ISBN 978-0-620-44584-9 Proceedings of the 16th International Symposium on High Voltage Engineering c 2009 SAIEE, Innes House, Johannesburg Copyright ° NEW COMPOSITE INSULATOR STRING DESIGNED FOR 400 KV OVERHEAD LINE USING STATISTICAL INSULATION COORDINATION PROCEDURE S. Vižintin1*, I. Uglešić2, B. Vertačnik3, H. Kifle3, B. Barl3, S. Lesjak3 Elektroinštitut Milan Vidmar, Hajdrihova 2, SI-1000 Ljubljana, Slovenia 2 University of Zagreb, Unska 3, 10000 Zagreb, Croatia 1 Elektro Slovenija, Hajdrihova 2, SI-1000 Ljubljana, Slovenia *Email: stane.vizintin@eimv.si 1 Abstract: The progress in the field of overhead line insulation, good experiences with composite insulators on the 110 kV level and appearance of some new insulation coordination computersupported techniques led us to the conclusion that for the next 400 kV overhead line project in Slovenia composite insulators should be used for the first time. In order to define the insulation level, a thorough overvoltage analysis was accomplished. Calculations and computer simulations were used to assess the temporary and transient (lightning and switching) overvoltages regarding their frequency and amplitudes. The lightning activity was represented by the actual highresolution lightning density map. With a bunch of data collected it was possible to perform a statistical insulation coordination procedure to determine the required insulation level. Some other system aspects were considered. The tests in the high-voltage laboratory showed appropriate dielectric strength for the insulator string that was designed at the beginning. Unfortunately the corona voltage inception level was too low due to unfavourable electric field distribution for the contemporary environmental requirements. Another solution is needed, so the insulator string with new metal fittings (potential rings, arching horns) is being designed. The paper shows results of the developed insulation coordination procedure in detail and dielectric tests. 1. INTRODUCTION 2.1. Electric power network The simulation scheme mapped the existing Slovenian 400 kV and 220 kV power transmission systems. The Beričevo-Krško OL is of a double-system with one ground wire. Its basic data are as follows: The insulation level for Slovenian 400 kV overhead lines was defined by using the deterministic insulation coordination procedure more than 40 years ago. The overvoltage data was established by means of calculations and more rarely by measurements. The standard insulator string was designed with glass cap/pin insulators. In the meantime a considerable progress in the field of insulation has been made especially by introducing composite insulators and some new insulation coordination computer-supported techniques. OL length: Number of towers: Medium span length: In our calculations of the switching overvoltages we took into account also the contribution of other 400 kV OLs. Five of them are those that connect power distribution transformer substations (TS) inside Slovenia (the length of individual OLs is between 31.1 km and 76.2 km), and four of them allow for connection with the Croatian, Austrian and Italian system (the length of individual OLs is between 32.2 km and 66.5 km). The two basic insulation levels (BIL) used in Slovenia are given in Table 1: In the last decade at the 110 kV level in Slovenia almost exclusively composite insulators were installed. Many discussions about their possible insulation properties degradation were raised in the past, but no problems have been experienced in service up to now. A thorough investigation [1] has revealed that aging has not left any serious deterioration on insulators. It is for this reason that a decision has been taken to design for the planned 2 x 400 kV Beričevo-Krško overhead line (OL) a new string with composite insulators with their dielectric parameters determined in the insulation coordination procedure. 2. 80.4 km 227 354 m Table 1: Standard insulation levels in Slovenia. MODELLING Um Insulation U250/2500 U1,2/50 420 kV BIL 1300 BIL 1425 950 kV 1050 kV 1300 kV 1425 kV On the 400 kV OLs there is solely single-pole automatic reclosure (AR) used. It is not likely that in the near future a three-pole AR will be used. The shortcircuit disconnection time does never exceed 0.5 s. Modelling of the Beričevo-Krško OL to allow for the calculation of the switching and lightning overvolatges was made by using the input data given bellow. Pg. 1 Paper E-58 Proceedings of the 16th International Symposium on High Voltage Engineering c 2009 SAIEE, Innes House, Johannesburg Copyright ° ISBN 978-0-620-44584-9 2.2. Beričevo-Krško OL 2.3. Equipments Geometrical circumstances (distances) for a 52.3 m high tower are shown in Figure 1: At both OL ends there are transformers in TS. Their basic data are given in Table 3: Table 3: Data of transformers in TS. Station Number (parallel) Power [MVA] Voltage ratio[kV] uk [%] Beričevo Krško 2 (autotransformer) 400 400/231 11,9 2 300 400/110 12,9 In our simulations we considered the magnetizing curve with a hysteresis for a typical 300 MVA transformer. In TSs, there are SF6 gas-filled circuit-breakers (CB) installed. As to their operational times, we considered the typical values: Closing time: Opening time: Breaking time: Most of the values obtained in our measurements of simultaneous closures of individual poles of modern (new) CBs (110 kV, 220 kV or 400 kV) were around 1 ms (between the first and the last pole). They never exceeded 2 ms. The warranty (manufacturer's) value was maximally 5 ms. The pessimistic values used in our simulations ranged from 2 ms to 5 ms. Figure 1: Spatial distribution of conductors (distances in [m]) for the Beričevo-Krško OL. While the distances between conductors do not change, tower heights Ht may vary. The actual values for Ht were taken into account in the analysis of lightning overvoltages. Dimension data are given in Table 2: In each of the TSs, the ZnO surge arresters (SA) are sited by transformers and there is none of them in the OL bays. We took into consideration the U-t characteristic for a typical SA of the rated voltage Ur = 336 kV (Ir = 10 kA, line discharge class 3). Table 2: Data for the medium heights of conductors and ground wires on the Beričevo-Krško OL. Conductor Phase x [m] y [m] 1 2 3 4 5 6 7 A B C A B C z.v. -6,2 -10,2 -7,0 6,2 10,2 7,0 0 30,0 21,5 14,3 30,0 21,5 14,3 43,6 Number of conductors D [m] 2 (horizontal) 0,4 1 - 60 … 90 ms 22 ms (± 3 ms) 40 ms 3. OVERVOLTAGES In our investigation we calculated all the types of overvoltages which are in accordance with the rules imposed on insulation coordination (IEC 60071) assessed to be of a decisive importance for the 400 kV OLs. The overvoltages were dealt with in the light of their cause. Value x in Table 2 means horizontal deflection of the conductor from the tower symmetral and y the average distance above the ground in which the sag is already taken regard of. The phase conductor represents a bundle of two Al/Fe 490/65 wires of the diameter dc = 30.6 mm (R20°C = 0,059 Ω/km) in the horizontal design and with the distance among themselves D. The ground wire is of the Optoflex ASB 4.1.3s type (Ay/ACS 131/25-13.4) with the diameter dgw = 18.0 mm (R20°C = 0.233 Ω/km). 3.1. Single-phase earth-fault overvoltages The usual causes for the occurrence of temporary overvoltages are: single-phase earth-faults, load rejections, ferro-resonance etc. The overvoltages important for the needs of insulation coordination in our circumstances are those taking place at singlephase earth-faults. Their value is determined with earth-fault factor k. Besides the characteristics of the power system (mostly the applied network grounding system), it is also the fault location that affects k. The measured specific ground resistance along the right-of-the way usually varies in the range from 100 Ωm < ρ < 1000 Ωm. The value used in our simulations was 500 Ωm. Our analysis was made with the PSS/E25 software package. Earth-fault factor k can be estimated by using complex impedance Z1 (direct system), Z2 (inverse system) and Z0 (zero system) with simultaneous Pg. 2 Paper E-58 Proceedings of the 16th International Symposium on High Voltage Engineering c 2009 SAIEE, Innes House, Johannesburg Copyright ° ISBN 978-0-620-44584-9 consideration of failure resistance R. We determined it for all the points of the Slovenian 400 kV electric power system (Table 4): 700 Table 4: Impedances of the 400 kV network. 100 Z [Ω] TS Cirkovce Šoštanj Beričevo Divača Krško Maribor Okroglo Podlog R1 1,18 1,31 0,91 1,02 0,70 0,94 1,34 1,06 X1 16,0 18,4 11,7 12,2 10,6 13,6 16,0 15,1 R0 3,58 1,39 1,97 4,18 0,90 2,77 3,26 1,63 [kV] 400 -200 Ratios Z X0 26,3 16,6 14,9 18,9 12,7 22,5 28,1 15,3 R0/X1 0,224 0,076 0,168 0,342 0,084 0,203 0,204 0,108 X0/X1 1,639 0,902 1,273 1,543 1,196 1,653 1,757 1,014 -500 R1/X1 0,074 0,071 0,078 0,084 0,066 0,069 0,084 0,070 -800 0,00 0,02 0,04 (f ile DV400kV.pl4; x-v ar t) v :X0009A v :X0009B 0,06 0,08 [s] 0,10 v :X0009C Figure 3: Phase voltages UA, UB and UC at the end of the open OL of the Krško TS at connection from the Beričevo TS. The other OL system was at this time loaded. As seen from Figure 3, at the beginning (10 % of the overall time), the induced voltage (not equal to zero) is in each of the three phases. The highest inter-phase voltages (close to 1200 kV) are also at the end of the open OL for the same case as above. The difference between the first and last CB pole closure time is approximately Δt ≈ 5 ms. Figure 4 shows another case - disconnection of OL at the Beričevo TS 5: Since the neutral point of the 400 kV network is directly grounded, k is determined by using the family of curves given in the IEC 60071-2 standard for R1 = 0. The family is shown in Figure 2: 400 [kV] 250 100 -50 -200 -350 -500 0 Figure 2: Short-circuit factor k for various values of R0 X 1 in X 0 X 1 at (R1 = 0). 10 20 (f ile DV400kV.pl4; x-v ar t) v :X0009A v :X0009B 30 40 [ms] 50 v :X0009C Figure 4: Phase voltages UA, UB and UC at the beginning of OL at the Beričevo TS at disconnection. As seen from values in Table 3, the earth-fault factor does not exceed the value k = 1.2 in none of the points of the power system. This means that low temporary overvoltages may be expected. Further we analysed the overvoltage phenomena by using a statistic CB with which closures take place at arbitrary selected times as foreseen by the Monte Carlo method; this is of course within the selected limits (depending on pole dissipation = 1,45 ms). The applied distribution was even with the pole set at some 3.2. Switching overvoltages It is known that the highest overvoltage amplitudes take place at CB closings at the end of an open OL. They depend on the moment of the CB pole closure so that the highest overvoltages occur at closure in the moment when the driving voltage is at its maximum. The amplitude of the supply phase (1 p.u = 343 kV) or inter-phase (1 p.u = 594 kV) voltage is used as a basis for calculation of the overvoltage factor [p.u.]. Connection of one system of a loadless OL is first analysed in a deterministic and then in a statistic way. The EMTP software was used. t = 10 ms 3 . We simulated 100 of such arbitrary selected closure times and statistically processed overvoltage values. Results for CB closings at the end of the open OL are given in Tables 5 and 6: Table 5: OL CB closing from the Beričevo TS. When connecting a loadless OL, the wave deflects and the voltage gets doubled at the OL open end. Figure 3 shows the voltage time development at the Krško TS for such case if connected from the Beričevo TS: Pg. 3 Line Beginning (CB) Open end Voltage [p.u.] Phase A Phase B Phase C A-B B-C A-C Umean 1,242 1,172 1,202 − − − Umean 1,694 1,546 1,541 1,430 1,386 1,535 St.d. 0,105 0,065 0,098 − − − U2% 1,44 1,34 1,43 − − − St.d. 0,153 0,206 0,204 0,140 0,188 0,120 U2% 1,96 2,10 1,95 1,74 1,80 1,75 Paper E-58 Proceedings of the 16th International Symposium on High Voltage Engineering c 2009 SAIEE, Innes House, Johannesburg Copyright ° ISBN 978-0-620-44584-9 Table 6: OL CB closing from the Krško TS. Line Beginning (CB) Open end Voltage [p.u.] Phase A Phase B Phase C A-B B-C A-C Umean 1,216 1,174 1,198 − − − Umean 1,748 1,587 1,570 1,428 1,393 1,579 St.d. 0,090 0,066 0,092 − − − U2% 1,39 1,33 1,40 − − − St.d. 0,181 0,197 0,214 0,136 0,149 0,141 Based on our analysis, we determined the values for the lightning current maximal amplitude Imax, median value Mp and standard deviation σp. The values are given in Table 7: U2% 2,10 2,95 1,98 1,73 1,63 1,85 Figure 5 shows a cumulative distribution overvoltages for one of the observed cases: Table 7: Properties of the lightning stroke currents for the Beričevo-Krško OL. of Lightning strokes Period [years] I max [kA] Mp [kA] σp [kA] 1587 8 142,9 12,5 12,5 The data were obtained by Slovenian lightning location system SCALAR. The average lightning stroke density on the OL right-of-the way is: 100 90 80 ng = 2.4 str./km2/year 70 % 60 Our simulations of lightning overvoltages, in which the lightning stroke parameters were changed according to the Monte Carlo method (current, shape, angle, and place of stroke), were made with the SigmaSlp software package. For each OL configuration we made 1000 simulations. The place of stroke was determined with an electro-geometrical model. 50 40 30 20 10 2,1 2,15 2 2,05 1,9 1,95 1,8 1,85 1,7 1,75 1,6 1,65 1,5 1,55 1,4 1,45 1,3 1,35 1,2 1,25 0 Napon [p.u.] p.u. Voltage Following our analysis of simulation results, the below major characteristics can be drawn: Figure 5: Cumulative overvoltage probability at CB closing at the end of the open OL (KrškoTS) phase B. - We also made overvoltage simulations at connection of OL terminated with a transformer at its no-load operation, connected additional OLs and during singlephase automatic fast reclosure. In each of the observed states, overvoltages were lower. - 3.3. Lightning overvoltages In our calculation of lightning overvoltages we took into account the actual values for the lightning stroke density ng and the probability distribution of currents in accordance with the log-normal distribution. Figure 6 shows a section of the right-of-the way of the Beričevo-Krško OL where colours represent the lightning stroke density from ng = 0.5 str./km2/year (yellow) to ng = 3.0 str./km2/year (blue). - - - - The number of strokes into the phase conductor (shielding failures) is substantially lower than the number of strokes into the ground wire. This share is approximately from 12 % to 15 % of the total number of strokes into OL. The majority of strokes into the phase conductor triggers a flashover (for any of the possible circumstances and insulation level). Their share is from 75 % to 100 %. The impact of the tower height or footing resistance Roz is rather irrelevant. When lightning strikes into the ground wire, it is only its small share that gives rise to a backflashover. For the footing resistance of up Roz = 30 Ω and insulation level BIL 1300 kV this share is less than 7 %. By using the strengthened insulation BIL 1425 kV, these values are further decreased and are below 6 %. As expected, the role of Roz at lightning strokes into the shielding wire when using the BIL 1300 kV and BIL 1425 kV insulation is important. The maximal values of the flashover shares are between 2.7 % (Roz = 20 Ω) to over 12 % (Roz = 40 Ω). When values of Roz are low, the majority of the BIL 1300 kV and BIL 1425 kV insulation flashovers are due to strokes into the phase conductor; when Roz values are higher, the share of back-flashovers may increase by over 50 % of their total number. The high trees along both sides improve OL performance very much. The flashover frequency decreases by over 75 % for BIL 1300 kV. The final criterion is a comparison of the 400 kV OL fault occurrence rate when using either BIL 1300 kV or BIL 1425 kV. Figure 6: Lightning stroke density ng between 0.5 str./km2/year (yellow) and 3.0 str./km2/year (blue). Pg. 4 Paper E-58 Proceedings of the 16th International Symposium on High Voltage Engineering c 2009 SAIEE, Innes House, Johannesburg Copyright ° ISBN 978-0-620-44584-9 4. Value of Urw-50Hz must be recalculated so as to obtain switching withstand impulse overvoltage Urw-250/2500 : INSULATION COORDINATION Based on our calculations and statistical processing of overvoltages, we determined the insulation level by combining the deterministic (temporary overvoltages) and statistic (transient overvoltages) method. U rw250/ 2500 1,7 U rw50Hz 1,7 361,7 614,9 kV since it covers Urw-50Hz. 4.1. Statistical method 4.3. Slow-front overvoltages The point of reference in the statistical method is the OL acceptable flashover risk or OL failure occurrence rate. Flashover probability Rfo for a certain overvoltage of the same type can be calculated on the basis of the overvoltage occurrence distribution f(U) and flashover probability on the insulator P(U) with the below expression: Judging from our computer simulations, the maximal voltage at OL connections on its open end is equal to U2% = 2.10 p.u., thus making the coordination insulation withstand voltage to be: 420 2,10 2 720,1kV 3 3 By taking into account the correction factors, it is increased up to the value of: U cw 2,10 2 Rfo f U PU dU 0 This means that besides the overvoltage there should also be a model provided to demonstrate flashover on the insulator string. The procedure is graphically illustrated in Figure 7: Um U rw250/ 2500 ks kaUcw 1,11,11 720,1 879,2 kV which determines the required withstand voltage. P(U) 4.4. Fast-front overvoltages 1 Our approach to fast-front overvoltages was to some extent different than the one given above. Our intention was to determine the increase in the flashover rate ∆N for the Beričevo-Krško OL if instead of BIL 1425 kV we use BIL 1300 kV. Our calculations were made for the values of OL tower footing resistances Rgr = 20, 25 and 30 Ω and OL tower heights Ht = 52, 57 and 62 m. The calculated OL flashover frequency for the two observed voltage levels is: 0,8 f(U) 0,6 0,4 0,2 Rfo Figure 7: Determination of flashover probability Rfo . N1425kV = 1.740 flashovers/year N1300kV = 1.831 flashovers/year According to IEC 60071-1, insulation of the 400 kV system belongs to range II. It is then dielectrically defined with the following withstand voltages: This means that by decreasing the insulation from BIL 1425 kV to BIL 1300 kV, the flashover frequency increases by 5.2 %. By taking into account the OL length lOL = 80.4 km and the average lightning stroke density ng = 2.4 str./km2/year we obtain (norming at lOL = 100 km and ng = 1 str./km2/year) the value N*1425kV = 0.901 flashovers/100km/year . U250/2500 standard switching impulse (250/2500 μs) U1,2/50 standard lightning impulse (1,2/50 μs) As foreseen by the insulation coordination procedure, we had to define an appropriate insulation level. To determine the required withstand voltages Urw, we used the appropriate insulation coordination factors for safety ks and above-sea-level altitude ka (maximally 1000 m). 4.5. Insulation level By considering results of our analysis of temporary and transient overvoltages, the required standard withstand overvoltages can be determined. For the switching withstand voltage Urw-250/2500 the following applies: 4.2. Temporary overvoltages For the maximum earth-fault factor that may occur in the Slovenian 400 kV network, k = 1.2 is assumed. With consideration of Um = 420 kV we determine the representative overvoltage that is the same as the insulation coordination withstand voltage, e.g. U rp 1,2 420 Temporary overvoltages: Slow-front overvoltages: 614,9 kV < 950 kV 879,2 kV < 950 kV As to the lightning overvoltages, when determining the lightning withstand voltage Urw-1,2/50, we find it reasonable to adopt a 5.2 % increase in the flashover frequency at N1300kV. This makes the standard insulation level to be: 3 291,0 kV . By taking into account insulation coordination factors, we get: U rw50Hz ks kaUcw 1,11,13 291,0 361,7 kV Pg. 5 Paper E-58 Proceedings of the 16th International Symposium on High Voltage Engineering c 2009 SAIEE, Innes House, Johannesburg Copyright ° ISBN 978-0-620-44584-9 On the potential ring, the corona (Figure 9) appeared at a relatively low voltage of 195 kV and extinct at 193 kV. U250/2500 = 950 kV U1,2/50 = 1300 kV In compliance with specifications of the Slovenian legislation, there shall be electrically strengthened insulation applied on certain locations. This in practice means that instead of the insulators rated BIL 1300 kV those rated BIL 1425 kV should be used. 5. INSULATOR STRING The adopted standard withstand voltages have to be considered in construction of insulator strings. The first to be defined are the basic demands for selection of composite insulators and amongst them the use of: - silicone rubber (HTV) for housing, and - ECR glass fibers for rod. Figure 9: Corona on the bottom part of the insulator at the voltage of 292 kV. The pollution state being moderate we opted for the minimum specific creepage distance of 20 mm/kV (pollution level II). As a result of a number of positive effects [2], the selected construction involves inbuilding of potential gradient control rings on both insulator ends. There were two types of the insulator ring constructed with an adjustable distance lad between arcing horns. The insulator length is lins = 2940 (BIL 1300 kV) or lins = 3280 (BIL 1425 kV). Our results of the dielectric tests are given in Table 8: Judging from the measurement results, with this construction we are quite far away from the limit value laid down for the corona evoking voltage which is set at > 291 kV and is being used as a criterion in several European countries. In re-designing the metal armours, the calculated maximal value of the single-phase short circuit of 37.5 kA (< 0.5 s) will be taken into account. 6. Following our experience, calculations and field measurements regarding the use of new 400 kV insulator strings, the conclusions to be drawn are: Table 8: Insulator strings withstand voltages. Impulse shape Polarity Single (suspension) BIL 1300 Double (tension) BIL 1425 U50% [kV] U10% [kV] U50% [kV] U10% [kV] 1,2/50 μs + − 1493 1647 1435 1583 1717 1844 1650 1772 CONCLUSION 250/2500 μs + − 1123 1124 1036 1037 1176 1330 1084 1226 - The insulators of our preference are composite insulators (silicone rubber, ECR glass). - Potential rings will be fitted on both insulator ends. - In the insulation coordination procedure, our analysis of overvoltages showed the appropriate insulation level to be U250/2500 = 950 kV, U1,2/50 = 1300 kV. - In the used design of the potential rings, the corona takes part at a considerably lower voltage than the adopted minimal limit value set at 291 kV, for which reason certain alterations shall have to be effected (larger diameter, double ring, etc.). The above withstand voltages were obtained at the flashover distance between the arcing horns lad ≈ 2655 mm (BIL 1300 kV) or lad ≈ 3000 mm (BIL 1425 kV). The acceptable withstand voltages for insulators (with no armours) are by 5 % to 10 % higher. We also made a radio interference (IEC 60437) and a corona voltage test (IEC 61284) on a single suspension string. The results of the first measurement at the atmospheric conditions (T = 17.0 °C, b = 96.4 kPa, hr = 37.5 %) are shown in Figure 8: The tower construction being old, the inter-phase insulation corresponds to the BIL 1425 level and was therefore not addressed in our analysis. 7. REFERENCES [1] S. Vižintin, A. Varl, S. Jamšek, M. Hrast, B. Barl, “State Analysis of 110 kV Composite Insulators after Being in Service in the Electric Power System of Slovenia for Several Years”, 14 th International Conference on High Voltage Engineering, Beijing, China, Aug. 2005. [2] WG 22.03, “Use of Stress Control Rings on Composite Insulators”, ELECTRA 143, pp. 69–71, Aug. 1992. Figure 8: Results of radio interference (RIV) test. Pg. 6 Paper E-58