Experimental, Empirical, and Analytical Comparison of Losses in a kA DC-Choke
Authors
Thomsen, Oskar Skov ; Laursen, Mads
Term
4. term
Education
Publication year
2026
Submitted on
2026-05-28
Pages
66
Abstract
This project examines how to determine power losses in a high-power DC choke (an inductor used to smooth current in power electronics). The main focus is on core losses at high frequencies, which are difficult to predict accurately. We evaluate three approaches: the classical Steinmetz equation, the Generalized Steinmetz Equation (GSE) for non-sinusoidal waveforms, and the finite element method (FEM), a computer simulation that captures geometry-dependent magnetic fields. Results are compared with manufacturer core-loss data under the choke’s nominal operating conditions. The deviations relative to the manufacturer data are 4.13% (Steinmetz), 19.57% (GSE), and 15.56% (FEM). These differences highlight the challenge of accurately predicting core losses in high-power magnetic components. To connect calculations with measurements, we built a calorimetric test setup that infers losses from heat and used it to fit the Steinmetz coefficients in the equation. The next step is to test the choke at nominal load in this setup to obtain a definitive power-loss measurement.
Dette projekt undersøger, hvordan man bestemmer effekttab i en højeffekts DC-drossel (en spole, der udglatter strøm i effektelektronik). Hovedfokus er kernetab ved høje frekvenser, som er svære at vurdere præcist. Vi vurderer tre tilgange: den klassiske Steinmetz-ligning, den generaliserede Steinmetz-ligning (GSE) til ikke-sinusformede bølgeformer og finite element-metoden (FEM), en computersimulering der kan modellere geometri-afhængige magnetfelter. Resultaterne sammenlignes med producentens kernetabsdata under de nominelle driftsbetingelser, som drosslen er designet til. Afvigelserne i forhold til producentens data er 4,13 % (Steinmetz), 19,57 % (GSE) og 15,56 % (FEM). Disse forskelle understreger udfordringerne ved at forudsige kernetab i højeffekt-magnetiske komponenter. For at validere resultaterne eksperimentelt byggede vi en kalorimetrisk testopstilling, der bestemmer tab ud fra varme, og brugte den til at tilpasse Steinmetz-koefficienterne i ligningen. Næste skridt er at teste drosslen ved nominelt læs for at opnå en endelig måling af effekttab.
[This apstract has been rewritten with the help of AI based on the project's original abstract]