Computational electromagnetics for electrical power applications and interactions with information technology devices

Research project P5/34 (Research action P5)

Persons :

Description :

The community of electromagnetic field modelling was until recently divided into two separate worlds: energy conversion (low frequencies, power systems) and wave propagation (high frequencies, information technology). However, with miniaturisation and increasing frequencies, these two worlds are approaching each other very fast. Modern electrical power applications are nowadays very often situated at the boundary of these two worlds.

The objective of this IAP project is therefore twofold. On the one hand, an enlargement of the experience of the network in the domain of the numerical modelling of electrical power systems, including material modelling. On the other hand, the development of a series of new numerical tools that will make it possible to analyse the various interactions of the power systems, either with information technology devices (the important problem of electromagnetic compatibility) or with their environment in a more general sense (electromagnetic shielding, ...), with a clear ecological imperative.

Schematically, the project will be centred on the four following themes, supported by substantial work on the appearing numerical aspects:

Even if this was the theme of phase 4, we plan to devote a part of our efforts in phase 5 to further research on the fundamental interactions between fields of different physical natures, i.e. constitutive laws and coupled problems, what we refer to as local interactions. Indeed, a fundamental drawback of the present simulation tools is still the lack of accuracy of constitutive laws. A better knowledge of materials and of their losses should be acquired as well as the tools to implement effectively and accurately realistic material models (hysteresis, ...) in numerical simulations.

In order to deal efficiently with the interactions at a distance between electromagnetic devices, the well-mastered FEM-BEM coupling will be further developed with other kinds of Green functions (Hankel, Helmholtz) and the solvers will be improved and made faster (multigrid preconditioner, Fast Multipole Method). Further developments are also needed to be able to take advantage of the symmetry properties of the structure in the boundary integrals. These developments will make it possible to investigate different strategies for electromagnetic shielding, radiative and capacitive effects in electronic circuits and important aspects of electromagnetic compatibility.

Practically, designers think in terms of models with only a few degrees of freedom that are well correlated to modifiable parameters like dimensions, measurable constraints or material properties. An important challenge of the project is to establish robust methods to project the results of detailed numerical simulations, with their enormous number of degrees of freedom, onto spaces spanned by a very small number of appropriately selected dynamical variables. We refer to this as nonlinear macro-modelling. Macro-modelling is intended to allow exploring quickly design spaces and to provide a representation of the dynamical behaviour of the device in an embedded system, which could be represented itself by other kinds of modelling tools (circuit solver, symbolic calculator, ...). Our objective is to determine how macro-modelling can be automated and which general methodology can be generated in order to use it as a common design practice.

Finally, all the developments will be done in tight collaboration thanks to a common development platform that will be set up and maintained by all the partners, under the direction of the group ELAP. The objective is to build one single program gathering all the capabilities of the partners, both these that were developed in the past and these that will be developed during the fifth phase. Eventually, this will give rise to an original numerical tool of high scientific interest and which will be able to address the real needs of modern CAD in electromagnetism.