An analysis of copper transport in the insulation of high voltage transformers

Examination of the paper insulation and copper stress braiding during stripdown of a number of Current Transformers (FMK type 400kV) has revealed the presence of dark deposits. Copper foils are often interspersed within layers of paper insulation and mineral oil found in transformer windings. The dark deposits were often found in association with these foils, affecting several layers of paper in addition to the layer in contact with the copper foil. This thesis describes the research undertaken to identify these deposits and establish a mechanism for the transportation through the paper layers. Preliminary investigation using scanning electron microscopy (SEM) in conjunction with energy dispersive X-ray analysis (EDX) has shown these dark deposits to be copper based. X-ray photoelectron spectroscopy was used to show that the transport of the copper deposit through the paper insulation was working under the influence of a diffusion controlled process, related to Fick's law. Laboratory studies in support of work designed to eliminate the problem have shown that corrosion of copper occurs in mineral oils containing a trace of oxygen. This corrosion is non protective in character and leads to migration of copper into adjacent layers of paper. It has been shown that the transport of copper through several layers of paper can be measured by XPS and that the concentration from one paper winding to the next declines in accord with Fick's law for non-steady state diffusion. Measurements of surface concentrations by XPS correlate well with measurements made with atomic absorption spectroscopy on solutions of extracts of the contaminated paper. The laboratory measurements have allowed determination of the diffusion coefficients and activation energy for the transport process and thus give a basis for interpretation of the diffusion profiles found in the transformer in terms of time and temperature of operation. The diffusion process is temperature dependant. The results have been used to produce long term prediction curves. (author)Available from British Library Document Supply Centre-DSC:DXN044619 / BLDSC - British Library Document Supply CentreSIGLEGBUnited Kingdo