of [7]. rise alog log lie due to Ferrati effect was also cosidered i calculatios, as a cause of TOV. Sequeces of causes for TOV, e.g. load rejectio origiatig from a groud fault, eed cosideratio, whe both overvoltages have comparable severity. I such cases, however, the amout of rejected load depedet o the fault locatio ad the MOSA locatio has to be carefully examied. Combiatio of causes such as groud faults ad load rejectio may result i higher TOV values tha those from the sigle evets. Whe such combiatios are cosidered sufficietly probable, the overvoltages for each cause have to be compouded takig ito accout the actual system cofiguratio. TOV aalyzed i this paper combie the effect of load sheddig, Ferrati effect ad groud fault. Substatios ad are coected with kv compact lie (Fig 7). MOSAs are istalled at both termials of the lie. The overhead lie trasmits P= MW ad Q= MVAR from substatio to substatio. Network with sigle-phase short circuit curret I sc =5 ka was aalyzed [8].

5 Figure 7. Model for calculatio of TOVs The followig evets are cosidered i order to evaluate the level of TOV ad the effect o MOSA: a) A sigle phase to groud fault i phase A occurs at the ed of the lie (substatio ) ad the a three-phase opeig of CB occurs. b) A double-phase to groud fault i phases B ad C occurs at the ed of the lie (substatio ) ad the a three-phase opeig of CB occurs. c) A sigle phase to groud fault i phase A occurs at the etrace of the lie (substatio ) ad the a three-phase opeig of CB occurs. d) A double-phase to groud fault i phases B ad C occurs at the etrace of the lie (substatio ) ad the a threephase opeig of CB occurs. Aalysis of MOSA eergy stress due to time differece betwee opeig of CB ad CB ad failure of relay protectio i substatio is coducted. The km log lie cosidered i this paper is a sigle circuit 5 kv lie equipped with groud wires, upgraded to kv without major modificatios of the desig of the towers. I this case the level of SFO ca exceed the switchig withstad voltage of the isulator strigs. Sequece data of equivalet etwork is show i Table I. TABLE I SEQUENCE DATA OF EQUIVALENT NETWORK Positive Zero I scrms sequece data sequece data R /X X /X (ka) (Ω) (Ω) I scrms (ka) R X R X V. SELECTION OF MOSA RATED VOLTAGE MOSA is a vital piece of equipmet ad a isurace agaist damage to the other equipmet i the substatio. Hece, it is essetial that the MOSA itself be stable uder all system operatig coditios. This, i tur, requires that the system behaviour, especially uder TOV coditios, be kow. Selectig MOSA for a specific applicatio is a compromise betwee protective level, TOV capability ad eergy capability. Icreasig the TOV capability (by additio of blocks i series) icreases the possibility of survival of the MOSA uder system voltage stresses but reduces the margi of protectio provided by the MOSA for a give isulatio level. MOSA with a higher eergy capability reduces the risk of failure. Optimizatio depeds o how well the actual MOSA stresses are kow or ca be estimated. The further steps will explai how to select the parameters of MOSA. The rated voltage U r of the MOSA should be equal to or higher tha the highest equivalet TOV obtaied. Whe protective levels lower tha that of the adopted MOSA desig are desired, rated voltages below the equivalet s TOV may be selected, provided the MOSA is able to absorb the eergy caused by system evets. I this case eergy absorptio calculatios should be carried out simulatig the system evets. Procedure for selectio of protective levels lower tha that of the adopted MOSA desig is show i Figure 8. Calculatios of SFO are ot preseted i this paper, oly TOV are cosidered. Modellig of compact trasmissio lie, surge arresters ad equivalet etworks Three phase load flow calculatios Calculatio of sigle-phase ad double-phase to groud fault TOV calculatios without surge arresters Selectio of surge arrester accordig to IEC 99-5 ad maufacturer guidelies Modellig of surge arresters Calculatio of eergy stress for differet types ad duratios of faults Eergy stress acceptable NO YES Selectio of higher eergy class YES Ed Calculatio of SFO Protectio level acceptable NO Selectio of arrester with lower Ur (lower protective level) Fig. 8. Procedure for selectio of protective levels lower tha that of the adopted MOSA desig For the example metioed above (Fig. 7), the MOSA with U r = kv ad eergy class (eergy capability 9 kj) was selected accordig to [7]. Calculatios of eergy stress were carried out for the MOSA with U r = kv ad for the MOSA with a lower rated voltage U r = kv (eergy capability kj). Static U-I R characteristics show i Table II were used for modellig of MOSAs i EMTP-RV. TABLE II STATIC U-I R CHARACTERISTICS Curret U r= kv U r= kv [A] [kv]

6 VI. SIMULATION RESULTS Three-phase load flow calculatio results are show i Table III i p.u., where p.u.=(/ ) =.99 kv. s at the begiig ad at the ed of the lie before the fault occurrece are determied. TABLE III RESULTS OF THREE-PHASE LOAD FLOW CALCULATIONS s at the begiig of the lie s at the ed of the lie U [p.u.] φ [ ] U a=. U b=.999 U c=.999 U a=.97 U b=.9758 U c=.975 φ a=.57 φ b=-. φ c=.9 φ a=.8 φ b=-.99 φ c=.89 Fig. 9. s at the begiig of the lie durig sigle-phase short circuit The results of three-phase load flow calculatios are used as iput parameters for the calculatio of TOV caused by siglephase ad double-phase to groud fault. Calculated amplitudes of TOV i the trasiet (U max ) ad steady state (U st ) are show i Table IV. Fig.. s at the begiig of the lie after opeig of CB TABLE IV RESULTS OF TOV CALCULATION a) b) c) d) U maxc U stc U maxb U stb U maxa U sta U maxc U stc U maxb U stb U maxa U sta Without MOSAs MOSAs with U r= kv MOSAs with U r= kv Fig.. s at the ed of the lie durig sigle-phase short circuit Fig.. s at the ed of the lie after opeig of CB s are expressed as p.u. values for the followig cases: begiig of the lie durig the fault; begiig of the lie after opeig of CB ; ed of the lie durig the fault; ed of the lie after opeig of CB. Aalyses of maximum TOV values i phases B ad C for differet times of fault occurrece i phase A were coducted. For case c) maximum voltages i phases C ad B are computed i each simulatio, i order to fid the time of sigle-phase to groud fault occurrece at which the overvoltages are the highest. Evet c) represets the most severe evet regardig eergy stressig of MOSAs. Results of TOV calculatios are show i Figs 9 -. Figures, ad 5 show the eergy stress ad the currets through the MOSAs with the rated voltage U r = kv for case c) i substatio. The eergy stressig the MOSAs with the rated voltages kv ad kv versus the duratio of the fault ad the time after the opeig of the CB is show i Table V. Eergy stress ad currets through MOSA at the ed of the lie are higher i the period after opeig of CB tha durig the fault. Fig.. Currets through MOSAs at the ed of the lie durig the fault Fig.. Currets through MOSAs at the ed of the lie durig fault ad after opeig of CB (t ope=. ms) Eergy stress of MOSA i phase C at the ed of the lie icreases rapidly after opeig of CB at t ope =. ms (Fig. 5) due to combied TOV - earth fault ad Ferrati effect.

7 Eergy (MJ) Phase A Phase A Phase C Fig. 5. Eergy stressig the MOSA i phase C at the ed of the lie durig the fault ad after opeig of CB (t ope=. ms) TABLE V RESULTS OF MOSA ENERGY STRESS [KJ/S] VERSUS TIME Eergy U r capability (kv) (kj) a) b) c) d) Begiig of the lie durig the fault Begiig of the lie after opeig CB Ed of the lie durig the fault Ed of the lie after opeig of CB Begiig of the lie durig the fault Begiig of the lie after opeig CB Ed of the lie durig the fault Ed of the lie after opeig of CB The combiatio of maximum allowed duratio of fault versus maximum allowed time after opeig of CB i substatio, for which the MOSA would stad the eergy stress is show i Fig. Fig.. Fault duratio versus time after opeig of CB If the duratio of fault (relay protectio settigs) is kow, from Fig. it is easy to determie whether the eergy capability of the MOSA is exceeded. I the case c) the eergy capability of MOSA with U r = kv at the begiig of the lie will be exceeded if the fault is ot elimiated i both substatios i less tha.7 s (Fig ). If the fault is elimiated i substatio (relayig problem) for example i ms, the eergy capability of MOSA will be exceeded if time after opeig of CB is greater tha ms. The eergy capability of MOSA will be exceeded for the combiatio of times that lie above the curves. VII. CONCLUSIONS This paper describes a procedure for the calculatio of the MOSA eergy stress durig TOV. Laboratory measuremets were coducted ad a MOSA model was developed i EMTP-RV software. Comparisos betwee measured ad calculated MOSA eergy showed a good agreemet. A method for the selectio of protective levels lower tha that of the adopted MOSA desig was described. By usig this method MOSA with low protectio level ca be selected without beig overstressed by TOV, ad, therefore, overvoltage protectio of compact lie ca be improved. Accordig to [7] a MOSA with a rated voltage U r = kv was selected. Calculatio results show that the MOSA with U r = kv could also be selected without beig overstressed by TOV, for certai duratios of TOV. This MOSA has a lower protectio level ad it reduces SFO ad fast-frot overvoltages more efficietly. After the selectio of the MOSA accordig to the show procedure, the study could be performed i order to check if the SFO are reduced to a level acceptable for the compact lie. Aother alterative for the limitatio of SFO is the applicatio of trasmissio lie arresters that ca be located alog the lie at selected poits to obtai a adequate overvoltage profile alog the lie [9] []. VIII. REFERENCES [] A. Legate, J. Bruke, J. Ray, E. Yasuda, Elimiatio of closig resistors o EHV Circuit Breakers, IEEE Tras. Power Delivery, Vol., No., Jauary 988, p.. [] J. Ribeiro, M. McCallum, A applicatio of metal-oxide surge arresters i the elimiatio of eed for closig resistors i EHV breakers, IEEE Tras. Power Delivery, Vol., No., Jauary 989, p. 8. [] H. Zhu, M. R. Raghuveer, Ifluece of represetatio model ad Harmoics o Metal Oxide Surge Arrester Diagostics, IEEE Tras. Power Delivery, Vol., No., October. [] J. Mahseredjia, S. Deetière, L. Dubé, B. Khodabakhchia ad L. Géri-Lajoie: O a ew approach for the simulatio of trasiets i power systems. Electric Power Systems Research, Vol. 77, Issue, September 7, pp [5] J. Mahseredjia, C. Dewhurst, Usig EMTP Tutorials ad Referece, Hydro-Québec/IREQ, 7. [] G. R. S. Lira, D. Ferades Jr. ad E. G. Costa Computatio of Eergy Absorptio ad Residual i Metal Oxide Surge Arrester from Digital Models ad Lab Tests: A Comparative Study, IPST, Lyo, Frace, Jue -7, 7. [7] IEC 99-5: Surge arresters Part 5: Selectio ad applicatio recommedatios, Editio.,. [8] B. Filipović-Grčić, I. Uglešić, A. Xemard, Selectio of statio surge arresters for cotrol of slow-frot overvoltages o compact lies, CIGRE C Colloquium o: Lightig ad Power Systems, Kuala Lumpur, Malaysia,. pp. -. [9] H. Seyedi, M. Saaye-Pasad, M. R. Dadashzadeh, Applicatio of Trasmissio Lie Surge Arresters to Reduce Switchig Overvoltages, IPST, Motreal, Caada o Jue 9, 5, Paper No. IPST5 8. [] L. Steström, M. Mobedjia, Limitatio of switchig overvoltages by use of trasmissio lie surge arresters, ABB Switchgear Swede, SC Iteratioal coferece CIGRE, Zagreb, Croatia 998.