Abstract
This paper computationally investigates partial discharges (PDs) in the form of self-sustained gas discharges. It presents two methods for predictive modeling: (1) a new low-fidelity algorithm for the PD inception voltage is introduced. The method is volume-resolved and describes both the strength of the self-sustained Townsend mechanism as well as the conventional streamer (or bulk) mechanism. It also intrinsically computes the inception region, i.e. the region where a first electron also leads to a discharge. (2) We apply a high-fidelity plasma model based on kinetic Monte Carlo, which self-consistently resolves the plasma dynamics during the PD process. The two models are complementary in the sense that the low-fidelity model provides the when and where the PD occurs, while the high-fidelity model resolves the PD process itself, starting from the first electron. Prediction and quantification of the PD processes is provided for four application cases: (1) protrusion-plane gaps, (2) spherical voids, (3) twisted wire pairs, and (4) triple junctions. Validation of the low-fidelity method is done through comparison with published experiments (where available), as well as virtual verification through comparison with the high-fidelity plasma model.