On-line partial discharge monitoring of statorwindings : comparison of different sensors Pemen, A.J.M.; Leeuw, de, W.; van der Laan, P.C.T.

Transcription

1 On-line partial discharge monitoring of statorwindings : comparison of different sensors Pemen, A.J.M.; Leeuw, de, W.; van der Laan, P.C.T. Published in: Proceedings of the 1th International Symposium on High Voltage Engineering, volume 4, Montreal, Quebec, Canada, August 1997 Published: 1/1/1997 Document Version Publisher s PDF, also known as Version of Record (includes final page, issue and volume numbers) Please check the document version of this publication: A submitted manuscript is the author's version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website. The final author version and the galley proof are versions of the publication after peer review. The final published version features the final layout of the paper including the volume, issue and page numbers. Link to publication General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. Users may download and print one copy of any publication from the public portal for the purpose of private study or research. You may not further distribute the material or use it for any profit-making activity or commercial gain You may freely distribute the URL identifying the publication in the public portal? Take down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. Download date: 2. Aug. 218

2 ISH '97

3 describe the measurement technique, by means of an example of a measurement on a 65 MW turbine generator. 2. Measurement technique Unfortunately, for the monitoring of pd's during regular operation, standard pd-detectors can not be used. The main reason for this is the presence of many sources of EM-disturbances in a power plant. Standard pd-detectors operate in a frequency range < 1 MHz. Within this range large interference is caused by the rotor excitation. In [1] we show that this problem can be solved by using a tunable narrow-bandwidth filter as a front-end for a standard detector. The filter is tuned to a frequency where partial discharges dominate and no interference is present. Our experience is that the best results are obtained at tuning-frequencies between 1 and 3 MHz. As a tunable narrow-bandwidth filter we use a spectrumanalyser in the "zero-span" mode. The output of the spectrumanalyser is connected to a modem pd-detector (Haefely TE571). This detector provides a digital registration of all pd quantities, such as the discharge magnitude, the time of occurrence as related to the 5 Hz voltage, the instantaneous inception voltage, the discharge energy and the quadratic rate. The results can be shown in many ways, as 2 and 3D patterns or pulse-height histograms. Also the detector has built in tools for recognition and classification of pd patterns [2]. Figure 1 shows a result measured in phase " 1" of the 65 MW generator "Hemweg-8". During the measurement the generator was operating at full load. The spectrum analyser is tuned to 2 MHz. The figure shows the phase resolved distributions HQmax (the maximum discharge magitude), HQn (the mean discharge magnitude) and H,. (the number of discharges). After calculating and storing a 'fingerprint' of 29 statistical parameters, the measurement can be classified. This means that the result is correlated with a database, which can contain fingerprints of: (i) earlier measurements on the same generator, (ii) measurements on similar generators, (iii) known defects and (iv) known ageing states. Figure 2 gives the classification of the measurement in Fig.I ("KWARTAAL 4/FASE 1") with reference to an earlier measurement on the same generator ("KWARTAAL 3/FASE 1 "). The correlation between the two measurements is nearly 1 %, which means that the discharge pattern has not changed. 3. Comparison of different sensors An important part of a pd measuring system is the sensor. For the measurements in the previous Section we used Rogowski-coils, installed around the high-voltage terminals of the generator. Besides Rogowski-coils, we developed capacitive sensors that are housed inside a standard support-insulator of the isolate-12hase-hus (ipb) (Fig.3). Both sensors are described in [4] and have a good sensitivity, bandwidth, EMC-properties and can easily be installed. Recently, we installed capacitive sensors in the new 165 MW turbine generator "Roca-3 ". For each phase, two sensors are installed, one at the generator side and one at the step-up transformer side. Besides the capacitive sensors, in each phase two other capacitors were available: "ring" capacitors and overvoltage capacitors. The "ring" capacitors are shown in Fig.3c. A metal ring (35 cm. wide) is installed inside the ipb. The ring is isolated from the metal ipb and connected to a N-connector. The overvoltage capacitors are installed inside the enclosure of the generator-circuit-breaker (Fig.4a). The equivalent circuit for all three sensor types is given in Fig.4b. The capacitance values are given in Table 1. Table 1 values for the three sensor types sensor Cl C2 R capacitive sensor 25 pf 1 pf 5 ring capacitor 16 pf 182 pf 5 overvoltage capacitor 132 nf 8 µf 5 (s) (b) (c) Fig.3 (a) Isolated phase-bus (ipb), with (b) a capacitive sensor inside a support insulator, and (c) a nng -capacitor inside the ipb (a) Enclosure of the generator-circuit-breaker Overvoltage capacitor (b) j_ :~tg Fig.4 (a) Overvoltage capacitor house inside the coaxial enclosure of the generator-circuit-breaker, and (b) equivalent circuit for capacitive sensors

4 Figure 5 gives the frequency spectra as recorded in phase W, with a capacitive sensor, a ring capacitor and a overvoltage capacitor. It is clear that the capacitive sensor has the best sensitivity, while the sensitivity of the overvoltage capacitor is the lowest. The spectra show several peaks caused by partial discharges. These peaks are a result of resonances in the statorwinding or reflections of the signal in the ipb. Preferably, for discharge measurements the spectrum analyser is tuned to one of these peaks and an important question is to which peak. To find the best tuning frequency, for each sensor a serie of zero-span measurements is done. Measurements are done at 3 different tuning frequencies (from 1 to 7 MHz, with an interval of 2 MHz) and the signal level at these tuning frequencies is recorded during several cycles of the 5 Hz voltage. The best results for phase U are given in Fig.6. For all sensors the best tuning frequency is 1 or 12 MHz. At these frequencies, the sensors have the largest sensitivity and no interference is recorded. As can be seen, the capacitive sensor at the generator side Cg has the largest sensitivity, but also the other sensor types give good results. -5 E' Cx:i ~-6 l S? -7-8 : ;~ - - capacitive sensor -- 'ring' sensor ovsrvoltage capacitor frequency (MHz) Fig.5 Frequency spectra as measured in phase W of generator #Roca-3 #, by means of J different sensors 6 G ~ -2 (.) 2 ~ -2 toot -1~=[ =======~. _.,,...,.,., 2 4 time (ns) Fig.7 Waveshapes of a partial discharge in phase V of generator # Roca-3 #' measured simultaneous with 4 different sensors Figure 7 gives the waveshapes of a partial discharge in phase V, measured simultaneously with all 4 sensors in this phase. From the transit times.1tl and.1t2 it is seen that this signal comes from the generator. In case of a discharge in the step-up transformer, the signal would first show up at sensor Ct. So installing 2 or more sensors in one phase enables the localisation of a discharge. The transit time.1tl is 42 ns (or 12.7 m) and corresponds with the distance between the capacitive sensor at the generator side Cg and the overvoltage capacitor OV-C. The transit time.1t2 is 44 ns (or 13.3 m) and corresponds with the distance between the overvoltage capacitor OV-C and the capacitive sensor at the step-up transformer side Ct. The distance between the ring capacitor and the capacitive sensor at the generator side Cg is 3 cm, and the corresponding transit time of 1 ns cannot be seen in Fig. 7. As with the spectrumanalyser measurements, the capacitive sensor at the generator side Cg gives the largest signal. This signal of about 25 m V can easily be detected with a digital oscilloscope. However, the signals from the ring capacitor and the overvoltage capacitor have an amplitude of about 2 m V and already show bit-noise. These amplitudes are close to the limit of what can be detected with an oscilloscope. This shows a great advantage of the spectrumanalyser: a much larger sensitivity so that small signals can be detected reliably. T 4. Discussion The results in the previous section show that, although the capacitive sensors have the largest sensitivity, all three sensor types can be used for on-line partial discharge measurements. The measurements can be done with a spectrumanalyser tuned to a frequency where partial discharges

5 dominate and less interference is present. This interference has two ways of distorting a measurement: (i) by coupling into the measuring system, and (ii) by coupling into the statorwinding. Coupling into the measuring system must be avoided by state of the art EMC measures [7]. For the measurements in the previous section, all sensor types have good EMC-properties and are shielded by the ipb or by the enclosure of the generator-circuit-breaker. All connectors and coax-cables have a low transferimpedance and the recording equipment is installed in an EMC-cabinet. Coupling of interference into the statorwinding causes 'legitimate' signals at the sensors, but can be rejected by tuning to the correct frequency. This tuning technique is not effective when interference couples into the measuring system, as we noticed during measurements on turbine generator "Claus-B". Here the overvoltage capacitors are installed in cabinets outside the ipb, and have long leads to the ipb and the grounding. system of the plant. A magnetic flux <I> can couple into the large loop (about 6 m 2 ) formed. This magnetic flux <I> can be a restilt from any source and results in an interference voltage v. = -d<i>/dt. Figure 8 gives the results for measurements in phase U of generator "Claus-B", for tuning frequencies of 1, 2 and 3 MHz. At all frequencies (also at 12 other tuning frequencies between 5 and 4 MHz) interference is recorded. However, off-line measurements show large discharge levels upto 5 nc. The discharges are caused by cracks in the endwindings. The cracks are a result of vibrations of the end-windings due to loose blockings. After consulting the manufacturer it was decided to rewind the generator. Surprisingly, the on-line measurements gave no indications for the presence of large partial discharges, since interference dominated the measurements. This proves that it is very important to suppress interference as much as possible, preferably by good EMC-measures. The use of a tunable narrow-bandwidth filter (or signal processing) in itself are not enough to reject interference and can only be used for further improvement. 5. Conclusions Several sensor types (capacitive sensors, overvoltage capacitors and "ring" capacitors) were used for partial discharge measurements on a 165 MW turbine generator. The discharges were recorded with a standard pd-detector, preceeded by a tunable narrow-bandwidth filter. The filter is tuned to a frequency where discharges dominate and less interference is present. All three sensor types give good results at tuning frequencies of 1 or 12 MHz; the best sensitivity is obtained with capacitive sensors. Good EMC-measures are of major importance for online pd-measurements. On-line measurements on poorly installed (from an EMC point of view) overvoltage capacitors of a 58 MW generator only showed interference, while off-line measurements showed large discharges of >5 nc time (ms) Fig.8 On-line measurement with a over-voltage capacitor, installed outside the isolated-phase-bus of the 58 MW turbine generator "Claus-B" Acknowledgements: this research was undertaken at request of the Dutch electric supply companies. The authors gratefully acknowledge the cooperation of EPON, EPZ, EZH and UNA. The contribution of the working group "Generator railsystemen'' is very much appreciated. References (1) A.J.M. Pemen, P.C.T. van der Laan: "Pitfalls of pd measurements on statorwindings of turbine generators", 7th DMMA, Bath, UK, September 1996, pp (2) E. Guiski: "Digital analysis of partial discharges", IEEE Trans.on Dielectrics and Electrical Insulation, Vol.2, No.5, October 1995, pp (3] H.G. Sedding, S.R. Campbell. G.C. Stone, G.S. Klempner: "A new on-line partial discharge test for turbine generators", IEEE Trans.on Energy Conversion, Vol.6, No.I, December 1991, pp [4] A.J.M. Pemen, P.C.T. van der Laan, P.T.M. Vaessen: "Sensors for pd monitoring of turbo generators", 28th UPEC, Staffordshire, UK, September 1993, pp [5] J.E. Timperley: "Incipient fault identification through neutral rf monitoring of large rotating machines", IEEE Trans.PAS, Vol.PAS-12, No.3, March 1983, pp (6) P. Grunewald, J. Weidner: "Possibilities and experience with off- and on-line diagnosis of turbine generator stator winding insulation", 1994 CIGRE Session, paper 11-26, Paris, 1994 [7] P.C.T. van der Laan, A.P.J. van Deursen: "Protection of electronics in high power installations; theory, guidelines and demonstrations", CIGRE Symposium on EMC, Lausanne 1993, paper 6-8 Adresses of authors KEMA Transmission & Distribution P.O. Box 935, 68 ET Arnhem, The Netherlands phone fax Eindhoven Univ. of Technology, High-Voltage & EMC group P.O. Box 513, 56 MB Eindhoven, The Netherlands phone , fax