This presentation explores the implementation and validation of homogenization techniques for modeling litz and stranded windings in 3D FEM simulations using the open-source solver ElmerFEM. The focus is on a practical question faced by magnetics engineers: how close can simulations get to real high-frequency measurements, and what are the pitfalls along the way? We compare simulation results with comprehensive frequency-sweep measurements on custom prototypes using litz and multi-turn windings, applying corrections to measurement data to account for parasitic capacitances. Across these studies, we underline both the strengths and the limitations of the homogenization approach: while it offers an efficient way to capture losses without meshing every strand, factors such as non-uniform strand distribution, twisting, and circulating currents remain outside its scope. The overall theme is pragmatism -FEM simulations are powerful, but reconciling them with real-world data requires awareness of their limits, careful correction of measurements, and a willingness to accept that models are approximations, not magic. Attendees will walk away with concrete insights into when homogenization can be trusted, when it cannot, and how to combine measurements and simulations for reliable design decisions in high-frequency magnetics.
