An alternative approach to model mechanical stress effects on magnetic hysteresis in electrical steels using complex permeability

• Published in: Computational materials science 2019-08, Vol.166, p.96-104
• Main Authors: Baghel, A.P.S., Sai Ram, B., Daniel, L., Kulkarni, S.V., Krebs, G., Blumenfeld, J.B., Santandrea, L.
• Format: Article
• Language: English
• Subjects: Complex permeability; Electrical steel; Engineering Sciences; Finite element method; Iron loss; Magneto-elastic behavior
• ISSN: 0927-0256; 1879-0801
• DOI: 10.1016/j.commatsci.2019.03.048

•A complex permeability based magneto-elastic model for magnetic material is proposed.•A frequency dependent tanh function derived from Maxwell’s equations is modified.•The model is quite amenable for numerical analyses and it reduces computational time.•Dynamic losses of magnetic materials as a function of stress can be predicted. The presence of mechanical stress significantly affects the performance of electrical machines, particularly in terms of permeability and losses of core materials. A precise modeling of electrical machines requires to consider such magneto-elastic couplings. An efficient approach needs a constitutive model to predict magneto-elastic hysteretic characteristics and its implementation into numerical analyses. However, it is a challenging task for engineers and researchers due to computational time and convergence issues. This paper deals with an approach to model the magneto-elastic behavior of electrical steels using complex permeability. The complex permeability function is used in this work for the first time in the literature to model the effects of stress. The proposed model is quite amenable to numerical analyses and it reduces computational time significantly. Although, the proposed approach is applicable to steady state or cyclic fields and it considers only reversible effects of stress. It provides an alternative way to consider magneto-elastic coupled behavior in numerical analysis of electrical devices involving magnetic materials. Computed results are in good agreement with measured ones with a maximum error of 2.5% for different stress levels and frequencies.