Shielding electrodes of different form are usually foreseen and specified for AC lines with 3 main purposes: To eliminate corona from insulator fittings (i.e. a corona shielding function); to grade the voltage potential along the string in order to limit the radio interference from insulators and, in the case of composite insulators, to reduce ageing phenomena (i.e. a grading function); and, to limit the effect of power arcs (i.e. a protective function). The following 2017 contribution to INMR by Alberto Pigini examined different shielding requirements for AC and DC lines. In particular, results of previous investigations of cap & pin insulator strings is analysed and summarized. Then, the case of composite insulators is studied in more detail by performing ad hoc electric field calculations under AC and DC voltage.
Cap & Pin Insulator Strings: Laboratory Experience
Grading electrodes are essential in AC to mitigate the voltage drop/electric field along insulators, especially on the line side. Fig. 1, for example, reports the voltage drop, in p.u., measured across each cap & pin disc for a 19 unit 420 kV insulator string, with and without shielding electrodes.
The same trend can be reproduced using electrostatic field calculations, taking insulation geometry and permittivity of the different materials into account. While evaluated under clean conditions, the voltage distribution can also be assumed valid under contaminated conditions whenever capacitive currents prevail over conductive currents. A more direct comparison between the measured voltage drops, with and without shielding electrodes, is shown in Fig. 2.
It is immediately evident that, without grading electrodes, the voltage drop on the first unit of the string can be very high, e.g. up to 3 times the average. Grading electrodes significantly reduce this stress and are therefore important to limit radio interference for cap & pin insulator strings under AC voltage.
The situation, however, is different in DC. Now, voltage distribution is dominated by resistive currents – both in clean and contaminated service conditions. Non-uniform distribution similar to that in AC can be obtained in the case of new cap & pin insulators, thoroughly cleaned at low humidity. Conductance values of 10-13 S with currents on the order of magnitude of nA were recorded on new insulators cleaned by ultrasound, while conductance values of 10-12 to 10-11 S were recorded on new insulators cleaned with alcohol, with surface currents in the tens of nA. On the other hand, if the insulator string and fittings are well designed from the corona point of view, the configuration can be considered corona free. In this case, ionic current in air is limited and on the order of only a few nA.
Under ideal laboratory conditions, with new and completely clean insulators, ionic currents in air can influence voltage distribution, being of the same order of magnitude of surface current and making voltage distribution close to that in AC, i.e. dominated by capacitances. However, under real conditions, surface conductance can easily overcome 10-9 S for weathered or only lightly polluted insulators, with surface currents on the order of magnitude of hundreds of nA or higher. This renders negligible the effect of ionic currents in air and consequently there is a tendency toward linear voltage distribution. For example, the voltage distribution measured on a 32 unit cap & pin insulator string for 500 kV DC, without grading electrodes, is shown in Fig. 3. The string was uniformly contaminated with low pollution (ESDD=0.01 mg/cm2) and exposed to a normal service environment with humidity less than 76%.
Voltage on the first insulators in the string is only marginally higher than the average. Therefore, measurements indicate that additional field grading by shielding electrodes is not needed. The above considerations hold true in the simplified case of nearly uniform conductivity along all the insulator string, in the absence of corona. In practice, however, non-uniform contamination can occur, with higher pollution accumulation at the high voltage side due to the influence of electric field in DC. This has been shown in tests performed inside a dust chamber on an ‘I-string’ of 7 cap & pin insulators, as shown in Fig.4. This further contributes to reducing voltage non-uniformity along the string.
Confirmation of the negligible impact of shielding electrodes in DC has been demonstrated by radio interference (RIV) measurements performed on a 21-unit insulator string, with and without shielding electrodes (see Fig. 5). Results confirm that influence of shielding electrodes is negligible up to about 450 kV and then actually becomes detrimental once corona starts from the shielding electrodes themselves.
Still, shielding electrodes of limited size and different from those adopted in AC may be necessary to prevent corona from insulator fittings. Beyond giving rise to unacceptable RIV, this can also contribute to enhancing pollution. Fig. 6 refers to tests inside a dust chamber with corona generated by a sharp 15 cm long needle projecting from the conductor, close to the line end of the insulator.
Simplified Representation for Electric Field Calculation
As example, reference is made to composite insulators for UHV applications. The design of suspension insulators adopted for the Chinese 1000 kV AC system is shown in Fig. 7.
Similar shielding configurations were adopted for a ±800 kV DC line, as shown in Fig. 8
Since the purpose of such calculations is to emphasize the different role of grading rings in AC versus DC, a simplified schematic of the composite insulator is as follows: Total insulator lengths and size of the shielding electrodes were taken as in Fig. 8, both for AC and DC. The composite insulator is represented by a rod, without sheds, with a 90 mm diameter of homogeneous material having a dielectric constant (ɛ) of 3 and conductivity (σ) of 10-13 S/m. For surrounding air, ɛ of 1 and conductivity σ of 10-14 S/m were assumed in the simplified case of absence of significant ionization by corona. Surface conductance was simulated assuming a pollution layer of 1 mm having same conductivity along the entire insulator. Different conductivities of the pollution layer were considered, representing clean laboratory conditions (ideal case) as well as weathered and lightly polluted insulators that had lost part of their hydrophobicity. Voltage distribution was evaluated by ANSOFT under AC and DC at the same voltage of 800 kV. To obtain an indication of the field gradient along the surface, reference was made to the voltage derivative along the surface
Calculation results are reported in Figs. 10, 11 and 12, which compare the voltage distribution and its derivative in AC and DC, with and without shielding electrodes. Namely, the following cases were simulated for the composite insulator in Fig. 8, suitably schematized:
• Application of DC voltage without rings;
• Application of AC voltage without rings;
• Application of DC voltage with rings;
• Application of AC voltage with rings.
In the ideal case, with conductivity of the surface layer equal to the assumed conductivity of the insulator simulation (10-13 S/m), voltage distribution under AC and DC voltages are superimposed and similar to those evaluated for the same insulator. Surface current is very low, with an order of magnitude of some 10-12 A and the voltage distribution under DC is practically identical to the typical ‘capacitive’ AC one. Also, influence of the shielding electrodes is the same.
The situation changes rapidly considering surface conductivities more representative of weathered and lightly polluted conditions. Assuming a conductivity in the layer of 10-10 S/m, surface currents with an order of magnitude of 10-9 are obtained, which are already sufficent to differentiate the DC distribution from the AC one, as shown in Fig. 11.
Already with conductivity in the layer of 10-7 S/m and surface current along the layer of 0.02 μA, surface current dominates voltage distribution. This leads, in the assumed case of uniform conductivity in the layer, to linear voltage distribution for DC. But voltage distribution in AC remains dominated by capacitive parameters, as shown in Fig. 12.
While shielding electrodes continue to control the distribution in AC, they are not influencial in DC. Even if the case examined of uniform conductivity along the layer is a simplified one, calculations indicate that, for DC, voltage distribution in the case of weathered insulators can be considered as controlled by the resistive parameters on the surface. There is no influence from the shielding electrode.
The behaviour of insulators under very clean/ideal conditions is similar in AC and DC, with shielding electrodes having the same role in terms of mitigating electric field. However, as soon as the insulators become weathered, with surface leakage current on the order of magnitude of 1 μA, potential distribution for DC is dominated by surface conductance. As a result, in the case of AC, shielding electrodes have significant influence on electric field and, in particular, may be required for composite insulators to avoid possibility of ageing due to dry as well as water-induced corona, e.g. allowable E-field of 0.42 kV/mm for more than 10 mm along the insulator surface and not exceeding 0.35 kV/mm on the end fitting seal have been recommended. By contrast, under practical conditions influence of shielding electrodes becomes negligible in DC. Shielding electrodes of reduced size may still be useful in DC if there is need to shield sharp metallic points on fittings to avoid corona. But they may not prove useful to control electric field along the insulator, as necessary for AC applications. An additional consequence is that corona and RIV tests on DC insulators, with the insulators new and in an ideally clean state, may not be representative of service conditions.
Actual insulator conditions may be more complex than that assumed in this example. For example, charges may accumulate in the space and along insulators as a consequence of ionization phenomena. Moreover, conductivity along the surface could be non-uniform due to different extent of contamination, e.g. higher accumulation at live side versus earth side. However, in all cases, surface condition will continue to dominate voltage distribution, thereby making negligible the capacitive influence due to shielding electrodes.
Summary & Conclusions
Shielding electrodes of different form are usually foreseen and prescribed for AC lines and cover the following three main functions:
1. To eliminate corona from insulator fittings (corona shielding function);
2. To grade the voltage potential along the string in order to limit the radio interference from insulators and for composite insulators to reduce ageing phenomena (grading function);
3. To limit the effect of power arc (protective function).
For DC, possible smaller and specific shielding electrodes may be necessary to cover function A, but they are useless in practice to cover function B. This is because, for weathered insulators with surface currents on the order of magnitude of hundreds of nA or more, surface conductivity dominates over voltage distribution. The influence of shielding electrodes remains valid only in the ideal condition of new and very clean insulators and with extremely low surface currents – a situation not representative of most service conditions. In this respect, laboratory RIV and corona tests made with insulators under ideal conditions may not be representative and may need to be reconsidered from a standards point of view. Shielding electrodes are also less necessary in DC to cover function C, being that power arcs are less of a problem in DC.