In-situ characterisation of the insulator leakage current of high voltage direct current transmission lines

Abstract

Leakage current (LC) monitoring of high voltage transmission line insulators is of interest as it is indicative of the degradation of the insulators, as well as the potential for flashover. Several high voltage direct current (HVDC) transmission schemes are being constructed worldwide, and issues of dimensioning and specification of the transmission line insulators are an area of debate. Further, advancements in composite insulator materials are touted to offer a better pollution performance; however, there is little operational information available in the literature to understand the benefits completely. This research addresses topics that relate to the characterisation of the LC under HVDC conditions. There is ample experience with LC under HVAC conditions, but far less so for HVDC. HVDC lines are harder to monitor than alternating current (ac) lines, since the conventional current transformers cannot be used to measure dc. Presently, there is no commercially available, non-intrusive, clamp-on type device for HVDC transmission line insulator LC measurements. Power utilities, like South Africa’s Eskom, have operational requirements for such a device for two critical applications. First, for continuous real-time monitoring of HVDC line insulators, and second, for use by live line workers to determine if it is safe to work on energised insulators. This thesis investigates various LC sensing techniques as well as their respective advantages and disadvantages. Commercially-off-the-shelf (COTS) sensors are used to perform initial LC measurements on insulators. The author subsequently developed a new dc LC sensor for HVDC transmission lines, which determines the current indirectly by measuring the magnetic field associated with it. The sensor was deployed in the Cahora Bassa HVDC transmission line scheme for an extended period. Such results are rare, and not readily available in the public domain. The results show that this sensor can measure LC across the full anticipated range non-intrusively; hence, it can be readily used by live line workers to determine the probability of flashovers. Two case studies have also been performed at Eskom’s corona cage high voltage testing facility on HVDC glass insulators. The first case study investigates the relationship between current and increased pollution layer conductivity of a glass insulator. The measurement results show that the LC increases with an increase in contamination level when the voltage level is kept constant, but also that flashover will occur for LC exceeding a critical threshold of 1.4 mA, irrespective of the conductivity layer applied to the insulator. The second case study investigates the probability of flashover occurring as a function of LC at a constant equivalent salt deposit density (ESDD) level of 0.03 mg/cm2. The results show that for currents exceeding 0.6 mA there exists a possibility of flashover occurring. This quantification of the LC that can lead to flashover is important as it indicates safe scenarios when live line workers can perform maintenance on energised HVDC transmission line insulators. In order to better understand LC behaviour on insulators, finite element method (FEM) simulations have been performed on dc insulators, and compared with that of ac insulators. Voltage, electric field and current density distributions are simulated in COMSOL Multiphysics for clean and uniformly polluted ac and dc energised insulators with layer thickness of 0.02 mm and conductivity of 0.07 S/m. To the best of the author’s knowledge, no voltage, electric field and current density distribution simulations have been conducted for glass dc insulators in literature. When insulators are polluted, a dramatic increase in current density for both the ac and dc insulators are observed, with the ac and dc insulator exhibiting current densities of 4.81 x 103 A/m2 and 5.34 x 103 A/m2, respectively. It is observed that the electric field on the surface of a polluted insulator is higher on an energised dc insulator compared to that on an energised ac insulator. This can lead to a dc insulator having a higher flashover probability than the ac insulator under the same excitation voltage. It is expected that arcs (that will lead to eventual flashover) will first emanate from the energised end of the insulator, since that is where the highest electric field occurs for both types of insulator. From the electric field simulations, it is also observed that interim maxima exist on the sheds of a polluted ac and dc insulator. These higher electric fields on the polluted insulator surface can lead to heating on the insulator surface and to the formation of dry bands, which increases the probably of a flashover occurring. These interim maxima are not observed on the clean insulator case. The simulations further show that the LC leakage current on ac energised insulators is purely resistive (conduction current) and that the contribution of capacitive currents on the current density is negligible. Various conductivities of air have been studied to emulate different altitudes and ambient humidity conditions. It is concluded that an increase in the air’s conductivity has a negligible effect on an insulator’s performance. The simulated current in COMSOL Multiphysics are compared with the laboratory LC measurements and the results agree well. The research concludes with in-situ measurements to characterise the behaviour of LC on the Cahora Bassa HVDC transmission lines, as representative of HVDC schemes at large. For this purpose, a glass and composite insulator have been installed on the terminal tower of the transmission line located at the Apollo Converter Station in Johannesburg, South Africa. Shunt resistors have been installed on both insulators, and the prototype sensor on the glass insulator, to measure the LC activity. (The prototype sensor measurements have not been used during the analyses of the data, but agreement with the shunt resistor measurements are shown.) Insulator LC measurements have been performed over an 11-month period. The influence of temperature, humidity, dew, rain and the HVDC line’s voltage and current on the behaviour of LC have been investigated. The measurements show that the composite and glass insulator LC behaviour is similar, except in cases of high humidity or rain. At the onset of rainfall and humidity (>90%), elevated LC levels are observed on glass insulators, while composite insulators demonstrate lower LC levels under the same conditions. LC activity is observed in the presence of continuous rain as well, but with low magnitude. A statistical analysis of all data shows the following correlation coefficients between LC and climatic and line conditions over the 11-month period for the two types of insulator: • LC and temperature: 0.40 (composite) and -0.12 (glass); • LC and humidity: -0.28 (composite) and 0.13 (glass); • LC and rain: -0.03 (composite) and 0.43 (glass) • LC and line voltage: 0.06 (composite) and -0.07 (glass) • LC and line current: 0.01 (composite) and 0.11 (glass) It is evident that the glass insulator’s LC increases when rain is present, but that rain has little effect on LC composite insulators. This may be the result of self-cleaning of the insulator that occurs due to its hydrophobicity properties. The HVDC transmission line current and voltage fluctuations have been found to have negligible influence on the LC levels. Overall, during the 11-month LC monitoring period, it has been evident that elevated LC activity occurs more frequently in the spring-summer period, which is the rainy season for Johannesburg, than in winter. Interestingly, under nominal weather conditions of no rain and low humidity, the LC measurements for both types of insulators exhibit an almost square-wave behaviour with LC switching between ≈ 20 μA and ≈ 60 μA with short transitions on a daily basis. These swift transitions can be attributed to condensation on the insulators, which has been found to be the primary determinant of the LC levels on these contaminated insulators. The field tests show that the line is working better than first anticipated by Eskom engineers, as higher levels of LC had been expected. A novel, linear approximation has also been determined between LC measured during laboratory tests and actual HVDC field tests. It was observed that by using the initial developed equation, the LC calculated for the field measurements agrees well with the actual LC measurements conducted on the HVDC line. This approximation can be especially helpful to support the design of new HVDC lines In conclusion, a novel dc LC sensor prototype has been designed, tested, calibrated and validated for the measuring of HVDC LC non-intrusively on HVDC transmission lines. Furthermore, the insulator LC of the Cahora Bassa transmission line has now been quantified and analysed over an extended period for both glass and composite insulators. These measurements have not previously been documented to this extent in the literature. Finally, a satisfactory correspondence between laboratory and live line LC measurements has also been established, which aids future research and development in this field.