Propagation of PD pulses through ring-main-units and substations Wagenaars, P.; Wouters, P.A.A.F.; van der Wielen, P.C.J.M.; Steennis, E.F.

Transcription

1 Prpagatin f PD pulses thrugh ring-main-units and substatins Wagenaars, P.; Wuters, P.A.A.F.; van der Wielen, P.C.J.M.; Steennis, E.F. Published in: Prceedings f the 9th nternatinal Cnference n Prperties and Applicatins f Dielectric Materials (CPADM) 5-9 August 009, Harbin, China DO: CPADM Published: Dcument Versin Publisher s PDF, als knwn as Versin f Recrd (includes final page, issue and vlume numbers) Please check the dcument versin f this publicatin: A submitted manuscript is the authr's versin f the article upn submissin and befre peer-review. There can be imprtant differences between the submitted versin and the fficial published versin f recrd. Peple interested in the research are advised t cntact the authr fr the final versin f the publicatin, r visit the DO t the publisher's website. The final authr versin and the galley prf are versins f the publicatin after peer review. The final published versin features the final layut f the paper including the vlume, issue and page numbers. Link t publicatin Citatin fr published versin (APA): Wagenaars, P., Wuters, P. A. A. F., Wielen, van der, P. C. J. M., & Steennis, F. (009). Prpagatin f PD pulses thrugh ring-main-units and substatins. n Prceedings f the 9th nternatinal Cnference n Prperties and Applicatins f Dielectric Materials (CPADM) 5-9 August 009, Harbin, China (pp ). Piscataway: nstitute f Electrical and Electrnics Engineers (EEE). DO: CPADM General rights Cpyright and mral rights fr the publicatins made accessible in the public prtal are retained by the authrs andr ther cpyright wners and it is a cnditin f accessing publicatins that users recgnise and abide by the legal requirements assciated with these rights. Users may dwnlad and print ne cpy f any publicatin frm the public prtal fr the purpse f private study r research. Yu may nt further distribute the material r use it fr any prfit-making activity r cmmercial gain Yu may freely distribute the URL identifying the publicatin in the public prtal? Take dwn plicy f yu believe that this dcument breaches cpyright please cntact us prviding details, and we will remve access t the wrk immediately and investigate yur claim. Dwnlad date: 05. Jan. 019

2 Prceedings fthe 9th nternatinal Cnference n Prperties and Applicatins fdielectric Materials July 19-3,009, Harbin, China E-10 Prpagatin f PD Pulses Thrugh Ring-Main-Units and Substatins Paul Wagenaars 1 *, Peter A.A.F. Wuters 1, Peter C.J.M. van der Wielen, E.Fred Steennis 1, 1 Eindhven University f Technlgy, P.O. Bx 513, 5600 MB Eindhven, The Netherlands KEMA, P.O. Bx 9035, 600 ET Arnhem, The Netherlands * p.wagenaars@tue.nl Abstract: Online partial discharge (PD) mnitring systems are traditinally installed at a single mediumvltage (MV) cable cnnectin between tw ring-mainunits (RMUs). t is mre efficient t mnitr tw r mre cnsecutive cables using a single mnitring system. Mrever, practical experience with the PD-OL system [1], shws that fr substatins, with many parallel MV cables, and RMUs installing the inductive sensr may be hampered r even impssible. n this paper the influence f RMUs and substatins n the prpagatin f PDs is studied. An RMU r substatin can be mdeled as a cmbinatin f cmplex impedances representing switchgear, transfrmer and MV cables. A PD pulse frm a cable encunters a lad impedance that des nt match the cable's characteristic impedance, resulting in partial reflectin and partial transmissin transmissin t ther cables. Mdels fr RMUs and substatins are prpsed and verified by measurements. Feasible ptins fr nline PD mnitring thrugh RMUs r substatins are determined. Keywrds: partial discharges; pwer cables; diagnstics; ring-main-units; substatins; mdeling NTRODUCTON n recent years, there is an increasing interest in nline mnitring systems that detect and lcate PDs in MV cables. These systems are usually installedn a single cable sectin between tw RMUs. Lcatin f the PD rigin can be achieved by installing a PD measurement unit at bth cable ends and by evaluating the difference in arrival time f the PD pulse at bth units. t wuld save mney and effrt t mnitr tw r mre cnsecutive cables, with ne r mre RMUs alng the cable cnnectin, using nly a single mnitring system (cnsisting f tw measurement units), see Fig. 1. Mrever, practical experience with the PD-OL system [1] with inductive sensrs, shwed that fr large substatins cmprising many cmpnents, and smetimes als fr RMUs, installatin is hampered r even impssible. Sme installatins, fr example, d nt prvide sufficient space fr installing the measurement unit at the desired lcatin at the cable. Mnitring tw cnsecutive cables, at bth sides f the RMUsubstatin, slves this prblem. An RMU r substatin alng the cable cnnectin that is being mnitred affects PD pulses prpagating thrugh it. An RMU r substatin acts as a cmplex impedance cmbining the influence f switchgear, trans- frmer, MV cables, and ther cmpnents such as line reactrs. Therefre, the lad impedance as seen by a PD pulse arriving frm a cable is nt matched t the cable's characteristic impedance. The pulse will partly reflect and partly transfer t utging MV cables, resulting in a distrtin f the pulse shape and amplitude. The significance n the perfrmance f the PD mnitring system is investigated in this paper. First, mdels fr typical RMUs and substatins are develped. Next, the mdels are verified by field measurements. Finally, simulatins using these mdels are perfrmed t investigate the influence f RMUs and substatins n prpagating PD pulses and the resulting influence n the perfrmance f the PD mnitring system. RMU AND SUBSTATON MODEL RMUs and substatins basically have a similar tplgy. There are ne r mre incming MV cables that are cnnected t a cmmn busbar via a switchgear. n additin, ne r mre transfrmers can be cnnected t the busbar. A mdular installatin cnsists f a series f cmpartments. Each cmpartment cnnects a single circuit r transfrmer t the busbar. The main differences between an RMU and a substatin are the number f cnnected cables and the dimensins f each cmpartment. A typical RMU applied in the Dutch grid has 1-5 cnnected cables, while a substatin has 5-30 cables. The width f each cmpartment in an RMU is typically in the range f 10-40em, while in a substatin the width ranges frm cm. The mdel presented in this sectin is based n a mre detailed mdel presented in [] and is adjusted s that it can be applied t bth RMUs and substatins. Als sme elements f the riginal mdel that have hardly influence in the frequency range 100kHz-5MHz are remved. Tw r mre cnsecutive cables, as cnsidered here, usually have a ttal length exceeding a few hundred meters. PD signals after having traveled this distance will hardly have energy abve 5 MHz. n Fig. the equivalent circuit f an RMUsubstatin is depicted. Each cmpartment has a lad impedance ZL in series with inductance L s. The impedance ZL represents the cmpnent that is cnnected t that cmpartment, usually an MV pwer cable r a transfrmer. The inductance l-«is the inductance f the lp frm the cnnected cmpnent t the busbar. The inductance Lbb is the inductancebetweenbusbar and earth ver the distance f the width f ne cmpartment $ EEE 441 Authrized licensed use limited t: Eindhven University f Technlgy. Dwnladed n April 3,010 at 1:00: UTC frm EEE Xplre. Restrictins apply.

3 ,... C'\' ~...!-~ ~ ~ -=i=-_ ~ '~ ]~ ~: "';""f- MVLV transf. ' * RMU MV cable inductive PD sensr -~~) : Cr- -) MV cable inductive i'lf]~ 'f ~ 1 MVLV transf. PD sensr MVLV transf. RMU * * RMU circuit under test Fig. 1: Setup mnitring tw cnsecutive cables with ne RMU alng the cable under test. cntains transfrmers and five cable cnnectins. Ring-main-unit measurements Fig. : Equivalent circuit f RMU r substatin with N cmpartments Cmp Cmp. Cmpartment 3 r-----y., t; Lbb t; \ : Rtf PLEC <. C tcc C" \ Earth TCC ~, _... Transf. cnnecting cables Transfrmer Fig. 3: Equivalent f RMU with tw MY cables and MYLY transfrmer n Fig. 3 the lad impedance ZL in each cmpartment is replaced by equivalent circuits f the cnnected cmpnents fr an RMU with three cmpartments (tw MY cables and an MYLY transfrmer). The MY cables are represented by their characteristic impedance Zc. The transfrmer is mdeled by the capacitance C tr between transfrmer windings and grunded cre and casing, inductance L tr f the windings and the lp f the cnnecting cables, and resistance R tr representing lsses in the transfrmer. The transfrmer cnnectin cables (TCC) that cnnect the transfrmer t the busbar are mdeled by the capacitance C tcc frm cable cnductr t earth screen, inductance 4cc f lp and earth cnnectin, and resistance R tcc representing lsses. MEASUREMENTS The prpsed mdel is verified by measurements n an RMU and a substatin. The RMU invlves three cmpartments, tw cable and a transfrmer. The substatin T be able t study the effect f an RMU r substatin n the prpagatin f PO pulses typical values fr the cmpnents in the presented mdel must be knwn. The mdel parameters were determined using measurements fr several RMUs. A pulse is injected inductively at ne cable end. n the RMU at the far end the resulting wavefrm is measured at three lcatins: arund the incming and utging cables (PLEC and PLEC), and arund the cmmn earth cnnectin f the transfrmer cnnecting cables (earth TCC). These lcatin are indicated in Fig. 3. Tw transfer functins are calculated: and H ( ) _ ltcc(w) tee W - lcl(w) where lcl is the current measured at PLEC (incming cable), c the current measured at PLEC (utging cable), ltcc the currentat earth TCe. H c is the transfer functin frm the incming t the utging cable, and H tcc the transfer functin frm the incming cable t the earth f the transfrmer cnnecting cables. An example f the measured H c and H tcc fr ne RMU is pltted in Fig. 4. The transfer functins H c and H tcc can ben expressed in terms f the mdel parameters in Fig. 3. The parameter values fr the mdel are fund by a fitting prcedure that minimizes the mean abslute relative errr between the measured transfer functins and the mdeled transfer functins. The mdel has nine parameters and therefre als lcal minima may exist. n rder t cnverge t the glbal minimum the starting values must be chsen accurately. Often, ne r tw resnances can be bserved, as is the case in Fig. 4 near MHz and 3 MHz. The prducts 4rCtr and LtccCtcc are chsen such that they match these frequencies. The capacitance f the transfrmer cnnectin cables can be estimated by multiplying the length with the capacitance value taken frm the cable datasheet (~ 140nF1m). Earlier impedance measurements [) shwed that the typical characteristic impedance (Zc) f a three-cre JOkY PLC cable is apprximately JOO and that the ttal inductance f the lp between tw installed MY cables is apprximately 00nH. The ttal inductance f the lp between the tw MY cables in the mdel in Fig. 3 is L s + Lbb. Because (1) 44 Authrized licensed use limited t: Eindhven University f Technlgy. Dwnladed n April 3,010 at 1:00: UTC frm EEE Xplre. Restrictins apply.

4 :c: t --- Measur ed H 46 '0 g.... ~":-': -~ -' ---: Mdeled H 46 - t Frequency (MHz) -t 4 6 Frequency (MH z) 10 Fig. 4: Measured (slid) and mdeled (dtted) RMU transfer functins H e (black) and H lee (grey). Fig. 5: Measured and mdeled rati f currents measured in cmpartment 4 and 6. the field measurements arc perfrmed n similar PLC cables and the same type f installatin the cnstraints L s + Lbb = 00 nh and Ze = 100 are kept fixed in the fitting prcedure. The mdeled transfer functins after the fitting prcedure are included in Fig. 4. The mdel parameters fr this RMU arc: L l r = 3..uH, C lr = 1.6nF, s; = 5.60, 4ee = 0.74.uH, C lee =.3nF, Rce =.40, L s = 345nH and Lbb = 110nH. Substatin measurement A measurement has been perfrmed t verify the mdel f Fig.. The measurement was perfrmed in a cmpact substatin with five cnnected MV cables and tw transfrmers. Each cmpartment is 4 x 10 x 70cm (WxHxD). The transfrmers are cnnected t cmpartments 1 and, and the cables t cmpartments 3-7. Fr the measurement a pulse was injected inductively arund the leftmst MV cable (in cmpartment 3). The injected current distributes ver the ther cmpartments. The injected current and the currents (4 t 16)thrugh the cables in cmpartment 4-6 are measured and the current transfer functins are calculated. Fr instance, in Fig. 5 the rati H46 = is pltted. Additinally, at each cable an impedance measurement is perfrmed t determine the impedance f that cmpartment (jwl s + Ze) in series with the rest f the substatin. The mdel parameters arc fitted by minimizing the mean abslute relative errr between mdel and measurement: L s = 50 nh, Lbb = 140nH and Ze = 0. At lw frequencies the influence f the inductances are negligible and the rati H 4 6 is determined by the characteristic cable impedances. Because Ze is equal fr all the cnnected cables the rati H46 pltted in Fig. 5 starts at 1. This means that the current injected arund the cable in cmpartment 3 distributes equally ver the ther fur cables. At higher frequencies the current distributin is mainly determined by the rati f L s and Lbb. EFFECT ON PD MONTORNG The prpsed mdels fr RMUs and substatins allw us t predict their effect n the PD wavefrm. The signal distrtin and the effect n lcatin accuracy is studied. Effect f ring-main-unit The effect f an RMU can be expressed in the ttal transfer functin H rrnu. This transfer functin is a cmbinatin f the transmissin cefficient Tel frm cable 1 t the RMU, and the transfer functin H e : Z e Hrrnu = Tel ' H e =.H e () z; + Z lad where Zlad is the RMU impedance as seen by a pulse arriving frm cable 1. H rrnu is pltted fr a typical RMU in Fig. 6. The parameter values fr this simulated RMU were btained by averaging the fitted parameters f measurements in five RMUs: L s = 340nH, Lbb = 10nH, 4r =.6.uH, = 10, u; = 1..uH, C lee = 1.9nF, C lr =.5nF, R lr Rce =.60, Ze = 100. The figure shws that fr frequencies up t 1.5MHz PDs pass thrugh the RMU almst unaffected. Because higher frequencies attenuate strnger than lwer frequencies the effect f an RMU in the cable under test is larger fr shrter cables than fr lnger cables. n rder t determine the effect f an RMU n the lcatin accuracy a simulatin has been perfrmed with a circuit with a ttal length f 1km and an RMUs at 400m and at 900m. The RMUs alng the cable have a ttal transfer functin H rrnu as depicted with the slid black line in Fig. 6. At its rigin a PD pulse is simulated by a delta pulse. The prpagatincefficient taken t simulate the PD prpagatin thrugh the cable has been measured n PLC cable sample. The lcatin accuracy is investigated by simulating PD measurements, as described in [3]. This simulatin cnsists f a prpagatin time measurement fllwed by PD measurements fr severallca- 443 Authrized licensed use limited t: Eindhven University f Technlgy. Dwnladed n April 3,010 at 1:00: UTC frm EEE Xplre. Restrictins apply.

5 H nnu ~..ġ Q) c:.~ g..j rr -rr Frequency (MH z) --- H suhs,56... subs,5 '-. "., R ~ "., ,.... "., Fig. 6: Ttal transmissin cefficient f RMU (black) and substatin (grey) and refiectin cefficient f substatin (dashed grey) tins alng the cable. The precise value f the arrival time f (PD) pulses is ambiguus and depends n the pulse detectin algrithm. A rbust methd, based n the signal energy criterin [3], is emplyed here. The "measured" PD lcatin is cmpared t the actual PD lcatin. n Fig. 7 the simulated lcatin accuracy is pltted. As a reference simulatin results f the lcatin accuracy fr the same circuit withut RMUs alng the cable under test is included. The intrductin f the RMUs alng the cable under test clearly intrduces a lcatin errr. The maximum errr is apprximately 0.4% f the ttal cable length. Fr shrter cables this relative errr will increase while fr lnger cables it will decrease. Effect f substatin The mdel has als been applied t simulate the effect f a substatin n PD pulse prpagatin. A substatin with 15 cnnected MY cables has been mdeled using L s =.uh, Lbb = 300nH and Zc = Q. These values are larger than fund fr the measurement in the previus sectin because the average substatin installatin has larger dimensins than the installatin f the measurement. A pulse frm cable 5 is transmitted t cable 6. The ttal transfer functin t the neighbring cable H su bs,56 is pltted in Fig. 6. The substatin transfer functin is smaller than H nnu ver the full frequency range. The transfer functins t cables at larger distance frm cable 5 (nt shwn) are even smaller. A substatin alng the cable cnnectin results in a large decrease in detectin sensitivity. Fr a substatin with many cnnected cables an alternative ptin fr PD lcatin can be cnsidered. A single-sided measurement, based n time-dmain reflectmetry, with the substatin at the far cable end is feasible. The lad impedance f the substatin is much lwer than Zc f the cable fr frequencies up t rughly 1MHz due t the many parallel cables. Fr higher frequency the lad impedance is much higher due t the relatively large O -0.5 '-- -'-- --'- ---L '--_---' PD rigin (m) Fig. 7: PD lcatin errr fr different PD rigins with tw RMUs alng the cable under test. inductances L s and Lbb. Because f this clear mismatch PD pulse reflectin is guaranteed. This is illustrated by the reflectin cefficient R su bs,5 depicted in Fig. 6. R su bs,5 is even larger than the RMU transfer functin fr almst the cmplete frequency range. CONCLUSONS Mnitring several cnsecutive cables with a single PD system is feasible, prvided that there are nly RMUs alng the cable under test. An RMU alng the cable under test des intrduce a lcatin errr, but fr mst cable cnnectins this errr is within a usually accepted range f 1% fr PD lcatin in pwer cables. Only fr shrt cables the errr will be beynd this limit. A substatin alng the cable under test results in a large decrease in sensitivity and is therefre nt recmmended. An alternative is t perfrm single-sided PD measurements, using time-dmain refiectmetry fr lcatin, with the substatin situated at the far end f the cable under test. The RMU and substatin mdels can be used t further study their effect n PD diagnstics when placed alng the cable under test r at the far end. Future research will be directed t the effect f RMUs and substatins n parameters imprtant fr PD diagnstics, such as the detectin sensitivity and the charge estimatin. REFERENCES [] P. C. J. M. van der Wielen and E. F. Steennis. Experiences with cntinuus cnditin mnitring f inservice mv cable cnnectins. n Prc. Pwer Engineering Sciety (PES) Pwer Systems Cn! & Exp. (PSCE), Seattle, WA, USA, Mar [) P.C.J.M. van der Wielen. On-line Detectin and Lcatin f Partial Discharges in Medium-Vltage Pwer Cables. PhD thesis, Eindhven University f Technlgy, Eindhven, The Netherlands, 005. [3] P. Wagenaars, P.A.A.F. Wuters, P.C.J.M. van der Wielcn, and E.F. Steennis. Accurate estimatin f the time-f-arrival f partial discharge pulses in cable systems in service. EEE Trans. Dielectr. Electr. nsul., 15(4): ,Aug Authrized licensed use limited t: Eindhven University f Technlgy. Dwnladed n April 3,010 at 1:00: UTC frm EEE Xplre. Restrictins apply.