3p.3 OVERVOTAGE MEASUREMENTS REATED TO IGHTNING- DETECTION SYSTEMS IN NORWAY H. K. Høidalen F. Dahlslett hans.hoidalen@elkraft.ntnu.no Norwegian University of Science and Technology Norway frank.dahlslett@energy.sintef.no SINTEF Energy Research Norway Abstract: The paper presents measurement of lightning induced voltages in a real low-voltage system (isolated neutral) in Norway. The measurements are compared with calculations based on lightning location system data (position and signal strength) showing reasonable agreement for ground conductivity.1 S/m. However, the unknown grounding conditions in the system plays a very important role. 182 lightning strokes were registered within 1 km from the measurement site during the summer of 21. For 12 of these, induced voltages above 5 V were measured. Keywords: Induced voltages, lightning location, measurements, calculations, ground conductivity. 1. INTRODUCTION Power quality has become a key quantity for Norwegian electric power utilities. The demand of stable and reliable energy supply is continuously growing. In addition, the regulator fines power outages above 3 minutes with up to 5 NKr/kWh. ightning is a major cause of power quality disturbances (voltage dips, surges, interruptions etc.). In Norway a larger project at SINTEF Energy Research has used PQNode instruments to compare power quality measurements with lightning data [1,2]. However, the sampling frequency and the clock precision were found insufficient. As a result a new PCbased instrumentation was developed and tested in the summer of 21, following the ideas in [3]. Modern ightning ocation Systems (S) [4] may provide extensive data on the lightning exposure of power systems. By combining such data with efficient system computation models and tools it should be possible to estimate the impacts of lightning on the power quality of a local power system, also taking the actual power system configuration into account. The possibility to reach this rather ambitious goal is mainly limited by: The detection efficiency and location accuracy of the lightning location system and peak current estimate. The availability of an efficient computation model for induced lightning surges The complexity of the power system, including the loads (and their representation) in the low-voltage system. The transformer modelling and the handling of attenuation effects in the system are particularly important and critical. As a second step towards the ambitious goal, calculations based on S-data from the year 21 are compared with measurements. The same system as in [2] is used. 2. INSTRUMENTATION The lightning-induced voltages are measured with passive high-voltage probes (1:1) with a 15 MHz bandwidth. The common grounding point is taken from a wall-mounted isolated conductor connected to a small grounding rod. The voltage probes are further connected to buffer circuits with bandwidth 7 MHz and located in an outdoor metal cabinet. The buffer circuits supply the coaxial cables (RG58) to an indoor PC. Figure 1 shows the setup of the instrumentation. A plate antenna for measurements of the vertical electrical field is also mounted on the wall, unfortunately too close to the power line. The signal from the antenna is also connected to a buffer and integrated so that the output signal is proportional to 1 kv/m with 1 volt positive offset error.
ICP 22 Cracow, Poland September 2-6, 22 221 The instrumentation is based on a data acquisition card in a PC. This card samples 4 channels continuously with sampling time of 1 MS/s and a 12 bit vertical resolution. The card uses the analogue trigger on channel 1 with a fixed trigger level of.5 volts. The pretrigger time is set to the maximum value 3.2 ms. A computer program based on abwindows is developed to control the cards and store the data after each triggering. The total length of the recorded event is 65536 microseconds (this gives a 512 kb binary disk file). The data card is ready to record a new event almost immediately, but the dead time between the second and the third event is about 1.5 seconds. Plate antenna overhead line Fig. 1. Measurement setup. Connection of voltage probes, vertical antenna and buffer. Node 1 in figure 2. The PC is also equipped and clock synchronised with an accurate GPS card. Due to problems with the digital trigger circuit it was not possible to find the exact time for the lightning event. Instead the GPS time was read just after the signal was recorded but before saving it to disk. This made is possible to identify and correlate the measured event to the lightning location data. 3. CACUATIONS 1:1 probes Cabinet for buffer circuits The calculations are based on the analytical approach from Rusck [5], modified to handle linear rising and falling lightning currents and finite line lengths. Besides, the lossy ground effect on the radial electrical field propagation is modelled with the Cooray-Rubinstein approximation [6,7]. The coupling between the electrical field and the overhead line(s) is modelled with the Agrawal model [8]. The induced voltages are calculated as outlined in [2, 9]. The main assumptions in this method are: The lightning channel is straight and vertical. The return stroke current propagates unattenuated upward with a constant velocity, v (T model). The shape of the lightning current is linear in time. Only the field from the return stroke is considered. The earth is homogenous with constant conductivity and permittivity. The overhead line is considered to be lossless. The lossy ground effect on the horizontal field is handled. Symmetry in connected electrical terminations (loads) is assumed and only the common mode system is studied. An analytical formulation of a two-phase overhead line from (x A, y A ) to (x B, y AB ) exposed by a lightning stroke in (x, y ) is implemented in the MODES language in ATP [1, 11]. The low-voltage network where the measurements and calculations are performed is shown in fig. 2. It consists of about 2.5 km of overhead line that mostly is performed as compact 2 or 3 phase XPE covered conductors. There is no neutral or ground wire in the system, which is typical for the Norwegian countryside. There are also no protective devices in the system. 4 3 3 (-14,36) 2 5 T 6 2 (-4,2) (, ) 4 (75,-7) -2 Cu High voltage lines -4 N Ex -6 1 (-255,-72) E -8-6 -4-2 2 4 6 8 1, Fig. 2. System configuration. Some simplifications are made on the northern part where 3 subscribers are connected. : subscribers installation (load), T: transformer. Units in meters. Solid line: low-voltage system (23V-IT), dotted line: high voltage overhead line (24 kv). Ex: 3-phase XPE covered compact line (one phase is open) from the HV line. Cu: 2-phase cu-line. The squares in fig. 2 marked with are loads connected to the system. The loads are modelled as shown in fig. 3 with a capacitor of 1 nf in series with a
222 26th International Conference on ightning Protection
ICP 22 Cracow, Poland September 2-6, 22 223 1, 5-5 lossless.1 S/m.1 S/m measured 8 6 4 2-1, -2 lossless.1 S/m.1 S/m measured -1,5 1 2 3 4 5 6 7 8 9 1 Fig. 6. Measured and calculated return-stroke induced voltages for lightning 467_1 (21-7, 12:4:57.52). Distance 97 m, peak current estimate: -47 ka. 2-2 -4 lossless.1 S/m.1 S/m measured -6 1 2 3 4 5 6 7 8 9 1 Fig. 7. Measured and calculated return-stroke induced voltages for lightning 467_2 (21-7, 12:4:57.54). Distance 936 m, peak current estimate: -12 ka. -4 1 2 3 4 5 6 7 8 9 1 Fig. 8. Measured and calculated return-stroke induced voltages for lightning 468_1 (21-7, 12:42:45.15). Distance 464 m, peak current estimate: -4 ka. 8 6 4 2-2 lossless.1 S/m.1 S/m measured -4 1 2 3 4 5 6 7 8 9 1 Fig. 9. Measured and calculated return-stroke induced voltages for lightning 468_2 (21-7, 12:42:45.2). Distance 334 m, peak current estimate: -3 ka.
224 26th International Conference on ightning Protection 4 2-2 -4-6 lossless.1 S/m.1 S/m measured -8 1 2 3 4 5 6 7 8 9 1 Fig. 1. Measured and calculated return-stroke induced voltages for lightning 469 (21-7, 12:5:19.31). Distance 386 m, peak current estimate: -1 ka. Both the strokes 467_1 and 469 result in a strange measured signal both in front of and after the return stroke as shown in fig. 11 for lightning 469. The same phenomenon was measured in the electrical field, except for the AC component and the increasing DC-component. Induced voltage [kv], phase R.8.6.4.2 -.2 5 1 15 2 25 3 -.4 -.6 -.8-1 -1.2 1 2 3 time [ms] Figure 11. Measured induced voltage on a millisecond scale. The voltage consists of the normal 5 Hz component with amplitude 23/ 3 volts, an initial high frequency signal 1, the induced voltage from the return stroke 2 (shown in fig. 1), a positive part after the return stroke 3, an increasing DC component 4. Around 21 ms a subsequent event is registered, but this is not reported by the lightning detection system. 5. DISCUSSION Figure 6-1 all show that reasonable agreement between measurements and calculations is obtained for a 4 ground conductivity s=.1 S/m. The correct sign of the induced voltage is predicted and in the correct order of magnitude. These comparisons must be regarded as a very good match when all the uncertainties (current amplitude/velocity/shape, low-voltage loads, grounding etc.) are kept in mind. A calculation assuming a lossless ground would have given completely wrong results. The grounding condition plays a very important role, however, and has a strong influence both on the steepness and the amplitude of the induced voltage. It is not obvious how to handle the down-conductor and the connection to ground in order the measure the real stresses on a low-voltage installation. All the five events are at a rather long distance (3-9 km) and the resulting overvoltages are moderate. It is likely that no flashover occurred in the system, even if fig. 6 and 1 indicate a strange behaviour at the front. In these two cases a maximum calculated voltages of 1.79 kv and 1.18 kv (+ AC component) respectively will occur at node 4 (far end, ref. fig. 2). The oscillatory behaviour did not occur in fig. 7 where the difference from fig. 6 is the current amplitude only. A similar but not so dominant oscillation is shown in fig. 12. Induced voltage [kv] 1.5 1.5 -.5-1 Phase R -1.5 1 2 3 4 5 Fig. 12. Measured induced voltage, lightning 43 (12-7, 16:1:9.21). 14.3 km right east, peak current amplitude 59 ka. The measurements for lightning 467_1 and 469 both include a signal with a very high frequency content in front of the return stroke. The source of this could be a preliminary cloud discharge or an inter-cloud lightning. ightning stroke 469 resulted in a high increasing "DC-component" as shown in fig. 11. This could be due to a continuous lightning current. Stroke 467_1 and 467_2 showed a more moderate DC-component, while stroke 468 had no such component at all. The system in
ICP 22 Cracow, Poland September 2-6, 22 225 fig. 2 is not grounded so even a leader or cloud charge could introduce such a DC component. The positive part after the return stroke, indicated with symbol 3 in fig. 11 is difficult to explain by the authors. The induced voltage shown in fig. 12 shows also a positive second peak, but with a shorter duration than for lightning 467_1 and 469. The 3 rd open conductor in the compact XPE insulate overhead line obtains approximately the same induced voltages as the other two energised phases, which means that the coupling between the conductors is large. The vertical electrical field measured by the plate antenna was proportional to the induced voltage. This indicates that the antenna was located too close to the phase conductors and was influence strongly by them. The sample frequency of 1 MS/s is a bit low to get reliable results for the high frequency part in fig. 11 and for the oscillatory front event in fig. 6 and 1. However, it is sufficient to get an overview of the event and is a good compromise between return stroke effects and the total lightning flash effects. In several occasions the record length is sufficient to include subsequent events, like 467_1 and 467_2, reported as two individual strokes by the S. The exact timing of the measurements needs to be improved. Now the time is read from the GPS card after recording the data. The normal time delay is 55 ms, but is for a subsequent event some seconds due to the disk writing process. This time deviation is no problem for identifying the actual lightning stroke, but the time stamp has no value in verifying the lightning location system. The timing can be improved to 1 ms precision by introducing a common digital trigger for the DAQ and the GPS cards. Besides 12 recorded lightning-induced voltages coinciding with S-data, over 1 other events of voltages above 5 V (in the 23 V system) where recorded. These are a mixture of distant- and cloudlightning, and switching overvoltages. 6. CONCUSIONS In order to relate damages/fires and power quality to S-data, advanced models for calculating lightninginduced voltages must be developed and verified by measurements. The paper presents measurement of lightning induced voltages in a real low-voltage system (isolated neutral) in Norway. The measurements are compared with calculations based on lightning location system data (position and signal strength) showing reasonable agreement for ground conductivity.1 S/m. However, the unknown grounding conditions in the system plays a very important role. 182 lightning strokes were registered within 1 km from the measurement site during the summer of 21. For 12 of these, induced voltages above 5 V were measured. 7. REFERENCES [1] H. Seljeseth, O. Rokseth, J. Huse, F. Dahlslett: ightning impact on power quality preliminary investigations in Norway. Proc. Int. conf. on Electric Power Quality and Supply Reliability, june 9-12, 1999 Estonia. [2] H.K. Høidalen, J. Huse, F. Dahlslett, T. Aalborg: "Impacts of lightning-induced overvoltages on power quality in low-voltage distribution systems", Proc. ICP, pp. 564-569, Rhodes-Greece, 18-22 Sept. 2. [3] H. Torres, et.al.: "ightning induced voltages measured in a high lightning activity zone and its comparison with theoretical values", Proc. ICP, pp. 553-557, Rhodes-Greece, 18-22 Sept. 2. [4] F. Dahlslett, A. Pleym, O. Rokseth, K. Solberg: "Analysis of the collected data (1996-21) from the Norwegian lightning location system", Proc. ICP'22. [5] S. Rusck, Induced lightning over-voltages on power-transmission lines with special reference to the over-voltage protection of low-voltage networks, Royal Institute of Technology, PhD Thesis, Stockholm, Sweden, 1957. [6] M. Rubinstein, An approximate formula for the calculation of horizontal electric field from lightning at close, intermediate, and long range, IEEE Trans. on EMC, vol. 38, No. 3, pp. 531-535, Aug. 1996. [7] C.A. Nucci, Cigré 33.1: "ightning-induced voltages on overhead power lines", Part I: Electra, No. 161, pp. 74-12, Aug. 1995. Part II: Electra, No. 162, pp. 12-145, Oct. 1995. [8] A.K. Agrawal, H.J. Price, S. Gurbaxani, "Transient respons of multiconductor transmission lines excited by a nonuniform electromagnetic field". IEEE trans. on EMC, Vol. 22, No. 2, May 198, pp. 119-129. [9] H.K. Høidalen: ightning-induced overvoltages in low-voltage systems, PhD Thesis, The Norwegian University of Science and Technology, 1997. ISBN 82-471-177-7. [1]. Dubè, I. Bonfanti, "MODES: A new simulation tool in the EMTP". European Trans. on Electric Power, vol. 2, no. 1, pp.45-5, Jan./Feb. 1992. [11] Alternative Transient Program (ATP) RuleBook, Canadian/American EMTP User Group, 1987-1998. [12] H.K. Høidalen, ightning-induced voltages in lowvoltage systems and its dependency on overhead line termination, ICP 98, paper 3a.8, 14-18 Sept., 1998, Birmingham-UK.