Assessing the reliability of linear dynamic transformer thermal modelling

X. Mao, Daniel Tylavsky, G. A. McCulla

Research output: Contribution to journalArticle

10 Scopus citations

Abstract

Improving the utilisation of transformers requires that the hot-spot and top-oil temperatures be predicted accurately. Using measured (noisy) data to derive equivalent linear dynamic thermal models yields performance that is superior to the ANSI standard model, but the reliability of these model coefficients must be assessed if the user is to have confidence in the model. By adding arbitrarily large amounts of data in the modelling process it was expected to make the reliability measures of these models arbitrarily small. When this did not happen, an investigation began that showed why there is a limitation to the accuracy of models derived from noisy data. It is also shown that a standard technique for assessing the reliability of model coefficients is invalid because of the absence of unmeasured driving variables. An alternative method for assessing transformer model reliability is provided.

Original languageEnglish (US)
Pages (from-to)414-422
Number of pages9
JournalIEE Proceedings: Generation, Transmission and Distribution
Volume153
Issue number4
DOIs
StatePublished - Jul 1 2006

ASJC Scopus subject areas

  • Electrical and Electronic Engineering

Fingerprint Dive into the research topics of 'Assessing the reliability of linear dynamic transformer thermal modelling'. Together they form a unique fingerprint.

  • Cite this