Epoxy res~ a major component of solid electrical insulation systems, degrades when subjected to electrical discharges. Identification of the epoxy resin degradation mechanism might indicate improvements which can be made in the chemical formulation of the resin to enhance the insulation systems. Samples of a commonly used epoxy res~ bisphenol-A epoxy crosslinked with phthalic anhydride, were manufactured and then aged by applying lOkV AC to an electrode 2mm above the resin surface. The following experimental conditions were altered: manufacturing system: (i) moulded slab and (ii) slice cut from moulded cylinder; atmosphere: (i) nitrogen, (ii) dry, (iii) normal and (iv) moist air; high voltage electrode:(i) brass and (ii) copper. In addition, ageing due to chemical, thermal and radiative stressing was also examined. The changes in the stressed resin samples were determined using the following techniques: diffuse reflectance Fourier Transform infrared spectroscopy (DRIFT), attenuated total reflectance FTIR (ATR-FTIR), thermogravimetric analysis (TGA), thermogravimetric FTIR (TG-FTIR) and atomic force microscopy (AFM). The changes in the electrode materials were determined using X-ray diffratometry (XRD) and Fourier Transform infrared (FTIR) techniques. The method of production of the specimens was shown to affect the degradation. Silicone release agent, used in the moulding of the resin slabs, was found on the surface of degraded moulded resin slabs: the contamination of the moulded samples was not detectable prior to partial discharge stressing. Crazing and flaking of the stressed resin surface were found on the moulded slabs but not on the slices of resin. Anhydride, acid and amine species were identified on the surface of the electrically stresses resin slices. Chemical reactions accounting for the changes found on the surface of the stressed resin slices are given. The radical species formed by methyl group dissociation, reacting with hydroxyl and activated oxygen species, lead to the formation of linear anhydrides, acids and peracids on the resin surface. Reactive nitrous oxide species in the discharge atmosphere react with the resin to form amines. Zinc formate dihydrate was identified on brass electrodes after the resin ageing process, whilst basic copper nitrate was identified on copper electrodes. The difference in deposit found on the electrode indicates that zinc reacts with carbon species from the discharge environment; copper reacts, not with carbon species, but with nitrogen species. The variation in chemical interaction at the high voltage electrode, dependant upon electrode material, has been found to correlate with changes on the resin surface. Correlations are made between the effects of partial discharge and other stresses applied. None of the applied stresses generated the anhydride structure found in partial discharge stressed resin samples. However, in common with p.d. stress, UV radiation increases the level of crosslinking in the epoxy resin and produces carbonyl structures, nitric acid fumes produce acid, peracid and nitroso structures.

Electrical trees have long been the interest of the electrical insulation community due to their role in power systems equipment failure at locations where high divergent fields might arise due to impurities, contaminants or voids. Even through trees take a long time to grow in real life, they can be grown experimentally in shorter times under various conditions so that their growth characteristics can be investigated. Different samples have been fabricated to investigate the effects of interfaces in electrical tree propagation. Initially, the impact of an interface perpendicular to the electric field, and the interface position, thickness and/or composition on the polymer’s lifetime was investigated. In the results acquired, the positive impact of interfaces positioned perpendicular to the electric field was evident: increasing the samples’ time to breakdown as well as the electrical tree inception time. Due to the encouraging results, further investigation has been focused on interface modification and how this might be used to control the electrical tree growth as well as the samples’ time to breakdown. Altering the interface’s surface roughness using a number of different methods was carried out. Results were graphically and statistically analysed so that the any conclusions are robust, and uncertainties clear. The statistical analysis used by generating regression model equations was a novel method to predict how different electrical tree parameters were affected/affecting by others. In this way the dielectric’s lifetime could be predicted with a certain level of confidence. The modification of the interface by coating the surface with either a thin layer of pure or nano-filled (hexagonal Boron Nitride) epoxy resin resulted in it being impervious thus preventing the electrical tree to propagate through it. This was a novel method that showed that specific modification methods can significantly enhance the dielectric’s lifetime when applied appropriately. Details of new sample fabrication techniques are described which enable better control of the materials and interfaces, and data on tree length growth characteristics are discussed.