Critical electric field strength and effective ionization coefficient measurements of nitrogen-oxygen mixtures with variable mixing ratio

Air is the most commonly used insulation gas in high voltage applications: Mostly ambient air for overhead power lines and air insulated substations, but also synthetic air in medium voltage metal-enclosed switchgear at slightly elevated pressures. Moreover, nitrogen and air have been proposed as suitable alternatives to SF6 in high-voltage insulation and interruption applications of gas insulated switchgear [1, 2, 3]. Air has a weaker insulation performance than SF6 and other proposed, strongly electron attaching, gases, but it has the advantage that its constituents are environmental benign and thus almost non-problematic in use. The motivation of this investigation is to find the optimum mixing ratio of the constituents of air (i.e. N2, O2, CO2, Ar) for electrical insulation. In this contribution we report on a systematic analysis of binary oxygen-nitrogen mixtures with varying mixing ratio, which has not been done before. The comparison of the different gas mixtures is based on macroscopic discharge parameters (swarm parameters), namely the critical density normalized field strength (E/N)crit and the effective ionization rate below and around (E/N)crit. The swarm parameters are measured in a Pulsed Townsend (PT) experiment. The experiment measures with a high precision and operates with a high degree of automation. This experiment is well suited to investigate a vast number of different gas mixtures. As expected, an increase of the attachment rate is observed with increasing oxygen content. Also (E/N)crit increases with increasing oxygen content, but shows a nonlinear behaviour. (E/N)crit increases by 7% if the oxygen content is increased from 8% to 16%, but only by 5% for an increase from 21% to 40%. It is known that oxygen shows pressure dependent detachment, which is an undesired property of an insulation gas. The role of oxygen cannot be answered conclusively today. Comparison of our measurement results to previously reported measurements and simulations are done and show the need for further investigations. In summary, it can be stated that, based on the results reported here, an improvement of the insulation performance in homogeneous and slightly inhomogeneous geometries of maximum 5% may be achieved by increasing the oxygen content compared to synthetic air. This is an encouraging basis for further optimization including mixtures with three or even more components.