Ultra highvoltage gasinsulated switchgear a technology milestone

EUROPEAN TRANSACTIONS ON ELECTRICAL POWER Euro. Trans. Electr. Power 2012; 22:60–82 Published online 18 May 2011 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/etep.582

Ultra high-voltage gas-insulated switchgear – a technology milestone Uwe Riechert*,y and Walter Holaus ABB Switzerland Ltd, CH-8050 Zurich, Switzerland

SUMMARY China is in urgent need of electrical power. Huge power plants are built all over the country and the enormous flow of electrical power to the large megacities has to cross several thousand kilometres from the source to the end user. At those dimensions, losses of the power lines can be significant. The State Grid Corporation of China (SGCC) is thus aiming for 1100 kV as the voltage level for AC transmission to keep losses as low as possible, a step into a new area of electrical grids. Asea Brown Boveri (ABB), together with its partners and suppliers, has developed the heart of such a system – a gas-insulated switchgear (GIS) design – that could pass all the tests with this groundbreaking technology. Many years of experience at voltage levels of up to 800 kV are available as the basis for developing GIS for 1100 kV. Nevertheless, the individual components such as circuit-breakers (CB) and disconnectors differ greatly from the known designs. As regards the CB, for instance, it emerges that a design with the closing resistor in a parallel tank is advantageous. The very fast transient overvoltages become more important for the disconnector design. Copyright # 2011 John Wiley & Sons, Ltd. key words:

ultra high-voltage; gas-insulated switchgear; disconnector switch; circuit-breaker; testing; very fast transient overvoltage

1. INTRODUCTION Reliable supply of electrical energy is one of the backbones of modern economies. A safe and reliable operation mainly depends on high-voltage switchgear – the core part of an electrical power system. The high-voltage circuit-breaker (CB) in this switchgear is often the last line of defense when big systems must be protected in the event of a short circuit. Electrical grids and the corresponding substations are well known as air-insulated systems in which the high-voltage is kept away from both the ground and people by distances of tens of metres. Another much more compact way of building high-voltage switchgear is the gas-insulated design – gas-insulated switchgear (GIS). Gas-insulated switchgear technology was introduced to the market in 1966 with the first 170 kV GIS underground substation delivered to the Zu¨rich city center (Figure 1). In 1976, Asea Brown Boveri (ABB) delivered the first 500 kV GIS to Claireville, Canada. With the installation of the first 800 kV GIS in South Africa in 1986, ABB has proven its technology leadership also at the ultra high-voltage (UHV) level. This so-called alpha substation has been in operation for more than 20 years without any failures or unplanned interruptions. The 500 kV GIS in Itaipu, Brazil is still the world’s largest installation but will soon be overtaken by the ABB GIS inside the Three Gorges Dam in China. China is a huge country where electric power generation happens mainly in the western parts and load centers are typically found in the coastal region – thousands of kilometres apart. Both AC and DC UHV systems are necessary to handle the increase in electric energy consumption and to back up the existing transmission system [1,2].

*Correspondence to: Uwe Riechert, ABB Switzerland Ltd, CH-8050 Zurich, Switzerland. y E-mail: [email protected] Copyright # 2011 John Wiley & Sons, Ltd.

ULTRA HIGH-VOLTAGE GAS-INSULATED SWITCHGEAR

61

Figure 1. ABB’s GIS history: from first research projects to the world’s largest installation within 50 years.

The State Grid Corporation of China (SGCC) – one of ABB’s biggest customers – began designing an AC system with a rated voltage of 1100 kV a few years ago [1]. This project initiated extensive research and development efforts in research institutes and at equipment manufacturers [3]. To finally determine the technical feasibility, a group of three Chinese and two Japanese GIS manufacturers and ABB were asked by SGCC to take part in the development of UHV GIS equipment for the Chinese UHV AC demonstration project. It was established in 2008 in central China and consists of almost 600 km of high-voltage lines and three substations – Jingmen, Nanyang and Changzhi. The first three switchgears in the UHV demonstration project are partially executed using gasinsulated design or as hybrid systems (Hybrid-IS), that is, as a combination of air-insulated and gasinsulated components. One of these switchgears is being supplied by ABB/Xi‘an Shiky as Hybrid-IS with a 2-CB layout (Figure 2). The demonstration project has already started operating in January 2009. The switchgear layout comprises virtually all GIS components such as the CBs with closing resistor, disconnectors, earthing switches, busbars, insulators, current transformers and bushings. Today, there is no international standardization for rated and test voltages at the 1100 kV voltage level. CIGRE´ WG A3.22 has worked on recommendations for standardization [3,4]. The rated voltages

Figure 2. Picture of ’Jingmen’ 1100 kV station. Copyright # 2011 John Wiley & Sons, Ltd.

Euro. Trans. Electr. Power 2012; 22:60–82 DOI: 10.1002/etep

62

U. RIECHERT AND W. HOLAUS

for the demonstration project are based on an SGCC specification. This specification was drawn up on the basis of extensive grid studies and investigations of voltage co-ordination [5]. Together with experience from the demonstration project, the specified requirements will also form the basis for further projects in China. The basis for the project is the Chinese Standard DL/T 593-2006, which largely corresponds to IEC 62271-1 with the addition of special Chinese requirements. The main rated values are shown in the following list:

Rated voltage 1100 kV. Nominal operating voltage 1000 kV. Rated lightning impulse withstand voltage (LIWV) to earth 2400 kV. Rated short-duration power frequency withstand voltage to earth 1100 kV. Rated switching impulse withstand voltage to earth 1800 kV. Rated power frequency 50 Hz. Rated normal current (switching devices/busbar)) 4000 A/8000 A. Rated short-time withstand current 50 kA, 3 seconds. Rated peak withstand current 135 kA.

2. ELK-5 ULTRA HIGH-VOLTAGE GAS-INSULATED SWITCHGEAR DEVELOPMENT PROJECT To design and install this 1100 kV GIS, ABB and Xi‘an Shiky, the biggest Chinese supplier of GIS, established a joint development project called ‘ELK-5’ (ELK is the name of ABB’s GIS systems; 5 indicates the new performance level). The focus for ABB in this joint effort was on the overall design of the Hybrid-IS and on the production and shipping of core components. A very demanding schedule was set by SGCC – after its start in November 2006 the first installation at Jingmen was to be energized by the end of 2008. Accomplishing this in 2 years would be a world record for upgrading a GIS to a new, demanding voltage level, during which time the development, verification, type testing, production and installation would also occur. The essential basis for the design consists of the specified dielectric requirements. If the values for withstand voltages from IEC 62271-203 are used together with the specification values, the scaling shown in Figure 3 is obtained. Figure 3 shows that the per unit (pu) values for impulse withstand voltages decrease with increasing rated voltage. Using the same dimensioning field strengths, the basic dimensions of a UHV GIS for static components would increase sub proportionally to the voltage. When disconnectors are switched, very fast transient overvoltages (VFTO) occur in SF6-insulated

Figure 3. Dependency of rated withstand voltages and VFTO on rated voltage as per IEC 62271-203, max. VFTO corresponds to 2.2 pu (Hybrid-IS) and 2.8 pu (GIS). Copyright # 2011 John Wiley & Sons, Ltd.



63

systems. As the rated voltage increases, the difference between the rated LIWV and the VFTO decreases. Hence, VFTO may become dielectrically dimensioned at UHV voltage levels. In addition to the dielectric design requirements, additional requirements for switching devices have to be taken into account. For this reason, the design of the CB and the disconnector is discussed in more detail below.

3. CIRCUIT-BREAKER – THE CORE COMPONENT 3.1. Design The CB is a switchgear component capable of safely turning on and off under all switching conditions, such as normal operation or fault clearance. The rated values of 1100 kV and 4000 A correspond to a rated power of 7600 MW for the three phases. This is more than the average electric power consumption of Switzerland [6]. With this rating the CB would be capable of turning on and off the electrical power of Switzerland. For the design of the complete CB, a comparison of required drive energies as a function of number of interrupter units connected in series is given in Figure 4. It shows the drive energy and contact speed that are required to reach a contact gap of 370 mm within 10 ms, calculated using Equation (1) and T100a ¼ 1650 J. 2 1 gap EdriveO;i ¼ ðm0 þ Smi Þ þ i ET100a 2 i tarc;min

(1)

It can be seen from this mechanical model, that there is an optimum for the required drive energy close to 4 interrupter units connected in series. The drive energy requirement for 2 interrupter units is as high as for 7 interrupter units and both are significantly higher than the optimum. A low contact speed reduces stress and increases robustness and mechanical endurance of the whole CB; it is also an advantage to have multiple interrupter units connected in series. With 4 interrupter units connected in series, the drive energy is at the optimum, and the contact speed and switching capability of ABB puffer interrupter units fits to the UHV specification. Consequently, 4 interrupter

Figure 4. Drive energy required for Ur ¼ 1100 kV as a function of number of interrupter units. Copyright # 2011 John Wiley & Sons, Ltd.


64


Figure 5. Circuit diagram for a circuit-breaker (CB) with closing resistor (CR) and closing resistor switch (CRS), CR parallel with CB (left), CR in series with CB (right).

units connected in series have been chosen for the UHV CB, as successfully used for years in a double stack in 550 kV CBs. For availability reasons, it is also sensible to work with one drive only. The specified closing resistor can be positioned in parallel or in series to the interrupter units (Figure 5). When connected in series, the closing resistor switch (CRS) short-circuits the resistor shortly after the CB is closed. The CRS must be able to carry the rated current and the short-time current. When connected in parallel, the CRS closes the resistor shortly before the CB makes contact. The CRS only needs to be able to carry a few kA for a short period. With series connection, all 4 interrupter units are positioned behind one another. Series connection with the closing resistor is not appropriate as the tank would become very long. This makes it advantageous to position the closing resistor parallel with the interrupter units. In addition to the volume of the closing resistor, the space requirement for the contact to switch on the resistor must be taken into account. Therefore, it is useful to place the CR in a separate tank. The benefits of series connection led to a decision in favour of the solution for an 1100 kV CB shown in Figure 6. Moreover, Figure 6 shows a size comparison with CBs at different voltage levels. The closing resistor is positioned in a separate tank, parallel with the interrupter units. The CRS is operated by the CB drive via a linkage. On closing the CB, the CRS performs a Close-Open operation. On opening the CB, the CRS does not move at all. This solution offers several advantages:

The force required for the interrupter units does not need to be diverted. The CRS does not need to carry rated or short-circuit current. The resistor stack can be adapted to specifications, independently of the CB. The tank diametres become substantially smaller, making them easier to manufacture. A dedicated gas compartment can be selected for the CRS and the resistor. Horizontal positioning in the layout minimizes the framework, allowing very good accessibility to all components without platforms. Because they are built on at the side, the drive, interrupter units, CRS and CR in a system can be inspected independently, without dismantling other GIS components.

Figure 6. 1100 kV circuit-breaker with separate CR: size comparison. Copyright # 2011 John Wiley & Sons, Ltd.



65

3.2. Development Specification of the fundamental design was followed by the development phase as such. The path from the design to the final product is the critical and most demanding part of the development process. To keep the development time as short as possible, the latest simulation and modelling methods were used (Figure 7): 3D design, dielectric calculations, dynamic field calculation for CB contacts to determine voltage co-ordination for making and breaking operations, simulation of voltage distribution for the interrupter units, mechanical calculations and simulations of bursting behaviour, simulation of earthquake behaviour, calculations of the mechanical function chain to determine the load values for the insulators and moving parts, simulations of the progression of movements under different drive and breaking and making conditions, simulation of deflection with static load and in case of short-circuit, flow simulation for the interrupter unit to optimize breaking behaviour, simulation of forces and temperatures in case of peak and short-time current loads, internal fault arc simulation, temperature calculation at rated current load, with the help of the thermal network method. Thanks to the use of the latest simulation and development technologies in conjunction with dimensioning criteria based on long experience and the most recent research findings, it is possible to keep the development time very short. At the same time, this approach achieves a high likelihood of optimal design with a small number of required development tests and high levels of safety and reliability for the type tests and during operation. 3.3. Voltage grading Certain special features regarding voltage distribution over the interrupter units should be noted on a 4-unit CB as opposed to a 2-unit CB. The voltage distribution depends on the number of interrupter units, the capacitances over the contact gap and to earth, and on which side the voltage or voltages are applied. To take optimal advantage of the breaking and making behavior of the individual interrupter units, the objective should be the most homogeneous voltage distribution that is possible. This can be achieved if grading capacitors are positioned parallel to each interrupter unit. As the number of interrupter units increases, the voltage distribution becomes less homogeneous. To ensure

Figure 7. Dielectric 3D field calculation. Copyright # 2011 John Wiley & Sons, Ltd.


66


Figure 8. 4-Unit CB, equivalent circuit diagram.

that voltage-grading factor is low, larger grading capacitances are therefore required for series connection of 4 interrupter units as compared to a 2-unit CB. Figure 8 shows the equivalent circuit diagram to calculate the grading capacitances Cp required for a voltage stress on one side, at which the maximum unevenness occurs. When calculating the voltage distribution, it has to be remembered that the grading capacitances show divergences in the capacitance value DCp for production-related reasons. Maximum unevenness occurs precisely when the voltage-side capacitor shows the lowest value, whereas all other grading capacitors are manufactured at the upper tolerance limit (’worst case’). The calculation results shown in Figure 9 reproduce the voltage-grading factors for the individual interrupter units both for the normal case (index n, no grading capacitor tolerances) and for the ’worst case’ described (index w, maximum grading capacitor tolerances). It can be seen that in the ’worst case’, the voltage stress on the first interrupter unit is about 3% higher than if the interrupter units are assembled with the specified capacitances. To keep the maximum voltage-grading factor for one unit as low as possible, that is, below 10%, and at the same time to enable the lowest possible capacitances to be used, different grading capacitors are used in the 1100 kV CB. The first grading capacitor has a capacitance increased by double the tolerance values, so the capacitance of the first interrupter unit is always greater than that of the second. A further increase in the capacitance of the first interrupter unit is not advisable; otherwise the voltage-grading factor is increased under phase opposition conditions. This principle must be applied on both sides of the CB. By using different grading capacitors according to these rules, the absolute voltage stress of the first interrupter unit can also be reduced by over 5% in the ’worst case’. A similar effect with identical grading capacitors on all interrupter units can only be achieved if the capacitance value is doubled. Figure 9 shows a comparison of all the cases described: Case A:Cp1 ¼ Cp2 ¼ Cp3 ¼ Cp4 ¼ Cp Case B:Cp1 ¼ Cp4 ¼ Cp þ 2 DCp; Cp2 ¼ Cp3 ¼ Cp Case C:Cp1 ¼ Cp2 ¼ Cp3 ¼ Cp4 ¼ 2 Cp

Figure 9. Voltage-grading factors: voltage stress on individual interrupter units, in relation to one quarter of the total voltage with voltage stress from one side. Copyright # 2011 John Wiley & Sons, Ltd.



67

Figure 10. Voltage stress for individual interrupter units, in relation to one quarter of the total voltage with voltage stress from one side: frequency-dependent (left), for impulse stress with a rise time of 400 ns (right).

The least favourable voltage-grading factor must be taken into account for dimensioning purposes. The ’worst case’ is also the basis for calculating the voltage stress for power tests in the form of halfpole tests, or on individual interrupter units (quarter-pole test, unit test). The voltage-grading factor calculations presented should not only be applicable for rated frequency voltage stress. In fact, the maximum voltage stress must apply for all types of voltage occurring in GIS systems, that is, they must not be higher even for lightning impulse voltage, up to and including voltages with extremely high rate of rise (VFTO). The voltage-grading factor is frequency-dependent here, because real grading capacitors contain inductances and resistances. As an example, the dynamic behavior shown in Figure 10 is obtained. By way of clarification, the voltage stress for individual interrupter units for an impulse voltage stress with a rise time of 400 ns is shown. In summary, the fact has been established that the individual interrupter units are not overstressed even when subject to VFTO stresses.

3.4. Closing resistor A closing resistor will be regarded as an integral part of the CB. Beginning with the dissipated energy a few properties of the resistor discs have to be considered (www.hvrint.com). The resistor discs are made of mixtures of clays, alumina and carbon. Diametres range from a few millimetres to 151 mm. For high-voltage applications most often the largest diametres are used due to their energy allowance. According to the manufacturer, the resistance of resistor discs is a function of temperature and voltage. Increasing temperature and/or voltage will decrease the resistance. During the insertion time of the closing resistor when assuming a stiff voltage source, the dissipated energy Ed can be calculated according Equation (2): Z Ed ðtÞ ¼

UðtÞ2 dt RðtÞ

(2)

As the insertion time is usually only a few millisecond the energy will increase the temperature of the resistor discs and will according Equation (3) reduce the instantaneous value of the resistor disc and increase the resulting dissipated energy. log10ðrðtÞÞ TCR ¼ 0:16 e 1:4 0:135 (3) The voltage across the resistor discs will decrease the resistance according Equation (4): VCR ¼ 0:62 rðtÞ0:22 Copyright # 2011 John Wiley & Sons, Ltd.

(4) Euro. Trans. Electr. Power 2012; 22:60–82 DOI: 10.1002/etep

68


Figure 11. Time progressions for resistance, energy and temperature during 2 times 11 ms closing time.

The resistance at time t can be calculated according Equation (5): RðtÞ ¼ R20

ð100 þ TCR dT þ VCR Ud Þ 100

(5)

The temperature rise of the resistor discs can be calculated using Equation (6): dTðtÞ ¼

Ed ðtÞ Vrd cm

(6)

The result from a simulation of two closing operations with phase opposition at 1000 kV is shown in Figure 11 with the progressions of the resistance value R(t) for a cold resistance of 560 V, the accumulated converted energy E(t) and the temperature W(t). To allow both closing operations to be drawn in one chart, cooling was mapped for 30 min by a temperature jump of 40 K at time 15 ms. The resistance progression clearly indicates the dependencies on temperature as well as voltage. In the least favourable case, the resistance value at the end of the simulation attains about 450 V with a temperature increase of 120 K. The converted energy in this case is 92 MJ for 2 times 11 ms of closing time. 4. DISCONNECTOR 4.1. Design Layout studies for GIS and Hybrid-IS systems have shown that a 908-angled disconnector (DS) offers most layout options and at the same time requires the lowest number of GIS components. This disconnector design is therefore used for the UHV GIS (Figure 12) [7]. Optionally, the disconnector can also be equipped with an earthing switch if one is required in the layout. For 1100 kV GIS and Hybrid-IS systems, the phase distance is typically 10–15 m, so a three-pole drive with a linkage between the phases is no longer appropriate. 4.2. Very Fast Transient Overvoltages Before starting the design of an UHV DS, it has to be clarified whether a VFTO damping measure is required. This decision shall be based on the maximum VFTO peak values that occur when switching Copyright # 2011 John Wiley & Sons, Ltd.



69

Figure 12. Cross-section of the UHV disconnector.

and its reference to the rated LIWV of the equipment. If the maximum VFTO reach the LIWV, it is necessary to design and maybe to test considering the VFTO level or to suppress severe VFTO for the insulation co-ordination. During switching of disconnectors in GIS a varying number of pre-strikes and re-strikes occur, depending of the speed of the switching device. Due to the very short duration of the voltage collapse of a few nanoseconds at the switching gap, travelling surges are generated in the busbar duct. The multiple refractions and reflections of these surges at impedance discontinuities within the enclosures create complex waveforms, which depend on the DS design, the operating conditions and the configurations of the GIS. Very fast transient overvoltage simulation is a well-known instrument for the calculation of overvoltages needed for the insulation co-ordination process. Because the accuracy of the simulation depends strongly on the quality of the model of each individual component, it is important to verify the simulation results by measurements. The maximum value of the VFTO depends on the voltage drop at the DS just before striking and the location considered. Trapped charge remaining on the load side of a DS must be taken into consideration. A trapped charge on the load side resulting in a voltage of 1 pu (2 pu across the DS) is normally taken into account as the most unfavourable case for high speed DS or phase opposition conditions. This precondition is normally used for the calculation of VFTO [8]. For this case, the maximum VFTO peak in GIS configuration has a typical value between 1.5 and 2.6 pu. As a basis for the design criteria, the peak value of VFTO under various switching conditions for the substations of the pilot project in China were calculated and analysed in comparison to Ref. [9]. The calculation results are summarized in Table I. For Hybrid-IS the VFTO levels will remain below 2.2 pu in any case. Hence, the maximum VFTO stress is always below the LIWV level of 2.66 pu. Therefore, it was concluded that a damping resistor is not required. There are differences between the different calculations as shown in Table I. A comparison between simulation and measurement can verify the accuracy of the simulation [10]. Therefore, the simulation method was also used during testing. According to IEC 62271-102, Annex F, VFTO amplitudes of at least 1.4 pu are required without pre-charging of the busbar. Figure 13 shows a comparison of simulated and measured VFTO without and with DC pre-charging of the busbar. The measured voltage progressions coincide very well with the simulation results as regards VFTO amplitude and rise time. During disconnector switching, VFTO of 2 MVoccur, corresponding to a pu value of 2.23. This value is above the VFTO which occur in real operation of Hybrid-IS. The successful test results therefore confirm both the accuracy of the calculation method and that no VFTO mitigation measures are required for the Hybrid-IS design. 5. TESTING Development and type tests at 1100 kV voltage level are a challenge to test laboratories. So far, no laboratory anywhere in the world is able to carry out all the necessary tests. This means that the tests Copyright # 2011 John Wiley & Sons, Ltd.


Copyright # 2011 John Wiley & Sons, Ltd.

No 2400 1.15 2087 2249 2.50 1.08 [5]

Switching resistor

GIS BIL (kV) Safety factor Protection level VFTO (kV) VFTO (kV) VFTO (pu) VFTO/protection level Refs. 2400 1.15 2087 1940 2.16 0.93

No

Single busbar

2400 1.15 2087 2260 2.502 1.08 [5]

No

Double busbar

2400 1.15 2087 2742 3.05 1.31 [5]

No

Double busbar future extension

2400 1.15 2087 1157 1.29 0.55 [5]

Yes 500 V

Yes 500 V 2400 1.15 2087 1250 1.39 0.60 [5]

Double busbar future extension

Single busbar

Jingdongnan (GIS)

2400 1.15 2087 1878 2.09 0.90 [5]

No

2400 1.15 2087 1204 1.34 0.58 [5]

Yes 500 V

Double busbar

Nanyang (MTS) Double busbar

Table I. VFTO calculation results for the Chinese pilot project.

2400 1.15 2087 1836 2.04 0.88 [5]

No

2400 1.15 2087 1409 1.57 0.68

No

Double busbar

2400 1.15 2087 1268 1.41 0.61 [5]

Yes 500 V

Double busbar

Jingmen (MTS)

70 U. RIECHERT AND W. HOLAUS



71

Figure 13. VFTO calculation and measurement when switching busbars with a GIS DS as per IEC 62271102, without pre-charging (left), with pre-charging (right).

have to take place in different laboratories. If the technical facilities are available, the type tests are carried out at the XIHARI Laboratories in Xi‘an, China. In addition, the STRI laboratory in Sweden and ABB’s laboratories in Germany, Switzerland and Sweden are involved. This entails outlay for intercontinental transportation. Transport by air-freight shortens transport times substantially. Due to the size of the test objects, maritime transport is the only possibility in some cases. Moreover, the time expended on setting up tests and gas handling is many times more than for 550 kV systems.

5.1. Dielectric tests The dimensions of the test objects require the availability of considerable space in the high-voltage laboratories. For high-voltage tests with combined AC and switching impulse voltages, for example, a minimum distance between bushings of 13 m must be maintained. The distance between live parts and walls must not be less than 10 m. These requirements already exceed the limits of most high-voltage laboratories. The size of the test objects (Figure 14) substantially increases the capacitance of the test objects as compared to 550 kV systems, that is, high performance high-voltage transformers must be available. Partial discharge (PD) measurements are an element of the dielectric type tests. These values are measured conventionally as per IEC 60270, that is, via a coupling capacitor. The apparent charge depends on the dimensions of the test object. Compared to a 550 kV GIS, the apparent charge for an

Figure 14. High voltage type test in the XIHARI laboratory: experimental set-up (left), UHF partial discharge sensor (right). Copyright # 2011 John Wiley & Sons, Ltd.


72


identical defect is lower by a factor of 2–3. To attain similar sensitivity with maximum AC test voltage, the background noise should be in the range from 0.1 to 0.2 pC. In this case, the background noise level is dependent on the size of the coupling capacitor and on the electromagnetic shielding of the highvoltage lab. The experimental set-up shown in Figure 14 has a test capacitance of about 3 nF. The transformer’s power limit requires the capacitance of the coupling capacitors to be relatively low, at 0.35 nF. Due to external interference from the test systems and corona, this experimental set-up results in a background noise level of about 7 pC at 1000 kV. A lower background level could be attained through the use of SF6-insulated AC voltage transformers. To achieve high sensitivity for the laboratory measurements, acoustic and UHF PD measurement have been used. PD measurement in the UHF range has become established as an acceptance criterion for on-site testing in recent decades [11,12]. To use the UHF method, it is necessary to install several field sensors which were integrated into the test object (Figure 14). One apparent ’drawback’ of the UHF measurement is that the UHF signal cannot be unambiguously correlated to the apparent charge from the PD source, so calibration as per IEC 60270 is impossible for physical reasons and factors related to measurement technology. However, a so-called ’sensitivity verification’ of the UHF sensors is possible [13]. The UHF-PD measurement not only allows very sensitive measurements but also localization of the PD source by means of a time-of-flight measurement. Impulse generators are not usually designed for the dimensions of 1100 kV test objects. As a result, overshoots of 7–10% occur in the peak range during the test with lightning impulse voltage. The overshoots are therefore greater than the definition in IEC 60060. For SF6-insulated systems, compensation of the lightning impulse voltage according to the revision of IEC 60060-1 in IEC TC42 is also considered correct for larger overshoots. As testing with the existing impulse generators will have to continue in the future, the recommendation is still that the requirement for dielectric type tests for the special requirements in the case of 1100 kV should be taken into account in the standardization. CIGRE´ WG D1.36 is currently working on this topic. Combined voltage tests should be conducted across open isolating distance and across open switching device. For this purpose, AC voltage is applied on one side and an impulse voltage on the other. In the case of 1100 kV, this test sets special requirements for the testing technology. The relatively high switch capacitance causes over coupling of the impulse voltage on the AC transformer during the combined voltage tests. The voltage drop must be compensated by a voltage increase. The over coupling can be reduced by additional lumped capacitances on the AC voltage side. Several nF are required. The potential capacitance value is nevertheless restricted by the power limit of the transformer. During the type tests, maximum over coupling of 20% of the impulse voltage occurred. The over coupling can be compensated on the impulse voltage side or the AC voltage side. In case of compensation on the impulse voltage side, the insulation to earth is stressed beyond the rated values. Full compensation on the AC voltage side leads to very long voltage stress close to the rated short-time power frequency withstand voltage. Combined tests with low voltage drop require AC transformers with current from 3 to 4 A on the high-voltage side.

5.2. Circuit-Breaker making and breaking tests Limits of the test laboratories may prohibit testing of the entire CB in a single test, especially in UHV testing. Therefore, the concept of unit testing or half-pole testing is developed and described in IEC 62271-100, in which only one or two interrupter units of the CB are tested with a limited voltage. To enable generation of an equivalent stress to the full-pole test, and hence a flow of current through all interrupter units, it is possible to work with two synthetics. For this purpose, it is necessary to manufacture a special CB with a centre connection and/or to set the CB tank to potential and insulate it to earth (Figure 15). All breaking and switching tests were performed at the XIHARI power laboratories as half-pole tests with a special CB tank with a centre connection. Two interrupter units were used as half-pole test CB. The other two interrupter units were used as auxiliary breaker. For half-pole tests, the TRV level corresponds to the level for tests on 550 kV CBs. Synthetic systems are not generally optimized for these tests. In particular, the relatively flat rate of rise of recovery voltage (RRRV) as compared to 550 kV can be problematic. The implementable RRRV causes higher Copyright # 2011 John Wiley & Sons, Ltd.



73

Figure 15. 1100 kV CB a, half-pole test set-up (right), full-pole test set-up (left).

dielectric stress after the short-circuit interruption in some cases (Figure 16). For T10 a possible future requirement for RRRV ¼ 10 kV/ms according to Ref. [3] has been taken into consideration. As described in Section 3.3, different grading capacitances were used for the inner and outer interrupter units in order to optimize voltage distribution. The voltage-grading factor must be indicated by the manufacturer in order to calculate the TRV values. The voltage-grading factor depends on:

how many interrupter units are included in the test, and which ones; whether a centre connection is present; the bushings where the voltage is connected; and whether a combination of several synthetic circuits is used. In this case, different voltage-grading factors must be respected for the individual partial voltages.

The voltage-grading factor for an interrupter unit with a voltage stress on one side is 8.6% (’worst case’). Two different marginal conditions may be used to determine the voltage-grading factors for half-pole tests: (A) Regardless of the tolerances for the grading capacitors built into the test CB, the voltagegrading factor for two interrupter units is stipulated such that the maximum percentage voltagegrading factor for one interrupter unit of 8.6% is not exceeded. This corresponds to the known procedure for a full-pole test on a 2-unit CB, as per IEC 62271-100. As a general rule, no special grading capacitors are built in at the tolerance limits for full-pole tests, in order to guarantee the maximum dielectric stress for an interrupter unit. This only becomes possible with unit tests. The authors regard the procedure as the preferred method for half-pole or unit tests. In this way, it is possible to indicate a percentage voltage-grading factor for all power tests:

Unit test with TRV from one side: 8.6%. Half-pole test with TRV on the outer connection: 4.1%. Half-pole test with TRV on the centre connection: 1.3%. Unit test with phase opposition: 5.4%.

Figure 16. 1100 kV circuit-breaker half-pole tests, required and prospective TRV for terminal fault T100a (left) and T10 (right). Copyright # 2011 John Wiley & Sons, Ltd.


74


Figure 17. Full-pole tests, required and prospective TRV for terminal fault T100a (left) test circuit (right).

Half-pole test with phase opposition: 1%. The percentage of distribution for a half-pole test fluctuates between 1% and 8.6%. Appropriate voltage-grading factors can also be specified for power tests with uneven voltage distribution, for example, short-line fault. (B) The voltage-grading factor is determined in relation to the capacitances of the installed grading capacitors or the voltage distribution measured on the test CB. This ensures that the maximum voltage-grading factor (’worst case’) for an interrupter unit is attained during each test. With this method, the voltage distribution must be calculated or measured before each test. This method was used for the tests with additional margin by application of a percentage voltagegrading factor of 10%. Moreover, it was requested that full-pole tests are carried out, in order to represent the correct voltage between the live parts and enclosure. This is because exhaust products during the breaking process are known to reduce the internal dielectric withstand capability. T100a and T100s were carried out by applying half the necessary voltages in opposite polarity to the CB terminals, with the CB on an insulating platform (Figure 15). The moving contact side was connected to the enclosure. The fixed contact side will obtain the positive polarity of TRV. In this way, 100% TRV will appear across the interrupter units and between fixed contact and enclosure. The test circuit and the TRV for T100a are shown in Figure 17. The so-called ’platform’ method has compared to other methods advantages. Auxiliary CB for only half of the rated voltage, smaller voltage dividers, smaller test halls and the current injection method are possible. Challenges still remain in the development of high power testing of UHV CB among others are the correct recovery voltage (RV) and the representation of transient stresses in unit tests. Regarding the correct application of AC RV after interruption special tests carried out by applying an AC or DC voltage to the enclosure and the TRV in opposite polarity to the CB terminals, with the CB on an insulating platform (Figure 18). The test peak value was increased to the RV peak value. In this way, 100% TRV and RV will appear between fixed contact and enclosure. The positive test results therefore confirm the interruption capability. The making tests T100s(a) were performed at the ABB power laboratories. The latest update of IEC 62271-100 does not allow to use a fuse wire for the T100s(a) making test. Therefore, the proper timing

Figure 18. Full-pole tests, required and prospective TRV and RV for terminal fault T100a. Copyright # 2011 John Wiley & Sons, Ltd.



75

Figure 19. T100s(a) making test set-up, synthetic ignition circuit and test circuit-breaker.

for making current was done using a synthetic ignition circuit (Figure 19). The test procedure aims to demonstrate the ability of the CB to close against a symmetrical current as a result of the pre-arcing commencing at the peak of the applied voltage. To achieve the longest pre-striking arc in the 1100 kV CB with synthetic ignition circuit for 550 kV, the SF6 pressure was reduced to 0.2 MPa absolute [14]. By doing so, the pre-arcing time of the 1100 kV CB with four interrupter units was similar to the prearcing time of a 550 kV CB with two interrupter units of the same design. The method has the advantage to perform multiple making operations without disassembly; it gives the right mechanical forces for the drive during making operation and results in the correct pre-arcing distance. As a result, this method with reduced gas pressure is a simple measure to perform a making test without having the full making voltage available. 5.3. Closing resistor tests The tests under phase opposition conditions for 1100 kV cannot be performed on a complete closing resistor. Only a representative section can be tested as it is important that the voltage across the section, the current through the section and the insertion time are as in real phase opposition conditions. The largest section that could be tested in ABB Power Lab in Baden (Switzerland) was 19% of the closing resistor. The voltage was applied for 11 ms. The voltage and the current were measured and the varying resistance was calculated by dividing voltage by current and is shown in Figure 20. Dividing measured signals is problematic when both values are close to 0 (measurement errors) and when the two signals

Figure 20. Resistance of closing resistor, comparison between measurement and calculation. Copyright # 2011 John Wiley & Sons, Ltd.


76


are not exactly in phase (as the current is not completely resistive due to the large setup). Therefore, the measured resistance in the grey sections of the figure cannot be interpreted as real and must be ignored. However, the measurement shows quite good agreement with the calculation. The measured resistance is significantly reduced – up to 17% – compared with the predicted up to 19%. The voltage coefficient (VCR) and the temperature coefficient (TCR) provided by the manufacturer of the resistor discs seem to reflect reality appropriately and can be used to design closing resistors without unnecessary conservatism. 5.4. Disconnector Switching tests Annex F of IEC 62271-102 describes the requirements for switching of bus-charging currents by disconnectors for rated voltages of 72.5 kV and above. Three test duties (TD) are defined to prove the correct design by special switching tests: TD 1: Switching of a very short section of busbar duct – TD 1 is a normal type tests and is mandatory for DS. The circuits for DS testing were chosen in such a way, that maximum pu values for VFTO peak were generated and it was assumed that they would also be the highest possible in the GIS. TD 2: Switching of parallel capacitors for CBs under 1808 out-of-phase conditions – TD 2 is required, if the CB is equipped with parallel capacitors. All UHV CB consist of two or more interrupter units. TD 3: Current-switching capability test – TD 3 serves only to indicate the current interruption capability of the DS when de-energizing long busbars or other energized parts. The development and type tests are carried out at the STRI laboratory in Sweden. Table II gives the test values for all TD. For these TD, test poles with the test DS and auxiliary DS for TD 1 and with DS and CB for TD 2 have been established (Figure 21). Internal capacitive sensors were used to measure the voltages. The maximum value of the VFTO depends on the voltage drop at the DS just before striking and the location considered. Trapped charge remaining on the load side of a DS must be taken into consideration. Depending on the design of the DS (especially contact speed, dielectric design of the contacts and SF6 gas pressure) the assumption of trapped charge resulting in 1 pu voltage is a very conservative assumption for VFTO calculations. The trapped charge voltage is specific for each design and could be analysed during type testing or simulated with high accuracy as basis for the insulation coordination. Totally 400, 520 records of measured trapped charge voltages during testing of DS have been evaluated for 550 kV, 1100 kV respectively. For the DS, the maximum trapped charge reaches 0.6 pu during the tests with a source voltage of 1.1 pu resulting in a most unfavourable voltage collapse of 1.6 pu. In case of 1100 kV DS 90% of the trapped charge voltages were limited to 0.45 pu (Figure 22). The distribution of the trapped charge voltages have been simulated to investigate if the measured distributions can be reproduced and to study the influence to the DS speed on these distributions. For the simulations two parameters of the DS have to be known. The first is the breakdown voltage characteristic depending on the contact distance. The electric fields on the relevant parts of the moving and the fixed contact side have been determined using dielectric calculations. The polarity effect of SF6, the surface roughness and the gas pressure were taken into account. The second parameter is the contact speed. The model simulates the test circuit for the relevant TD according IEC62271-102 Annex

Table II. Test values for disconnector switching on UHV GIS DS. Test duty

Test voltage Source side U1

Load side U2

TD 1

700 kV ACrms

990 kV DC

TD 2 TD 3

700 kV ACrms 635 kV ACrms

700 kV ACrms —


Number of CO operation for development tests 100 and 100 with reversed terminals 100 100 Euro. Trans. Electr. Power 2012; 22:60–82 DOI: 10.1002/etep


77

Figure 21. Test pole for TD 1 (left) and TD 2 (right) switching tests.

F. The capacitance values of the source and load side have been determined according the length and typical value of GIS bus duct capacitance. The starting point of the contact separation has been varied in 18 steps all around 3608. It can be seen that the measured trapped charge distributions can be reproduced with good agreement for the 1100 kV DS (Figure 23, left). The trapped charge voltage distribution depends strongly on the contact speed. For contact speeds exceeding 1 m/s, the simulation results suggest that the 90% trapped charge voltage increases significantly from around 0.5 to 0.8 pu (Figure 23). However, depending on the design of the disconnector (especially contact speed, dielectric design of the contacts and SF6 pressure) the assumption of trapped charge resulting in 1 pu voltage is a very conservative assumption for VFTO calculations. The evaluation of type test results for the 1100 kV disconnector have revealed that the 90% voltage associated with trapped charges where 0.4 pu at a source voltage of 1.0 pu (Figure 23). That means for this 1100 kV disconnector having such a trapped charge behaviour, a safety factor of around 1.43 (¼2 pu/1.4 pu) is included when 1 pu trapped charge voltage is assumed for the VFTO calculation. The out-of-phase condition is relevant when the DS on the transformer side is required to be closed before synchronizing the system. TD 2 is necessary if the CB is equipped with parallel capacitors. For the tests, a CB in open position was used to have the proper capacitive current during testing (Figure 24, left). When pre-striking shortly after an opening occurs, the amplitude of the temporary power frequency overvoltage results from the combination of the trapped charge voltage and the over coupled voltage from the CB grading capacitors. The over coupled voltage correlates to the capacitance ratio

Figure 22. Measured trapped charge voltage during DS switching under laboratory conditions, probability density (left), approximation of absolute values by normal distribution (right). Copyright # 2011 John Wiley & Sons, Ltd.


78


Figure 23. Measured trapped charge voltage during DS switching under laboratory conditions, comparison between measured and calculated values at 1100 kV (left), influence of contact speed, statistic values (right).

Figure 24. Simplified simulation test circuit (left) and calculation result with trapped charge voltage of 0.45 pu (case a) or 1.1 pu (case b) (right).

between source and load side CP/(CP þ Cbx). The dependence of the voltage across the gap as a function of Cbx is shown in Figure 24 to the right. Considering a trapped charge voltage on the load side of 1.1 pu (case b), it can be seen that the voltage across the gap reaches 3.3 MV for a typical minimum capacitance to earth of 150 pF. Considering the trapped charge behaviour of the DS the voltage across the DS is lower. For case a, a remaining trapped charge voltage of 0.45 pu covering 90% of all cases the maximum voltage across the open gap reaches 2.9 MV. Figure 25 shows the comparison of the voltage across the DS and the rated voltages for different voltage levels for the two different cases: case a: Trapped charge voltage of 0.45 pu (90% probability value at 1.1 pu source voltage). case b: Trapped charge voltage of 1.1 at 1.1 pu source voltage. The amplitude of the temporary power frequency overvoltage greatly exceed those of the rated withstand voltages, for both cases. For case a, only the combined lightning impulse voltage covers the temporary overvoltages across the DS during TD 2. It has to be mentioned that the duration of the voltage stress during the tests is in the order of some 10 minutes. Also the temporary voltage against earth at the load side could be higher compared to the rated voltages. For testing, it has to be considered that the test arrangement reflects the arrangement in service. Especially the distance between DS and CB should be as short as in real service. Moreover, TD 2 can be used to verify the fast transient stress withstand capability of the grading capacitors of the CB. In testing of metal enclosed CBs with grading capacitors, unit tests may not represent the transient stresses that occur due to unequal dielectric behaviour of the interrupter unit. In unit tests, stresses on grading capacitors such as occur in pre-strikes are not represented. Therefore, the CB was used as test device for switching of bus-charging current by DS. The perfect behaviour of the Copyright # 2011 John Wiley & Sons, Ltd.



79

Figure 25. Power frequency test voltage across the open DS (left) and against earth at the load side (right) in comparison with the rated values according to IEC 62271-203.

Figure 26. Dependency of bus-charging current, load capacitance (right) and equivalent busbar length (left) on rated voltage as per IEC 62271-203.

grading capacitors during the DS switching tests confirms dielectric capability of the grading capacitors. Disconnector switches have a capacitive current interrupting capability. The bus-charging current was specified as 0.8 A for the 1100 kV system, which corresponds to 4000 pF load side capacitance or a length of the load side busbar of 80 m. Figure 26 shows the load capacitance and the equivalent length of GIS for different voltage levels depending on the specified bus-charging current. The bus charging currents, evaluated based on the existing UHV substation layout, reach a maximum value of 0.84 A. Schemes of future UHV substations with maximum busbar lengths up to 200 m are concluded to be rather exceptional and a bus-charging current of 1 A is sufficient also for future applications [3]. Nevertheless, TD 3 tests with 2 A have also been performed. The bus-charging current corresponds to 10 000 pF load side capacitance or a length of the load side busbar of 220 m. Unavoidable resonance effects of the test transformer lead to overvoltages in the range of 2–3 pu. Therefore, a currentswitching capability test with a bus-charging current higher than 1 A and using common high-voltage transformers is not possible at the moment.

6. THE FIRST ULTRA HIGH-VOLTAGE GAS-INSULATED SWITCHGEAR SUBSTATION After development and the successful type testing in 2007 and 2008, ABB and Shiky began to assemble and ship the first equipment to the substation at Jingmen. This substation includes an almost complete set of GIS components, such as CBs with closing resistors, disconnectors, earthing switches, current transformers, busbars, bushings and insulators. Extensive layout studies to find the optimum arrangement of the GIS components proved that a ‘flat’ setup with good accessibility would be best Copyright # 2011 John Wiley & Sons, Ltd.


80


Figure 27. ‘Jingmen’ 1100 kV substation.

suited for the Jingmen Hybrid-IS substation [7]. The layout has the following characteristics (Figure 27): All GIS switching equipment is placed close to ground level. The flat arrangement improves robustness against seismic stresses. All the drives are placed at a height within 1.5 m of the ground, which provides convenient and safe access for operators during installation and maintenance. No platforms or ladders are needed. The layout can be easily extended in the busbar direction. It requires a minimum of steel construction as a bay structure. The on-site workload is small and allows for fast installation.

7. CONCLUSIONS The 1100 kV substation was installed in 2008 near the city of Jingmen in Central China. The substation has been in operation since January 2009 without any failures or unplanned interruptions. It transmits part of the energy produced by the Three Gorges power plant to the northern part of China. Meeting the challenge the ELK-5 development project was a big challenge in many respects: a pioneering design in an unprecedented execution time and a cross-continental cooperation with suppliers and partners in Europe and China, who with very different cultural backgrounds worked closely together. By utilizing the latest simulation and development technologies in conjunction with design criteria based on empirical values, the development time can be kept short while increasing the safety and reliability of the design. The switching components require specific attention. For the CB, it emerges that a design with four interrupting chambers in series and a closing resistor in a parallel tank is advantageous. Suitable selection of the grading capacitances enables a substantial reduction in overall capacitance. The closing resistor is a special challenge with regard to the thermal capacitance. By measurement and VFTO calculation it was possible to confirm both the accuracy of the calculation method and that no VFTO mitigation measures are required for the Hybrid-IS design. The trapped charge voltage must also be taken into account. Depending on the trapped charge voltage characteristic of the disconnector the resulting safety factor can be >1.4 for the traditional VFTO calculations assuming 1.0 pu trapped charges. Type tests of the ELK-5 components were carried out simultaneously in Chinese, Swedish, German and Swiss laboratories. Tests on UHV equipment represent a major challenge for the laboratories. Many tests reach – and in some cases exceed – the limits of the laboratories. This poses new problems also for the standardization sector, in order to take into account the special requirements for the UHV level. The experience gained by the manufacturers and institutions participating in the UHV demonstration project are of help here. Copyright # 2011 John Wiley & Sons, Ltd.



81

This project was not only the start of a new era in UHV transmission but also a powerful demonstration of the combined engineering capabilities of the world’s technology leaders.

8. LIST OF SYMBOLS AND ABBREVIATIONS r(t) A cm Cp DCp dT dT(t) Ed(t) Edrive gap L m pu R(t) R20 tarc, min TCR U(t) Ud Ur VCR Vrd ABB CB CR CRS DS GIS Hybrid-IS LIWV PD RRRV RV SGCC TRV UHF UHV VFTO

R(t) A/L resistivity in V cm area of resistor disc (cm2) specific heat capacity of resistor discs grading capacitance divergence of grading capacitance temperature rise in 8K temperature rise of resistor discs at instant t dissipated energy at instant t required drive energy contact gap thickness of resistor disc (cm) mass per unit resistance of the closing resistor at instant t resistance at 208C minimum arcing time temperature coefficient (%/8K temperature rise) voltage across closing resistor at instant t voltage across one cm of resistor rated voltage voltage coefficient (%/kV/cm) volume of all resistor discs Asea Brown Boveri circuit-breaker closing resistor closing resistor switch disconnector switch gas-insulated switchgear hybrid-insulated switchgear lightning impulse withstand voltage partial discharge rate of rise of recovery voltage recovery voltage State Grid Corporation of China transient recovery voltage ultra high frequency ultra high-voltage very fast transient overvoltages

REFERENCES 1. Int. Conference of UHV Power Transmission Technology, Peking, 2006. 2. Symposium on Int. Standards for Ultra-high voltage, Peking, 2007. 3. CIGRE´ Working Group A3.22: ‘‘Background of Technical Specifications for Substation Equipment Exceeding 800kV AC’’, Brochure, to be published in 2010. 4. CIGRE´ Brochure 362 WG A3.22: ‘‘Technical Requirements for Substations Equipment Exceeding 800kV – Field experience and technical specifications of Substation equipment up to 1200 kV’’, December 2008, ISBN: 978-2 85873-049-0. Copyright # 2011 John Wiley & Sons, Ltd.


82


5. CIGRE´ Working Group B3.22: ‘‘Technical Requirements for Substation Exceeding 800 kV’’, Brochure 400, December 2009. 6. [5th] Annual Report of SwissEnergy 2005/2006. 7. Holaus W, Riechert U, Sologuren D, Kru¨si U. ‘‘Development and Testing of 1100kV GIS’’; The second IEC – CIGRE´ International Symposium on International Standards for UHV Transmission, 29–30 January 2009, New Delhi, India, Proceedings pp. 128–141. 8. CIGRE´ Working Group 33/13-09: ‘‘Monograph on GIS Very Fast Transients’’, Brochure 35, July 1989. 9. Liangeng B, Zutao X, Sen SW, et al. ‘‘Estimation of VFTO for GIS and HGIS of China 1000kV UHV pilot project and its suppressing countermeasures’’; IEC/CIGRE´ UHV Symposium, Beijing, China, July 18–21, 2007, paper 2–3–4. 10. Riechert U, Kru¨si U, Sologuren-Sanchez D. ‘‘Very Fast Transient Overvoltages during Switching of Bus-Charging Currents by 1100kV Disconnector’’, CIGRE Session 2010, paper A3_107. 11. CIGRE´ Working Group 15.03: ‘‘GIS Insulation Properties in Case of VFT and DC Stress’’, Report 15–201, 36th CIGRE´ Session, Aug. 25 to 31, 1996, Paris, France. 12. Riechert U, Kudoke M, Strehl Th. ‘‘Monitoring und Diagnose von gasisolierten Schaltanlagen–Sinnvoller Einsatz von Teilentladungsmessungen’’ HighVolt Kolloquium 2003, Mai 22–23, 2003, Dresden, Konferenzband, paper 7.4, Konferenzband, Seite 249–258. 13. CIGRE´, Joint Task Force 15/33/03.05: ‘‘Partial Discharge Detection System for GIS: Sensitivity Verification for the UHF Method and the Acoustic Method’’, E´lectra, No. 183, April 1999, pp. 75–87. 14. Riechert U, Holaus W, Kru¨si U, Sologuren D. ‘‘Gas-Insulated Switchgear for 1100kV – Challenges in Development and Testing’’, CIGRE´ 6th Southern Africa Regional Conference, Colloquium of CIGRE´ SC A2/A3/B3, 2009, Somerset West, 17–21 August 2009, South Africa.

