High speed catastrophic events in technical insulating liquids

As liquid breakdown modeling is still in its infancy, dielectric breakdown tests remain one source of input regarding the design of an insulation system for power apparatus. The relevance of small scale tests has long been known to be limited. It is known that designers would never dare to use the strengths seen in such tests. The paper argues that one aspect that may describe the reason for the discrepancy is that the physics controlling the breakdown are different for the small scale tests when comparing to the breakdowns observed in large technical insulation systems. Experimental data are presented showing that catastrophic events of technical interest arise in different liquids at different voltage levels. Voltage differences obtained with the impulse overvoltage method are not seen when 'traditional' small scale breakdown tests are performed.