Reduction of Power Transformer Core Noise Generation Due to Magnetostriction-Induced Deformations Using Fully Coupled Finite-Element Modeling Optimization Procedures

This paper focuses on the development of an algorithm for the prediction of transformer core deformation, using a fully coupled magneto-mechanical approach. An imposed magnetic flux method coupled with finite-element modeling is employed for the magnetic resolution. The constitutive law of the material is considered with a simplified multi-scale model describing both magnetic and magnetostrictive anisotropies. Magnetostriction is introduced as an input free strain of a mechanical problem to get the deformation and displacement fields. A 3-D estimation of acoustic power is carried out to evaluate the global core deformation and acoustic noise level. The numerical process is applied to a three-phase transformer made of grain-oriented FeSi core, leading to a set of noise estimation for different flux excitations and core geometries.