Gas-insulated systems under DC voltage have a different characteristic compared to AC voltage stress. Particularly some specific effects of DC voltage such as surface and space charges as well as polarity effects have a great influence on the electric performance of a gas-insulated system. The DC insulation system design has to cope with these specific characteristics. Besides, typical PD defects may occur in DC gas-insulated systems also, e. g. particles, protrusions, parts on floating potential and voids. The origins of these defects can be in manufacturing, in assembling of components and due to friction or vibration in operation. These defects may reduce the dielectric performance of the system; thus they have to be detected, located and identified. The phase resolved partial discharge (PRPD) pattern, commonly used for PD defect identification under AC voltage, cannot be applied for DC voltage due to the missing phase correlation. Therefore, basic investigations using another method, the pulse sequence analysis (PSA), were carried out. PSA is based on the comparison between magnitudes or magnitude differences respectively and time differences of consecutive partial discharge impulses. The corresponding quantities are correlated to PSA patterns. For each type of defect different PSA patterns can be established. This investigation evaluates the suitability of six evolved PSA patterns. In addition to commonly used PSA methods, further related quantities are used for creation of additional patterns to increase the diagnostic significance. By use of logarithmic axis scaling, charts are optimized for UHF-partial discharge measurements and defects with a higher variance of time differences. The considered defects are basically protrusions and free moving conductive particles. The tests were carried out at different gas pressures with a gas mixture of Nitrogen (N2) and Sulfurhexafluoride (SF6). The encapsulated test set-up consisted of a real-sized DC gas-insulated line (GIL) arrangement for DC voltage up to +/-550 kV, containing different insulators and a set of UHF sensors. The analysed PD pulse sequences were recorded by the UHF partial discharge measurement. Moreover, the impacts of the measuring method, the applied measurement system, the locations of the UHF sensors and the characteristics of each defect e.g. the particle shape, were examined. The aim of the investigation is a reliable detection of defects, their classification, respectively their identification by means of advanced PSA patterns. The ability of the method could be proven during onsite measurements on a DC GIL prototype installation, at which a PD defect was detected. By means of the knowledge of the PSA patterns gained at the investigations described, the PD defect could be identified.
展开▼
