It is important that research into the development of breakdown models of electrical insulation should be encouraged, with the aim of producing more reliable designs of electrical equipment and alleviating the need for extensive laboratory testing. At the present stage in development of breakdown prediction, the two most useful models are the integration method and the column model. Of these two, the integration method is perhaps the best, because it can be used for any electrical insulation, whereas the column model is restricted to positive impulse breakdown in air. This thesis will demonstrate the applicability of the two models and act as a guide to their use. Experimental and analytical work will cover a range of standard and non-standard lightning and switching impulse waveshapes.
The integration method has been mainly used to predict breakdown characteristics of air insulation. Previously, its applicability to other electrical insulation has received little attention. A comprehensive set of experimental volt-time curves for transformer oil has been obtained, and these results have been submitted for evauation by the integration method. Analysis has shown that the integration method yields acceptable agreement between calculated and measured breakdown characteristics for times to breakdown less than approximately 50µs. For longer breakdown times, errors occur, because the U-curve effect of the V50 breakdown strength can not be modelled accurately. Steep-fronted volt-time curves of 11kV class surge diverters have also been subjected to similar analysis by the integration method. It was found that the results of only two standard tests - the front of wave test and the standard 1.2/50µs test - were needed to predict the complete volt-time curve of surge diverters by the integration method with an accuracy of better than 4%.
An improved method of predicting the breakdown voltage and volt-time characteristics of wood-porcelain combinations is described. The model accounts for the different non-standard voltage waveshapes that stress each element of the combination. The natural variability of component impedances and breakdown strengths is also modelled. The integration method is used to evaluate the breakdown, and the time to breakdown, of each element, from which are derived the insulation strength and the volt-time characteristic of the combination. This method can reproduce laboratory results to within 10% for a wide range of wood-porcelain combinations.
By treating the positive leader as the early stage of an arc, certain empirical formulae relating the electric gradient, the leader current, the arc temperature and the time of a per unit volume of arc can be calculated by solving the energy balance equation for an arc. The equations derived by this method compare favourably with formulae assumed by Jones and Hutzler, for the column model. One of the simplifying assumptions used to derive these equations was that Local Thermal Equilibrium (LTE) exists in the leader channel. Recently, Gallimberti has proposed that LTE does not exist in the leader channel. A theory of two electron populations, one in LTE and one in non-equilibrium co-existing in the leader channel, has been offered to link the two opposing theories.
Experiments are described which show that the switching impulse strengths of a rod-plane gap and tower window can be increased by placing a floating insulating barrier at the mid-gap region. Both the minimum and V50 breakdown voltages were increased by about 15% for positive switching surges; for negative switching impulses there was no perceived change. Field strength measurements at the plane were recorded and used to help explain this phenomenon. The increased strength is maintained even though the barrier has been punctured many times, and investigations revealed that the hole size in the plastic barrier was the determining factor.