Oil testing has found widespread use in assessing the quality of the insulation system of a transformer. Techniques such as Dissolved Gas Analysis (DGA) have been proven to be reliable in this task, but are expensive, laborious and time consuming. It is also not available for field use. This paper proposes the combined use of gas and optical sensing technology for the testing of transformer oil. It provides a low-cost, portable technology capable of fast and reliable field screening of transformers. The method consists of an oil handling and unit as well as a gas and optical sensor array. Gas sensing targets moisture, hydrocarbons and other volatile compounds dissolved in transformer oil; optical measurements provide information on the absorption properties of transformer oil, within a band of frequencies made possible by the recent advances in Blue-LED technology. These measurements are then combined through a pattern recognition system producing a collective decision on the state of the transformer. The performance of the method was evaluated on a database of 26 transformer oil samples, using a leave-one-out validation technique. Our results indicated that the method was capable of categorizing transformer oil into three classes, acceptable, marginal and bad with reasonable accuracy, based on the acidity and furfuraldehyde levels estimated from gas-optical measurements. The results need to be further validated using a larger database of transformer oil samples.
©2011 Walter de Gruyter GmbH & Co. KG, Berlin/Boston