Propagating uncertainty using IEEE STD C57.104-2019 dissolved gas analysis methodology for transformers

Research output: Contribution to journalConference articlepeer-review

Abstract

Dissolved Gas Analysis is a well-established tool for transformer health monitoring with published Standards to help with its interpretation. However, even though it is known that there is measurement uncertainty regarding the true value of sampled gas, there is less available guidance regarding the practical implications. This paper proposes a method for propagating the measurement uncertainties through the methodology presented in the IEEE Std C57.104-2019 to provide the degree of support for its potential outputs. This is done relying on the simplifying assumption that the measurement uncertainty can be expressed as a symmetric triangular distribution for a given gas sample, and that gas samples are independent. The joint probability function is derived in general terms analytically and then a stratified sampling approach is proposed to numerically solve the function. In addition, a modification is made to allow for a more constrained sampling space by deriving and using a simplifying marginal probability without impacting accuracy. These are presented via a use of a case study to demonstrate the efficacy of the proposed approaches.
Original languageEnglish
Pages (from-to)396-401
Number of pages6
JournalIET Conference Proceedings
Volume2023
Issue number46
Early online date9 Apr 2024
DOIs
Publication statusPublished - 6 Aug 2024
Event23rd International Symposium on High Voltage Engineering (ISH 2023) - Glasgow, United Kingdom
Duration: 28 Aug 20231 Sept 2023

Keywords

  • constrained sampling space
  • propagating uncertainty
  • measurement uncertainty
  • stratified sampling approach
  • symmetric triangular distribution
  • transformer health monitoring

Fingerprint

Dive into the research topics of 'Propagating uncertainty using IEEE STD C57.104-2019 dissolved gas analysis methodology for transformers'. Together they form a unique fingerprint.

Cite this