Modelling, design and optimization tools

The designers and operators of electrotechnic devices are more and more lead to use in an intensive way softwares of electromagnetic fields. Indeed, these tools allow to manage virtual prototypes in order to estimate the performances of a device without the effective realization of a real prototype, as example, the figure 1 presents a study of a synchronous machine. In an industrial context, these numerical models must be accurate and fast. Then, the aim of the OMCO axis is to develop methods and tools in order to answer to those concerns.

modelisation-conception-optimisationSeveral themes of study are investigated :

  • The development of analytic and numeric models taking into account the multi-physical aspects of electrotechnic devices. We can distinguish two types of coupled problems. In fact, the magnetic behavior laws of the materials can be impacted by thermal and/or mechanical constraints during the manufacturing or operating steps. Usually, the electric actuators are modelled outside of their environments. Then, to improve the accurate of the numerical models, these should be coupled with the electric, thermal and/or mechanical equations to take into account the system of electric supply, the electrical or mechanical charges.
  • The development of optimization approaches in order to have fast and reliable tools of design.
  • The quality of the solution and the computation time require to solve numerical models. Indeed, for a designer of electric devices, a numerical model is viable, if the compromise quality of solution/computation time is strong. The precision of the solution is crucial for a good interpretation of the results, as example, figure 2 presents the distribution of the numerical error for an induction machine. In an industrial context, the computation time to solve a numerical model should be the lowest possible.
  • The consideration of uncertainties on the geometrical dimensions of a device and on the behavior laws of materials. The idea is to introduce the uncertainties in the numerical model in order to study theirs influences on the interest quantities such as electric currents or mechanical torques.

Some of our industrial partners involved :

EDF, Valeo, Eomys, Bernard et Bonnefond, Auxel, Jeumont Electric, etc.

Our R&D axes