Differences in streamer initiation and propagation in ester fluids and mineral oil

The properties of ester fluids and mineral oils differ in many cases in a qualitative way. This means that these liquids may look like very similar in some experiments while in others they behave very differently. These differences must be understood and taken into account in the design of ester filled transformers to avoid unexpected failures. Classic models such as the two-parameter Weibull statistics of breakdown voltage are in some respects too simple to properly catch the underlying physics. We discuss the importance of segregating initiation and propagation processes to properly interpret results of experiments with esters and mineral oil. In transformer design it is in some parts more important to avoid initiation while in others it is more important to stop propagating streamers. Furthermore, to obtain low probability statistics from high probability breakdown experiments, both average and spread of breakdown voltages are required. The difference in spread of initiation and propagation voltages is discussed. Esters and mineral oils behave in a qualitatively similar way regarding initiation while they differ substantially in the propagation process. In this paper we compare some different experiments in terms of initiation and propagation.