The breakdown mechanism of compressed SF6 in gas insulation is known to be controlled by stepped leader propagation. This process is reasonably well understood for strongly non-uniform insulation gaps (‘point-to-plane’) and in the absence of pre-breakdown discharge activity (‘corona stabilization’). Open questions still remain for weakly non-uniform insulation gaps with small electrode protrusions (particles, surface roughness), in which pre-breakdown partial discharge (PD) activity is present. This paper presents a first attempt to derive a consistent picture under these conditions, which are characteristic for practical gas insulation systems. Experiments were carried out in a uniform field gap with a short protrusion on one electrode. This configuration was studied at various pressures from 0.1 to 0.5 MPa and both polarities using electrical and optical diagnostics. The results are interpreted using a quantitative model and order-of-magnitude estimates. The emerging picture allows prediction of most of the technically relevant aspects of the discharge processes and their main parameter dependences. It comprises statistical time lags, formative time lags including pre-breakdown PD activity and breakdown fields as a function of gas pressure, protrusion length and polarity.

The breakdown mechanism of compressed SF6 in gas insulation is known to be controlled by stepped leader propagation. This process is still not well understood in uniform and weakly non-uniform background fields with small electrode protrusions, such as particles or surface roughness. In a previous publication an investigation of partial discharges and breakdown in uniform background fields that focused on streamer and leader inception mechanisms was presented (Seeger et al 2008 J. Phys. D: Appl. Phys. 41 185204). In this paper we present for the first time a physical leader propagation model that consistently describes the observed phenomena in uniform background fields in SF6. The model explains two different types of leader breakdown; these can be associated with the precursor and the stem mechanisms. It also yields the parameters of stepped leader propagation, which include step lengths, associated step charges, step times and fields and temperatures in the leader channel. Further, it explains the features of arrested leaders in uniform background fields. The model predicts the range of parameters under which arrested and breakdown leaders occur in good agreement with the experimental data.

No abstract

SF 6 has for decades been widely used in high voltage insulation and switching applications, e.g. in gas insulated switchgear. Despite its widespread use some important parameters, like the properties of streamers, are still not sufficiently understood. Since breakdown in SF 6 always occurs via the streamer-leader transition the streamer properties are decisive for leader inception and, therefore, breakdown of the insulation. Important parameters are, for example, the streamer radius and the streamer propagation length of arrested streamers. Such properties enter in breakdown prediction models. In the present study the streamer radius and the propagation length were investigated experimentally at 50 and 100 kPa for both polarities using strongly and weakly non-uniform background fields. No experimental information was available so far for negative polarity. The resulting streamer radius scaling agrees with previous experimental results for positive polarity and with expectations from breakdown models for negative polarity. These results were similar for strongly non-uniform and weakly non-uniform background fields. A difference between the two setups was observed for the streamer lengths. It was found that for strongly non-uniform fields the streamer length scales as expected with the critical electric field but with a different field for weakly non-uniform background fields. This was similar for both polarities.

No abstract