TRANSMISSION AND DISTRIBUTION Development of standing waves on long distance HVDC transmission lines by B R Duncan and J M Van Coller, University of the Witwatersrand The Cahora Bassa HVDC scheme in Southern Africa can transmit 1920 MW from Mozambique to South Africa. The DC line is experiencing numerous insulator flashovers. This paper describes an investigation into the possible existence of standing waves on the DC line. The simulation study used the Cigré benchmark model for HVDC studies, adapted to model the Cahora Bassa HVDC scheme. Steady state and transient studies were performed investigating the development of standing waves on the DC line. The results show that under certain operating conditions it is possible to have a standing wave present on the DC line, which could increase the number of insulator flashovers, thereby reducing the overall performance of the DC line. HVDC, composite resonance, standing waves Cahora Bassa The continued growth of economies has led to an increased demand for reliable electrical power [1]. Dwindling resources near load centres increases the need to access remote resources and transmit electrical power over longer distances. This has led to the use of high voltage direct current (HVDC) for bulk power transmission [2]. There are currently two operational HVDC schemes in Africa, namely the Cahora Bassa scheme between Mozambique and South Africa, and the Inga-Shaba scheme in the Democratic Republic of Congo (DRC) [3]. The reliability of an HVDC scheme will be reduced if there are factors that increase the number of DC line insulator flashovers. One such factor is where the voltage along the line is increased by the presence of standing waves, and this paper describes a simulation study to investigate this possible problem. Standing waves on transmission lines For a transmission line, a standing wave is formed by the superposition of two waves travelling in opposite directions [8]. This is a result of reflections at the transmission line ends due to impedance mismatches. The wave appears to be standing still and is characterised by a series of nodes and anti-nodes (maxima and minima) at fixed positions along the transmission line. Designing long AC overhead transmission lines is not as simple as designing shorter lines [7] and under certain conditions, reflections can cause standing waves to develop. This problem is solved by breaking lines up into shorter segments with large loads or generation at nodes, to ensure that the length of segments is kept well below /4 [5]. This is not always possible with DC lines since most HVDC Fig. 1: Impedance at the inverter AC bus. a) Cahora-Bassa model. b) Cigré benchmark model. Fig. 2: Impedance at the rectifier AC bus. a) Cahora Bassa model. b) Cigré benchmark model. schemes are point-to-point transmission systems and generally do not have tapping points along the line. Standing waves should not generally be a problem on DC lines. This is true for pure DC conditions, but since conversion between DC and AC takes place at both ends of the line, characteristic and noncharacteristic ripple frequencies can lead to the development of standing waves [5]. These are normally small in magnitude and do not create serious problems, unless the length of the line corresponds to a multiple of the half wavelength. Resonances associated with HVDC schemes In HVDC schemes, resonances are present on both the AC sides and DC sides of the converters. Parallel resonances are seen as high impedances and series resonances are seen as low impedances. These AC and DC resonances are interconnected energize - June 2010 - Page 50 via the converters. This has led to the concept of the composite resonance [6], a resonance that arises when a parallel resonance on the AC side is coupled with a series resonance on the DC side. Instabilities occur when the overall resistance of the resonant circuit is negative at the composite resonant frequency. The capacitive nature of the reactive compensation required in HVDC schemes together with the inductive nature of the connected AC system produces a parallel resonance. The impedance measurement tool in PSCAD can perform an impedance scan of a system over a specified frequency range, providing insight into the actual resonant frequencies. The impedance is determined assuming that all transformers are unsaturated and that all power electronic devices are in their “off ” state [9]. This means that there is no access to impedances of HVDC systems under actual operating conditions and TRANSMISSION AND DISTRIBUTION hence time domain simulations are required. Modifications to the Cigré benchmark model The Cahora Bassa model was based on the Cigré benchmark model [10] implemented using PSCAD. The AC filters at the Songo end were modelled in the same manner as at the Apollo end, changed only to accommodate the different voltages and reactive power compensation. The DC filters and smoothing reactors were also modified to match the equipment installed on the Cahora Bassa DC lines. The Songo end converter rating was changed from 1000 MVA, 345 kV to 2000 MVA, 220 kV. The Apollo end converter rating was changed from 1000 MVA, 345 kV to 2000 MVA, 275 kV. Fig. 3: Impedances of the 1400 km and 3000 km DC lines and the Cigré DC line. The scheme topology was changed from 12-pulse monopole to 24-pulse bipole. The controllers used for each pole were those of the Cigré benchmark model. The converter transformers were modified to the appropriate voltage and MVA ratings and the transformer impedances were dropped from 0,18 to 0,14 pu to more accurately represent the transformers used in the Cahora Bassa HVDC scheme [11]. The DC line model was changed from a single nominal-pi model to nine identical distributed parameter model sections with the same line geometry as for the Cahora Bassa DC line. This allowed for the monitoring of the voltage at multiple points along the line – thus making it possible to derive a voltage profile along the transmission line. Fourier analysis was used to determine the amplitudes of the ripple frequency components along the DC line. Simulation results – resonances Steady state simulations were performed for various line lengths and with various rectifier constant current-control (CCC) gains. The rectifier CCC gain has a significant effect on composite resonances since the rectifier controller can present a negative resistance to the resonant circuit [6]. Varying the length of the DC line altered the series resonance on the DC side, thereby also affecting the composite resonance. The AC and DC side impedances were measured, using PSCAD, to verify the composite resonance condition. Figs. 1, 2 and 3 are impedance plots of the Cahora Bassa model for comparison with the Cigré benchmark model. Fig. 1a shows the AC bus impedance, as a function of frequency, at the inverter end of the Cahora Bassa model (Apollo), and is compared with the inverter end of the Cigré benchmark model in Fig. 1b. It is clear that both systems have a parallel resonance near the second harmonic (100 Hz). Fig. 2a shows the AC bus impedance, as Fig. 4 (a+b) : Voltage profile along the 1400 km transmission line – low rectifier CCC gain a) Harmonic components. b) Voltage profile along the line. a function of frequency, at the rectifier end of the Cahora Bassa model (Songo), and is compared with the rectifier end of the Cigré benchmark model in Fig. 2b. Again it is clear that both systems have a parallel resonance near the second harmonic (100 Hz). Fig. 3 is an impedance plot of the DC lines that were modeled – the 1400 km line based on the Cahora Bassa DC line geometry, and an extended-length version of the same line geometr y for easy comparison. The extended-length line was chosen to be 3000 km since this is approximately the length for the Westcor project, which would transmit power from Inga in the Democratic Republic of Congo, to Cape Town in South Africa [4]. graphs. The bottom left plot displays the harmonic content close to the rectifier end (Songo). The bottom right plot shows the harmonic content at the inverter end (Apollo). The base frequency was 5 Hz and 31 harmonic amplitudes were calculated. Fig. 4 shows the voltage profile along the 1400 km line operated at a low rectifier CCC gain of 0,5. As can be seen in Fig. 4a, the harmonic content, although small, is greater near the centre of the DC line. Fig. 4b shows the voltage profile along the DC line with the measured ripple magnitude superimposed. Simulation results – standing waves Fig. 5 is the voltage ripple amplitude profile of the same system operated at a higher rectifier CCC gain of 0,9. It is clear that the composite resonance level is greater at higher rectifier CCC gains. The composite frequency harmonic magnitude also increases with time as seen when comparing Fig. 5a to Fig. 5b. It is important to note that the center of the DC line appears to have significantly higher ripple amplitude than the ends of the DC line. The voltage profile along the line, in Fig. 5c, verifies that near the ends of the line, the peak-to-peak standing wave voltage is less than at the centre of the line. The harmonic content of the voltage waveform at various positions along the DC line is shown using a set of ten bar This is ver y similar to a half-wavelength standing wave at the composite resonance frequency of approximately 65 Hz, with a Similar to the Cigré benchmark model [10], both the 1400 km line and the 3000 km line have a series resonance near the fundamental frequency (50 Hz). Since these resonant frequencies are both similar to that of the Cigré benchmark model, we would expect that under certain conditions a composite resonance at approximately 70 Hz could be excited [6]. energize - June 2010 - Page 52 TRANSMISSION AND DISTRIBUTION rectifier, rather than at the standing wave maxima. The peak voltage on the line was measured to be 0,072 pu greater than the DC voltage at the rectifier. Conclusions Fig. 5 (a+b): Voltage profile along the 1400 km line (rectifier CCC gain of 0,9). a) Harmonic components, T = 1,5 s. b) Harmonic components, T = 5 s. For the levels of composite resonance m o d e l l e d, t h e p e a k-t o - p e a k r i p p l e voltages along the 1400 and 3000 km transmission lines were shown to reach 0,14 and 0,2 pu at the maxima, while the peak-to-peak ripple voltage magnitude was less than 0,07 and 0,1 pu at the line ends. Understanding wave conditions the insulators along the transmission line will experience a greater stress than under ideal operating conditions – 0,035 pu for the 1400 km DC line and 0,072 pu for the 3000 km DC line. For a polluted insulator suffering high leakage currents, this could make the difference between withstand and flashover. The presence of standing waves could therefore worsen the DC line performance. Acknowledgement Fig. 5 (c): Voltage profile along the 1400 km line (rectifier CCC gain of 0,9). c) Voltage peak-to-peak amplitude profile along the line T = 4,9 to 5 s. peak-to-peak standing wave voltage magnitude of 0,14 pu near the center of the DC line. It important to note that the voltages at the ends of the DC line show a peak-to-peak ripple magnitude of less than 0,07 pu under these conditions. We can also see that although the standing wave maximum amplitude is measured to be near the centre of the DC line, that when superimposed on the DC voltage profile of the transmission line, the highest peak voltage is found closer to the rectifier end. The peak voltage along the transmission line was measured to be 0,035 pu greater than the DC voltage at the rectifier end. This could cause an insulator to flash over where it would otherwise have withstood the voltage stress. The DC line length was then extended to 3000 km, approximately twice that of the previous simulations. This also corresponded to the approximate length of the proposed DC line for the Westcor project [4]. The line geometry was assumed to be the same for easier comparison. The simulation was conducted in the same manner as the previous two, but with a significantly higher rectifier CCC gain of 1,6. The results showed a harmonic voltage profile along the line with an additional node near the centre of the DC line as seen in Fig. 6a. The voltage amplitude profile along the line shown in Fig. 6b shows a 70 Hz standing wave. The peak-to-peak standing wave amplitudes were 0,2 pu at the maxima and 0,1 pu at the line ends. Similar to the 1400 km line, the maximum peak voltages are found close to the Fig. 6: Voltage profile along the 3000 km line. a) Harmonic components, T = 5 s, b) Voltage peak-to-peak amplitude profile along the line T = 4,9 to 5 s. energize - June 2010 - Page 54 This paper was presented at the Cigré 6th Southern Africa regional conference: Somerset West 2009 and is reprinted with permission. References [1] C D Barker, N M MacLeod, R A Mukhedkar and R L Sellick. “Extra High Voltage DC Bulk Power Transmission – Opportunities and Challenges.” HVDC 2006 Congress, Durban, South Africa, July 2006. [2] A Williamson, U Åström, V F Lescale, D Wu and B Westman, “Latest development in transmission with 800 kVDC.” HVDC 2006 Congress, Durban, South Africa, July 2006. [3] N M Ijumba, A C Britten, L Rajpal, and L Pillay. “Development of Research Capacity in HVDC Technology in Southern Africa.” IEEE Africon, 2004. [4] L Pillay, A C Britten, P Naidoo, D Muftic, F F Bologna and N M Ijuba “Eskom's Proposed Strategic Research in Long-Distance HVDC Transmission.” HVDC 2006 Congress, Durban, South Africa, July 2006. [5] J P Reynders. “Elen 437 - Power Transmission and Protection Course Notes.” Department of Electrical and Information Engineering, University of the Witwatersrand, JHB. [6] J Arrillaga “High Voltage Direct Current Transmission – 2nd Edition”. , London, United Kingdom, 1998. [7] HVDC Transmission Course – Trans Africa Projects, Midrand, South Africa. 21-23 November 2007. [8] Definition – Standing Wave www.its.bldrdoc.gov/ fs-1037/dir-034/_5083.htm. [9] PSCAD Help files. [10] M Szechtman, T Wess, C V Thio, “First Benchmark Model for HVDC Control Studies,” Electra, No. 135, April 1991. [11] Personal communication with Eric Greyling. Line Manager, Apollo Substation. Contact B Duncan, University Witwatersrand, Tel 011 717-7211, b.duncan@ee.wits.ac.za or John Van Coller, Tel 011 717-7211, j.vancoller@ee.wits.ac.za