Enhancing magnetic core properties by computational design of amorphous powder packing fractions

Amorphous soft magnetic materials are ideal for high-frequency applications due to their low coercivity, high permeability, and reduced core loss. However, optimizing packing density while maintaining superior magnetic properties remains a challenge, particularly for high-efficiency magnetic cores in electric vehicles and renewable energy systems. This study addresses this challenge by integrating discrete element method simulations, machine learning, and experimental validations to optimize powder packing density and evaluate its impact on magnetic properties. Tri-modal amorphous powders were mixed at various ratios, achieving over 90% relative density with derived ratios by machine learning, significantly outperforming conventional trail-and-error methods. Enhanced core density improved magnetic properties, including a 49.7% reduction in coercivity and an 8.72% increase in saturation magnetization. These improvements were attributed to reduced porosity and optimized compaction strategies. Core loss analysis further demonstrated lower hysteresis and eddy current losses in high-density cores.