Periodic Reporting for period 4 - MULTIMAG (Multiscale Magnetic Models for Emerging Energy Conversion Applications)
Reporting period: 2024-09-01 to 2025-08-31
Power electronics is used for converting electrical energy from one form into another. Although the efficiency of modern power electronic systems is high, a large amount of energy is annually wasted as power losses in such systems. This project was related to modeling of power losses in magnetic components used in power electronics. In particular, material models were developed both for ferrites and tape-wound magnetic cores, and numerical computation techniques were developed for copper windings comprising many thin parallel strands. As applications, inductors, transformers and wireless power transfer systems were considered.
Losses arising in the magnetic cores and windings comprise a major part of the total losses in a power electronic converter. Accurate analysis of the losses is challenging, since the physical phenomena related to the loss mechanisms in the magnetic materials are not properly understood and since conventional computational models are too heavy to account for the fine geometrical details in the materials and windings. The modeling tools developed in the project provide new insight into the loss mechanisms with lower computational burden, which is important since the use of power electronics in the electricity system is rapidly increasing. Improved loss models allow designing more material- and cost-efficient energy-conversion systems.
Losses arising in the magnetic cores and windings comprise a major part of the total losses in a power electronic converter. Accurate analysis of the losses is challenging, since the physical phenomena related to the loss mechanisms in the magnetic materials are not properly understood and since conventional computational models are too heavy to account for the fine geometrical details in the materials and windings. The modeling tools developed in the project provide new insight into the loss mechanisms with lower computational burden, which is important since the use of power electronics in the electricity system is rapidly increasing. Improved loss models allow designing more material- and cost-efficient energy-conversion systems.
The power losses in magnetic cores depend on the intrinsic material parameters, i.e. the electrical conductivity, permeability and permittivity. To model the losses in cores with arbitrary dimensions, these intrinsic parameters need to be known. However, since the distributions of the electric and magnetic fields in ferrite cores depend on the core dimensions, the intrinsic material parameters cannot be directly measured. We developed inverse-problem-based characterization methodologies in which we iteratively search for material parameter values input to a simulation model such that the simulated core impedances match with measured ones. In such a way, an estimate of the intrinsic material parameters is obtained.
In magnetic cores wound from hundreds of layers of thin (e.g. 20 um thick) amorphous tapes, the power losses at high frequencies arise both because of eddy currents in the tape layers and the magnetization dynamics at the level of magnetic domains. To model such effects, a 2-D simulation model was coupled to a 1-D model of the eddy currents in a single tape layer which also accounts for the magnetization dynamics at the level of magnetic domains. The approach allows considering the tape-wound inductor or transformer cores in a homogenized manner without the need of modeling each tape layer separately.
Multiscale modeling methods for granular magnetic materials were developed to account for grain- and particle-level effects in macroscale simulations. Computationally efficient time-domain models were introduced for stochastic simulation of materials with a random microstructure. Such models can provide improved understanding of the physical loss mechanisms and their dependnce on the grain and particle structures.
For windings, we developed efficient numerical techniques to account for several thin parallel conductors in the simulation of complete wound components and derived new dimensionality reduction techniques to avoid heavy 3-D simulations. 3-D effects in an inductor were modeled by coupling several 2-D slices together instead of building a full 3-D model. The approach allows performing calculations for such coil geometries which would be impossible to model with a 3-D model. We also developed methods for modeling wireless power transfer coils with multistrand windings by starting from a sub-model of an individual conductor and recursively constructing higher-level models by utilizing pre-computed results from the sub-models. This makes it possible to perform accurate but very fast calculations of multistrand windings.
In magnetic cores wound from hundreds of layers of thin (e.g. 20 um thick) amorphous tapes, the power losses at high frequencies arise both because of eddy currents in the tape layers and the magnetization dynamics at the level of magnetic domains. To model such effects, a 2-D simulation model was coupled to a 1-D model of the eddy currents in a single tape layer which also accounts for the magnetization dynamics at the level of magnetic domains. The approach allows considering the tape-wound inductor or transformer cores in a homogenized manner without the need of modeling each tape layer separately.
Multiscale modeling methods for granular magnetic materials were developed to account for grain- and particle-level effects in macroscale simulations. Computationally efficient time-domain models were introduced for stochastic simulation of materials with a random microstructure. Such models can provide improved understanding of the physical loss mechanisms and their dependnce on the grain and particle structures.
For windings, we developed efficient numerical techniques to account for several thin parallel conductors in the simulation of complete wound components and derived new dimensionality reduction techniques to avoid heavy 3-D simulations. 3-D effects in an inductor were modeled by coupling several 2-D slices together instead of building a full 3-D model. The approach allows performing calculations for such coil geometries which would be impossible to model with a 3-D model. We also developed methods for modeling wireless power transfer coils with multistrand windings by starting from a sub-model of an individual conductor and recursively constructing higher-level models by utilizing pre-computed results from the sub-models. This makes it possible to perform accurate but very fast calculations of multistrand windings.
The developed material models allow separating the losses in magnetic cores into different components based on different material parameters or physical phenomena. This challenges the common engineering approaches in which the loss is understood as a lumped value valid for a specific core geometry. The winding models enable calculations that are extremely heavy or even impossible with existing modeling approaches. They can thus be used to provide relevant information on the distribution of currents and power losses in the conductors of multistrand windings.