Skip to main navigation Skip to search Skip to main content

Predicting the no-load noise of power transformers through the multiscale boundary-coupled finite element model

  • Fan Zhang
  • , Dehua Zhao
  • , Zhenjie Zhao
  • , Feng Li
  • , Jihu Zheng
  • , Ping Zhang
  • , Shengchang Ji
  • Xi'an Jiaotong University
  • Ltd

Research output: Contribution to journalArticlepeer-review

1 Scopus citations

Abstract

The no-load noise from the magnetostriction of electrical steels significantly influences the power transformers' sound pressure level (SPL). This paper introduces a multiscale boundary-coupled finite element model to predict the no-load noise of power transformers. The proposed model integrates a three-dimensional (3D) magnetic model to calculate the main flux and its harmonics, a 3D anisotropic magnetostriction model of the core, a two-dimensional (2D) axisymmetric model to represent the vibro-acoustics of windings, and a 3D model of the whole transformer with the surrounding air domain. In the 2D axisymmetric model, the sound emission from windings in oil is characterized by a cylindrical sound source boundary, and the top and bottom clamping rings are defined as displacement boundaries. Such boundaries are coupled between the 2D and 3D models to ensure the transmission of vibration and pressure between the two models. The error in predicting the SPL of two 110 kV and one 220 kV transformer is only -0.4 dB(A), -2.9 dB(A), and 1.1 dB(A), respectively. The proposed model can be used to predict the SPL of newly designed transformers and implement noise-reduction strategies.

Original languageEnglish
Article number112706
JournalElectric Power Systems Research
Volume254
DOIs
StatePublished - May 2026

Keywords

  • Anisotropic magnetostriction
  • Finite element model
  • Multiscale boundary
  • No-load noise
  • Power transformers

Fingerprint

Dive into the research topics of 'Predicting the no-load noise of power transformers through the multiscale boundary-coupled finite element model'. Together they form a unique fingerprint.

Cite this