Partial Discharges of High Frequency Transformer for Space Application in Near Vacuum

Abstract

This paper presents partial discharge measurements and a model-based investigation for corona activity on a prototype of a high frequency transformer from Flux A/S, designed to operate on a space travelling vessel. Previous iterations of the prototypes failed external partial discharge tests, and the environment that the transformer was designed to be installed and tested in, was suspected to cause the partial discharge to onset. In normal operation on the space vessel (and during the required design-tests), the transformer has its terminals exposed to the surrounding atmosphere, whose pressure varies from standard atmospheric pressure to near vacuum conditions during the launch of the space vessel. The most critical condition for external partial dielectric breakdown will be encountered under its intended operation, similar to the knee point in the Paschen curve. The conditions at this point yields the largest stresses on the dielectric, in this case the surrounding atmosphere of the transformer. A predictive method for evaluation of corona activity onset is therefore included. This is done based on calculation of the effective ionization coefficient along the critical electrical field line, $\bar{\alpha }$ obtained by Finite Element Method (FEM) models and is evaluated for a given set of pressure values to emulate the intended operating conditions of the transformer. $\bar{\alpha }$ itself is a function of the electrical field strength along the critical line, as well as the relative air density. The FEM models of the transformer were therefore designed to calculate the electrical field strength distribution around the transformer, and to locate the critical field line. It was found that 2D models yielded a satisfactory accuracy for the intention of the simulation, and that 3D models would only yield a slightly improved accuracy for a substantially increased computational burden.

For the partial discharge measurements, the test voltage was 1 kV at 50 Hz, and the geometry of the transformer and distances to objects in proximity to the transformer will remain constant, leading to the electrical field strength distribution also being constant. The only varying parameter which is affecting $\bar{\alpha }$ is the variation of the atmospheric pressure. The corona activity onset condition is evaluated by evaluating the integral value of $\bar{\alpha }$ over its region that yields a net positive ionization and compare this value with the criteria for the Townsend mechanism. The simulated pressure range includes only values that are realizable with the available equipment in the HV-laboratory of Aalborg University, which is 1.0 to 0.2 bars of absolute pressure. This is done with the intention to compare the predictive method for corona onset with actual partial discharge measurements. The PD measurements showed that no external partial discharge activity was present for the given experimental conditions and pressure range, and the new design of the transformer prototype was therefore improved.