Impulse-driven surface breakdown data: a Weibull statistical analysis

Mark Wilson, M Given, Igor Timoshkin, Scott MacGregor, Tao Wang, M.A. Sinclair, K.J. Thomas, Jane Lehr

Research output: Contribution to journalArticlepeer-review

25 Citations (Scopus)
683 Downloads (Pure)


Surface breakdown of oil-immersed solids chosen to insulate high-voltage, pulsed-power systems is a problem that can lead to catastrophic failure. Statistical analysis of the breakdown voltages, or times, associated with such liquid-solid interfaces can reveal useful information to aid system designers in the selection of solid materials. Described in this paper are the results of a Weibull statistical analysis, applied to both breakdown-voltage data and time-to-breakdown data generated in gaps consisting of five different solid polymers immersed in mineral oil. Values of the location parameter γ provide an estimate of the applied voltage below which breakdown will not occur, and under uniform-field conditions, γ varied from 192 kV (480 kV/cm) for polypropylene to zero for ultra-high molecular weight polyethylene. Longer times to breakdown were measured for UHMWPE when compared with the other materials. However, high values of the shape parameter β reported in the present paper suggest greater sensitivity to an increase in applied voltage – that is, the probability of breakdown increases more sharply with increasing applied voltage for UHMWPE compared to the other materials. Analysing peak-applied-voltage data, only PP consistently reflected a low value of β across the different sets of test conditions. In general, longer mean times to breakdown were found for solid materials of εr more closely matched to that of the surrounding mineral oil
Original languageEnglish
Pages (from-to)2449 - 2456
Number of pages8
JournalIEEE Transactions on Plasma Science
Issue number10
Early online date23 Jan 2012
Publication statusPublished - Oct 2012


  • weibull statistical analysis
  • breakdown voltage
  • pulse power systems
  • flashover
  • dielectric breakdown
  • weibull distribution

Cite this