In this paper a mathematical model, based on experimentation, to estimate the axial moisture distribution in power transformer windings was developed. The model considers the variables of moisture content in oil, top oil temperature, as well as the transformer's design parameters and ageing of the insulation system. The model was validated using a small-scale experimental setup, which represents a 75 MVA transformer. The developed model proved to be more accurate than existing models. It considers the axial moisture distribution in paper, in 4 thermal zones equidistantly distributed along the height of the winding. This paper compares the results of the proposed model with those provided by other authors.

Power transformers are some of the most important equipment for the transmission and distribution of electric power. A single failure in a transformer causes disturbances in the electric network and may cause severe conflicts in hospitals, banks, industrial installations or urban areas in general. In Mexico, the transmission network is composed by 350 power substations and 2,580 power transformers. The capacities of these transformers are typically 375, 225 and 100 MegaVoltAmpere (MVAS), with a nominal tension of 400 kV, 230 kV and lower. Approximately 27% of these transformers have more than 30 years in operation. For this reason, it is important to observe and register the amount and type of failures that have presented the transformers in the country. Table 1 shows the type of transformer failures from 1997 to 2007 (CFE, 2010).

Detecting transformer failures at early basis represents enormous economical and technical advantage for a utility company. One approach reported in the literature is the vibration analysis for the detection of mechanical failures in transformers. The basic idea is the characterization of the normal vibration during the operation of the transformer, and the recognition of variations in the vibration patterns of the transformer when a failure is present. This paper presents the development of a probabilistic vibration model used for the detection of incipient failures in transformers. Vibration measurements are taken all around of the transformer tank and a probabilistic model is constructed using automatic learning algorithms developed in the Artificial Intelligence community. The models are Bayesian networks that relate probabilistically all the variables in the experiments. Later, inference algorithms are used to estimate on-line, a probability of a failure in the transformer. This project is in collaboration with Prolec General Electric, the largest constructor of transformers in North America. Experiments were carried out at Prolec GE laboratories on a power substation transformer (PST). A discussion of the experiments and their results are included in this paper.