Development of Ferromagnetic Metallic Glasses into Low Loss Power Transformer Cores

Author: Floyd, Michael Cameron Dawley

Year: 2018

Degree: Dissertation (Ph.D.)

Advisor: Johnson, William Lewis

Committee Members: Johnson, William Lewis; Schwab, Keith C.; Demetriou, Marios D.; Greer, Julia R.; Faber, Katherine T.

Option: Materials Science

DOI: 10.7907/Z9QF8R27

Abstract

Currently, 3% of energy losses in the U.S. electrical grid occur at power transformers. With a transition to Metglas, transformer efficiency could be increased, but is Metglas the best replacement material for power transformers?

With this in mind we develop a Fe-based metallic glass for its glass forming ability and soft magnetic properties. During this development we identify a redox reaction of boron oxide by Si during melt fluxing of the Fe-based glass, which promotes an unexpected exchange of Si and B in the alloy. Taking this reaction into account, a unique optimization strategy is implemented, enabling oxide purification of the melt coupled with a significant but predictable shift in composition. This leads to an optimized Fe-based glass demonstrating a global peak in glass forming ability. Following boron oxide fluxing in the high temperature melt, alloy with composition Fe57.5Co20.2Si10.2B2.05P10.05 transforms to Fe57Co19.2Si6.8B7.4P9.6, and increases its critical rod diameter from 1 mm to 5 mm. The alloy also demonstrates excellent soft ferromagnetic performance characterized by a magnetic saturation of 1.53 T.

While developing the above alloy, we also analyzed the effect of varying thickness of a Fe68Mo4Ni3Co5Si1P11.5C5B2.5 transformer core as a function of frequency to discover if there was a minimum in the losses. We did not find a single minimum, but found that the optimal thickness exhibits a logarithmic dependency on frequency. This dependence suggests the optimal thickness of a core ranges from 100−400μm, instead of in the < 50μm range currently used. These larger optimal thicknesses are unexpected if anomalous losses are not considered, but the dominance of the anomalous losses at low frequencies, or for thin samples, validates the need for thicker power transformers. While other amorphous metals and casting techniques will yield varying results, the logarithmic dependence on frequency and the 100−400μm optimal thickness range should be broadly applicable.

Files