Introduction of Numerical Homogenization, and its Application to Electromagnetic Field Problems

  • TAGAMI, Daisuke (Kyushu University)

Please login to view abstract download link

When numerically evaluating the AC resistance of Litz wires used in electronic devices to transmit high-frequency AC currents, directly modeling the complex microstructure of the wire, such as bundling multiple stranded wires, leads to extremely large-scale computational models. Then, its computational costs become extremely high, and unrealistic. Therefore, the homogenization method is known as a modeling technique that significantly reduces the computational cost by treating the wire as a single equivalent homogeneous material. Mathematical frameworks for the homogenization method have been proposed by Allaire (1992) based on the 2-scale convergence, and by Allaire--Briane (1996) based on the multi-scale convergence, and numerical schemes based on these convergences have been proposed by Terada--Hori--Kyoya--Kikuchi (2000) and Allaire--Brizzi (2005). Furthermore, another numerical homogenization method proposed by Altmann--Henning--Peterseim (2021) and a method using continued fractions have been proposed by Igarashi (2025). In this paper, we consider the numerical homogenization method as a numerical scheme based on the homogenization one based on two-scale convergence, attempt to extend it to electromagnetic field problems. Then, we establush some mathematical results such as error estimates from the mathematical points of view. Furthermore, we show some numerical examples to validate the error estimates of the numerical homogenization method using the manufactured solution problem, and to evaluate the efficiency of the numerical homogenization method by useing application to the corresponding engineering problem.