| dc.description.abstract | The post digital era in high voltage impulse testing of transformers witnessed renewed activity in an otherwise dormant field due to the introduction of the transfer function (TF) approach. The TF method provides a more meaningful assessment of the entire impulse test data and is therefore superior to conventional fault diagnosis methods. However, despite its advantages, TF is not yet an approved method in any relevant standard. From an examination of the matter, it emerges that issues related to implementation and interpretation of TF are the two major bottlenecks. Solutions to these are urgently desirable if the true potential of the TF tool is to be exploited. The latter issue is discussed in this thesis.
The nature and structure of TF is governed primarily by the winding type, and to some extent by the configuration and connection of neighbouring windings. Knowledge of the TF structure is considered useful during TF interpretation. EHV transformers usually employ partial or fully interleaved windings, while disc or layer type windings are preferred at lower voltages. There are differences in the TF structure of these two winding types, and certain features specific to the TF of an interleaved winding have, over the years, been attributed to the increased series capacitance of interleaved windings. However, why this increase introduces such differences in the TF, why only a few natural frequencies are present in the TF of interleaved windings, etc., is still unresolved. Ascertaining these reasons can lead to improved understanding of winding behaviour in particular, and permit better diagnostics of transformers in service. Theoretical explanations for the structure of TF are presented in this thesis, which have, until now, neither been quantitatively analysed nor properly reasoned.
Considering the ladder network equivalent circuit representation (consisting of series capacitance, series inductance, series resistance, shunt capacitance, and mutual inductance) of the transformer winding, analytical expressions for neutral current and TF are derived in terms of the series and shunt capacitance. From these expressions, poles and zeros are determined, whose examination has yielded the following explanations for the observed differences in the TF structure of an interleaved winding:
It has been shown, for the first time, that in interleaved windings, pole–zero cancellation almost occurs, and this is the primary reason why its TF contains only a few low frequency poles.
Network loss causes reduction of pole magnitude and widening of the pole-two more factors that shape the TF. This makes it impossible to resolve higher frequency poles.
Advantages of employing the swept frequency method for direct estimation of TF-rather than the time domain route-are suggested, especially for interleaved windings. Problems associated with processing time domain data in such situations are also discussed.
These findings have been compared with TF obtained from time domain records computed using PSPICE. Further, practical measurements performed on a model transformer coil validate the theoretical results. The implications of these results on the diagnostic capability of TF, especially for interleaved winding transformers, are highlighted. Certain guidelines for data acquisition during impulse tests are also suggested, which can improve diagnostic capability.
In summary, it is believed that these results will lead to overall improvements in diagnostics of large HV transformers. | |
