An Engineering Model for Predicting Probabilities of Dielectric Breakdown in Gas-insulated Systems

In high-voltage engineering the so-called “streamer integral” is used widely to predict dielectric breakdown voltages of gas-insulated systems. The simple, deterministic nature of the model facilitates application to complex system geometries but comes at a price: most notably, the physical relation between the predicted “inception voltage” for breakdown and the experimentally-observable, statistical behavior of a system is lost. Consequently, the inception voltage is typically associated ad hoc with the voltage, at which breakdown occurs with two percent probability. This complicates the determination of the model parameters and limits their applicability across diverse system geometries. In this work, we introduce a probabilistic extension of the streamer integral model, which permits predicting how the probability of breakdown depends on the applied voltage. An initial validation is performed based on breakdown experiments in air with lighting-impulse voltages.