On-line PD Measuring System Modeling and Experimental Verification for Covered-Conductor Overhead Distribution Lines

Transcription

1 Proceedings of the 1th Mediterranean Conference on Contro & Automation, Juy 7-9, 7, Athens - Greece T1-1 On-ine PD Measuring System Modeing and Experimenta Verification for Covered-Conductor Overhead Distribution Lines G. Murtaza Hashmi and Matti Lehtonen Power Systems and High Votage Engineering Laboratory Hesinki University of Technoogy (TKK) 1, Espoo, FINLAND Abstract In this paper, EMTP/ATP simuation environment is used to mode on-ine singe-phase partia discharge (PD) measuring system using Rogowski coi for the monitoring of faing trees on the covered-conductor (CC) overhead distribution ines. The CC is modeed as a distributed parameters ine and Rogowski coi is modeed based on its equivaent circuit as a saturabe current transformer having inear magnetizing characteristics. The simuation resuts are compared with those obtained from the aboratory measurements. The mode can be used to estimate the ength of the CC ine at which PDs due to faing trees can be detected; thus, deciding the number and positioning of the sensors over a particuar ength of the CC overhead distribution ine. I. INTRODUCTION The presence of the partia discharges (PDs) is one of the most prominent indicators of defects and ongoing degradation process of eectrica insuation systems. Therefore, it has been widey recognized as the most effective diagnostic method for on-ine condition assessment. For eectric power distribution industries, continuous monitoring of instaed and operating high votage (HV) apparatus is of particuar importance from safety and reiabiity point of view. A reativey new appication is conducting on-ine high frequency PD measurements for the monitoring of faing trees on the medium votage (MV) covered-conductor (CC) overhead ines [1]. Automatic detection of the faing trees reduces visua inspection work after storms and improves the reiabiity and safety of the system []. Rogowski coi is used as a PD sensor, because it is non-intrusive and provides the needed bandwidth for this appication. Extracting the features of PD from the measurements to detect and ocate such fauts on a compicated transmission ine network is a chaenging task. The chaenge for on-ine PD measurements is to find the optima ocations for these sensors with respect to their sensitivity, interference eve, signa distinction, and universa appicabiity [3]. To work with these design aspects of a sensor, the wave propagation characteristics of the CC ine has aready been determined [4,]. The The first author gratefuy acknowedges and thanks FORTUMIN SÄÄTIÖ to provide partia financia support for this project under Grant Nos and 7-47, for the years 6 and 7, respectivey. measurement of the attenuation can be used as a design aid for the sensor ocation to get the required signa. Therefore, attenuation of the PD puse is an important consideration for deciding the number of sensors and their positioning. In this way, the CC ines wi be more reiabe and the costs reated to visua inspection work wi aso be reduced. The attenuation of the CC ine approximated from the Rogowski coi measurements is much higher than measured using time domain refectometry (TDR) method or cacuated from the theoretica mode of the ine [1,4,]. As the theoretica mode of the CC ine has aready been experimentay verified using TDR measurements [], the higher attenuation in Rogowski coi measurements is an issue that needs a deeper anaysis. To resove this matter, the PD measuring system (incuding CC ine and Rogowski coi) is modeed in EMTP/ATP simuation environment to investigate the source of high attenuation. The ATP simuation resuts are verified by comparing with experimenta resuts, which prove that PD measuring system has successfuy been modeed for the detection of faing trees on the CC overhead distribution ines. The simuation resuts show that CC ine has very ow attenuation, and mosty the PD signa ampitude dies in the over damped resonant circuit of the Rogowski coi. The paper is organized in the foowing pattern. In Section II, the on-ine PD measuring system is depicted. Section III presents the behavior of Rogowski coi for high frequency measurements. Section IV describes the basic principe and construction of Rogowski coi. Equivaent circuit of the Rogowski coi is drawn in Section V. ATP simuation parameters are cacuated in Section VI, whie the simuation resuts are presented in Section VII. The concusions are drawn in Section VIII. II. ON-LINE PD MEASURING SYSTEM The on-ine singe-phase PD measuring system is arranged in a HV engineering aboratory at Hesinki University of Technoogy (TKK), Finand. Fig. 1 depicts the PD measuring set-up. The experimenta system consists of: MV overhead CC ine ( 9m) having poyethyene insuation; fexibe Rogowski coi (without integrator) mounted around the CC ine; puse caibrator; digita oscioscope; capacitor ( pf) for connecting conductor-end to the ground; and the computing system (aptop) for data acquisition from the measurements.

2 Proceedings of the 1th Mediterranean Conference on Contro & Automation, Juy 7-9, 7, Athens - Greece T1-1 Figure 1. On-ine singe-phase PD measuring set-up A caibrator puse U is sent from one end of the conductor and the Rogowski coi measurements are taken at points P 1 and P, at the distances of 6 and 3.7 m from the point of insertion of the caibrator puse, respectivey, as shown in Fig.. The votage puses captured by the Rogowski PD transducer at point P 1 and P for nc caibrator puse are shown in Fig. 3. The decreasing ampitude at point P is the effect of the CC ine attenuation during the propagation of signa. Puse caibrator P 1 P U 6 m 17.7 m. m C Ground Figure.. Singe-ine diagram for on-ine PD measuring system 1 III. ROGOWSKI COIL FOR HIGH FREQUENCY APPLICATIONS The conventiona PD detectors experience severe imitations when it comes to on-ine monitoring due to the infuence of background noise, absence of nonintrusive sensors, and ack of processing faciities. Moreover, they have a imit in the detection frequency range, especiay for CC overhead ines, due to the attenuation of high frequency PD signas. Low detection frequency for the detector imposes a fundamenta imitation on positioning the PD ocation that is one of the major concerns of the insuation monitoring. Recenty, the detection frequency range has been extended up to the radio frequency band with the deveopment of new sensors e.g. Rogowski cois. These cois have the foowing features: (i) There is no conductive couping between the sensor coi and the high votage test circuit. Therefore, it becomes non-intrusive sensor which is a very important aspect for on-site/onine PD monitoring. (ii) It possesses high signa to noise ratio with wide frequency bandwidth. It assures very good inearity due to absence of the magnetic materias. The high frequency behavior of the coi, in particuar (iii) its bandwidth and susceptibiity to high frequency osciations, is significanty infuenced by the termination impedance. There is a trade-off between the bandwidth and the sensitivity of the coi. If the coi geometry is not symmetricay positioned around the CC ine, the experimenta resuts are ogicay infuenced, however, no significant effect has practicay been found in time or frequency domain behavior measurements (see Fig. 4). Ampitude (mv) 1 - Symmetrica geometry Asymmetirca geometry Ampitude (mv) Ampitude (mv) Figure 3. Rogowski coi responses for nc caibrator puse at point P 1, and point P Ampitude FFT Symmetrica geometry Asymmetrica geometry Figure 4. Rogowski coi responses for nc caibrator puse at point P for its different geometries around the CC ine in time domain, and frequency domain; CC ine is in the midde of the Rogwski coi (symmetrica geometry), Rogowski coi is touching the surface of the CC ine (asymmetrica geometry)

3 Proceedings of the 1th Mediterranean Conference on Contro & Automation, Juy 7-9, 7, Athens - Greece T1-1 IV. OPERATING PRINCIPLE OF ROGOWSKI COIL The Rogowski coi operates on the basic principe of the Faraday s aw. The air-cored coi is paced around the conductor, where current puses produced by PDs are to be measured. This variabe current produces a magnetic fied and the rate of change in current induces a votage in the coi given as: di v rc ( t) = M. (1) dt where v rc (t) is the votage induced in the coi by the current i(t) fowing in the conductor due to the mutua inductance M between the main current and the coi, which is independent of the conductor ocation inside the coi oop. To prevent the infuence of nearby conductors carrying high currents, Rogowski coi is designed with two wire oops connected in eectricay opposite directions [6]. This wi cance a eectromagnetic fieds coming from outside the coi oop. The first oop is madeup of turns of the coi, and the other oop can be formed by returning the wire through the centre of the winding as shown in Fig.. Rogowski cois are wound either on a rigid toroida core or a fexibe bet-ike core form. Fexibe Rogowski cois are generay more convenient to use than rigid cois, but are ess accurate. A typica mutua inductance of a standard fexibe Rogowski coi (used in this study) is 3nH and its resonant frequency ies in the high frequency spectrum [7]. V. EQUIVALENT CIRCUIT OF ROGOWSKI COIL In this study, Rogowski coi is used to measure PDs, which typicay ast few nanoseconds, therefore, the high frequency behavior of the Rogowski coi is of paramount importance. Up to now, two different modes of Rogowski coi have been deveoped: the distributed and the umped parameters mode [8-1]. The distributed parameters mode can hep cacuating the coi sensitivity H (V/A) of the Rogowski coi used in ATP simuations; and the transfer function wi be extracted from the umped parameters mode to anayze its bandwidth [11]. A. Dustruibuted Parameters Mode A mode of the distributed parameters to anayze the high frequency behavior of the Rogowski coi has aready been deveoped [1]. In this mode, the system is considered as a distributed ine with per unit ength parameters; resistance R d (Ω/m), inductance L d (H/m), and capacitance C d (F/m). For a Rogowski coi having ength (m), characteristic impedance (Ω), propagation constant γ, and is terminating with an impedance (Ω), the transfer function that reates the induced votage V rc (V) with the output votage per unit ength V outd (V/m) measured at the terminating impedance is given as [11,1]: Figure. Construction of the Rogowski coi where s is the Lapace variabe. and γ are given as: = R ) / sc (3) ( sld + d γ = sc sl + R ) (4) d ( d d The output votage can be expressed as [11]: V outd d = MI () L The coi sensitivity can be cacuated using aforementioned equations as: d Voutd H = (6) I B. Lumped Parameters Mode For simpified anaysis, the behavior of the Rogowski coi with terminating impedance can be represented by its equivaent circuit of the umped parameters as shown in Fig. 6 [11], where R (Ω), L (H), and C (F) are the umped resistance, inductance, and capacitance of the coi, respectivey. The transfer function (V out /V rc ) of the Rogowski coi umped parameters mode (see Fig. 6) can be cacuated as: V V out rc = (7) s L C + s( L + R C ) + ( R + ) sld + Rd V outd =. V rc () γ 1+ e 1+ + γ 1 e Figure 6. Rogowski coi equivaent circuit (umped parameters mode)

4 Proceedings of the 1th Mediterranean Conference on Contro & Automation, Juy 7-9, 7, Athens - Greece T1-1 VI. ATP SIMULATION PARAMETERS CALCULATIONS A. Rogowski Coi Simuation Parameters The geometric characteristics of the circuar cross section Rogowski coi are given in Tabe I. For a toroida coi having a circuar cross section, the umped parameters can be cacuated as foows [11]: R w = ρc (8) πr TABLE I. GEOMETRY OF THE ROGOWSKI COIL Geometrica parameters Inner diameter a Outer diameter b Transducer diameter Specifications 16.4 mm 191 mm d rc 14.3 mm Length of the wire w m Radius of the wire r Length of coi 1 mm 6 mm L µ N d rc b = og (9) π a TABLE II. MEASURED LUMPED MODEL PARAMETERS OF ROGOWSKI COIL Lumped mode parameters Measured vaues C 4π ε + = ( b a) b + a og b a (1) where ρ c is copper resistivity and ε is air permitivity. In order to provide the coi with appropriate damping, the approximate vaue of the can be determined as [8]: π L = (11) C The number of turns of the coi N can be cacuated using the approximate vaue of M as: πm N = (1) b µ d rc og a The measured parameters of the Rogowski coi umped mode are given in Tabe II. The Rogowski coi parameters are measured at a frequency of 1 KHz with the hep of Agient 463B LCR Meter. As the high frequency behavior of Rogowski coi is being investigated, the measured umped parameters are preferred for reiabe simuation resuts. The frequency response of the Rogowski coi can be investigated by drawing bode pots from the transfer function using (7), and is given in Fig. 7. The resonance frequency cacuated using Rogowski coi measured umped parameters mode is 9 MHz. This can aso be verified in Fig. 7, where, the maximum gain is obtained at resonance frequency. The distributed parameters can be cacuated dividing the umped parameters by the ength of the coi. By using the distributed parameters vaues in (6), the sensitivity of the coi can be cacuated. H is taken as.1 (V/A) from the manufacturer's data sheet [13]. In ATP simuations, H and N are used to mode the Rogowski coi as a saturabe current trasformer having inear magnetizing characteritics [14]. These characteristics wi simuate the behavior of an air-cored Rogowski coi. As the vaue of M is not given by the manufacturer, it is assumed to be nh in this study, and N comes out to be 431 using (1). Gain (db) Phase (deg) Resistance R.11 Ω Inductance L.6 µh Capacitance C.3 pf Terminating impedance kω Figure 7: Bode pots for measured umped parameters mode of Rogowski coi B. Covered-Conductor Line Simuation Parameters The CC ine is mounted at an approximate height of 3 m above ground eve in the experimenta set-up. The frequency-dependent CC ine characteristics can be cacuated theoreticay []. As the high frequency PD signas propagation is being investigated, the average vaues of the ine characteristics at MHz frequency range are used. The cacuated CC ine characteristics using theoretica mode are used in simuation as: resistance, Ω/m; characteristic impedance, 3 Ω; and propagation veocity, 3 m/µs. The transmission ines are represented using distributed parameters Carke mode. VII. ATP SIMULATION RESULTS AND DISCUSSION The on-ine singe-phase PD measuring system is drawn using ATPDraw. The ATPDraw is used as a graphica interface and the corresponding network of the PD measuring system incuding Rogowski coi and CC overhead ine is shown in Fig. 8. The measured and the simuated nc caibrator puses are shown in Fig. 9.

5 Proceedings of the 1th Mediterranean Conference on Contro & Automation, Juy 7-9, 7, Athens - Greece T1-1 It has aready been determined that PDs produced due to eaning of a tree on the CC ine are 3-1 nc [1,4]. This amount can vary depending upon the size, weight, and the species of the eaning tree as we as the environmenta conditions prevaied. Therefore, nc caibrator puse is simuated in the PD measuring system to make rea anaysis of an average singe tree eaning on the CC ine. A comparison of the measured and simuated votage puses captured by the Rogowski coi at point P (see Fig. ) is given in Fig. 1. The comparison is carried out considering the time domain performance and fast Fourier transform (FFT) anaysis. From Fig. 1, it is cear that simuated PD measuring system response has a cose match with the measurements, both in the time and frequency domain. It is reveaed that resonance occurs at MHz. Athough, the resonance frequency of the Rogowski coi is 9 MHz, however, due to the effect of cabing, stray inductances are introduced into the system, resuting in the ower vaue of the resonance frequency of the coi (see Fig. 1). The resonance effect can aso be verified by determining the simuated transfer impedance function of the Rogowski coi as show in Figure 11. The measured and the simuated votage puses captured by the Rogowski coi at point P 1 (see Fig. ) are given in time and frequency domain as shown in Fig. 1. The ampitudes of the measured and the simuated puses are not cosey matched in the first few cyces, and the puses are aso distorted in phase. It can be expained on the fact that the resonance frequency of the Rogowski coi has different vaues (in measurements) at different distances aong the CC ine. This can be due to the effect of stray capacitances, introduced during the measurements taken at point P 1, which is nearer to the exposed meta components ying in the aboratory. Another reason to expain the odd response of the Rogowski coi can be due to the unequa cearance of CC above ground in the experimenta set-up, resuting in different vaues of the ine capacitances. The higher attenuation cacuated from the Rogowski coi measurements can be due to the effect of its varying resonance frequencies at different distances from the puse caibrator [1]. On-site PD measuring system can be more reiabe in terms of its fixed geometrica parameters. Ampitude (mv) Ampitude FFT Measured puse Simuated puse FFT of measured puse FFT of simuated puse Figure 1: Rogowski coi response for nc caibrator puse at point P in time domain, and frequency domain Magnitude (Ω) Figure 8. ATPDraw circuit for on-ine singe-phase PD measuring system Ampitude (V) Measured caibrator puse Simuated caibrator puse Figure 9: nc caibrator puses (measured and simuated) Phase (deg) Figure 11: Simuated transfer impedance of the Rogowski coi; magnitude, and phase

6 Proceedings of the 1th Mediterranean Conference on Contro & Automation, Juy 7-9, 7, Athens - Greece T1-1 Ampitude (mv) Ampitude FFT FFT of measured puse FFT of simuated puse Measured puse Simuated puse Figure 1: Rogowski coi response for nc caibrator puse at point P 1 in time domain, and frequency domain The simuated mode can be used in order to estimate the maximum ength of the rea CC ine that can be monitored with a PD sensor. For this purpose, the CC ine characteristics for a rea situation can be cacuated from the theoretica mode, and are used in simuation as: resistance,. Ω/m; characteristic impedance, 46 Ω; and propagation veocity, 9 m/µs []. As the PD source is of nc, which is approximatey equivaent to a tree eaning on the CC ine [1], the maximum distance at which the faing tree can be detected on the CC ine from the point of measurement, can be determined. VIII. CONCLUSIONS ATP simuations are performed to mode Rogowski coi and the resuts confirm that this environment can be used to simuate the transient behavior of Rogowski sensor for on-ine PD measurements. The on-ine singe phase PD measuring system is simuated in ATP. The mode can be used to estimate the ength of the CC ine at which the PDs due to faing trees can be detected; thus, deciding the number and positioning of the sensors over a particuar ength of the CC ine. The simuation resuts show that CC ine has very ow attenuation, and the higher ampitude puse measured by the Rogowski coi near the source of PD is due to the effect of its varying resonance frequencies at different distances aong the ine in experimenta set-up. The Rogowski coi used in this study is not good for high frequency measurements beyond MHz. Therefore, a higher bandwidth Rogowski sensor shoud be used for rea time anaysis of PD signas produced by faing trees. It woud be more practica and demanding if an on-ine three-phase (instead singe-phase) PD measuring system can be modeed in ATP to anayze and resove the rea probems faced by the oca utiities. With those resuts, we woud have the basis to deveop requirements for wireess sensors used in this specific appication. Wireess technoogy is fairy inexpensive and it can be integrated into a modern protection reay and to the distribution automation system for detecting faing trees on CC ines. ACKNOWLEDGMENTS The authors woud ike to thank Professor Akihiro Ametani at Doshisha University, Kyoto, Japan, for giving feedback and vauabe comments on this paper. Discussions with Mr. Nagy Ekaashy and Mr. Abdesaam Ehaffar at TKK to deveop PD measuring system mode are highy appreciated. Many thanks to Hannu Kokkoa for arranging measuring set-up. REFERENCES [1] G. Murtaza Hashmi, Mikae Nordman, and Matti Lehtonen, "A Partia Discharge Detection Concept for Wireess Sensors in Covered-Conductors Distribution System", Proceedings of Europe's premier conference on eectrica insuation (INSUCON 6), Birmingham, UK, May 4-6, 6. [] P. Pakonen and V. Latva-Pukkia, "On-ine Partia Discharge Measurements on Covered Conductor Lines", Proceedings of Nordic and Batic Workshop on Power Systems, Tampere, Finand, February 4-,. [3] P. C. J. M. van der Wieen, J. Veen, and P. A. A. F. Wouters, Evauation of Different Types of Sensors and Their Positioning for On-ine PD Detection and Locaisation in Distribution Cabes, Nordic Insuation Symposium, Tampere, Finand, June 11-13, 3, pp [4] G. Murtaza Hashmi, Mikae Nordman, and Matti Lehtonen, Determination of the Wave Propagation Characteristics for Partia Discharge Monitoring in Covered-Conductor Overhead Distribution Networks, Proceedings of the Modern Eectric Power Systems conference (MEPS6), Wrocaw, Poand, September 6-8, 6. [] G. Murtaza Hashmi and Matti Lehtonen, Covered-Conductor Overhead Distribution Line Modeing and Experimenta Verification for Determining its Line Characteristics, Paper accepted for IEEE PowerAfrica 7 Conference, Johannesburg, SA, Juy 16-, 7. [6] Lj. A. Kojovic,"Rogowski Coi Transient Performance and ATP Simuations for Appications in Protective Reaying", Proceedings of the Internationa Conference on Power Systems Transients (IPST'), Montrea, Canada, June 19-3,. [7] D. A. Ward, and J. La T. Exon,"Using Rogowski cois for transient current measurements" Engineering Science and Education Journa, June [8] W. F. Ray and C. R. Hewson,"High Performance Rogowski Current Transducers", Industry Appications Conference, Rome, Itay, October 8-1,. [9] John D. Ramboz, Machinabe Rogowski Coi, Design, and Caibration, IEEE Transaction on Instrumentation and Measurement, Vo. 4, No., Apri [1] Donad G. Peinen, M. S. Di Capua, S. E. Sampayan, H. Gerbracht, and M. Wang, Rogowski coi for measuring fast, high-eve pused currents, American Institute of Physics, Review of Scientific Instruments, Vo. 1, Issue 11, pp , November 198. [11] Marta Argueso, Guiermo Robes, and Javier Sanz, Impementation of Rogowski coi for the measurement of partia discharges, American Institute of Physics, Review of Scientific Instruments, 76, 617 (June ). [1] J. Cooper, On the High-Frequency Response of a Rogowski coi, Pasma Physics (Journa of the Nucear Energy Part: C), Vo., pp. 8-89, [13] Instruction sheet, FLUKE ifex, Fexibe AC Current Probe. [14] H. W. Domme, Eectromagnetic Transients Program - Rue Book, Oregon, 1984.