21745 articles – 15574 Notices  [english version]
HAL : inria-00589763, version 1

Voir la fiche concise  BibTeX,EndNote,...
A new method for inverse electromagnetic casting problems based on the topological derivative
Canelas A., Andre Antonio N., Roche J. R.
Journal of Computational Physics 230, 9 (2011) 3570-3588 - http://hal.inria.fr/inria-00589763
Articles dans des revues avec comité de lecture
Mathématiques/Optimisation et contrôle
A new method for inverse electromagnetic casting problems based on the topological derivative
Alfredo Canelas () 1, Novotny Andre Antonio 2, Jean Rodolphe Roche 3, 4
1 :  Instituto de Transporte y Estructuras (UDELAR)
UDELAR
Uruguay
2 :  Laboratorio Nacional de Computação Cientifica / National Laboratory for Scientific Computation (LNCC / MCT)
http://www.lncc.br
Laboratorio Nacional de Computação Cientifica
LNCC, Av. Getulio Vargas, 333, Quitandinha, 25651-075, Petropolis, RJ
Brésil
3 :  Institut Elie Cartan Nancy (IECN)
http://www.iecn.u-nancy.fr/
CNRS : UMR7502 – INRIA – Université Henri Poincaré - Nancy I – Université Nancy II – Institut National Polytechnique de Lorraine (INPL)
France
4 :  CALVI (INRIA Nancy - Grand Est / IECN / LSIIT / IRMA)
CNRS : UMR7005 – INRIA – Université de Strasbourg – Université de Lorraine
France
The inverse electromagnetic casting problem consists in looking for a suitable set of electric wires such that the electromagnetic field induced by an alternating current passing through them makes a given mass of liquid metal acquire a predefined shape. In this paper we propose a new method for the topology design of such inductors. The inverse electromagnetic casting problem is formulated as an optimization problem, and topological derivatives are considered in order to locate new wires in the right position. Several numerical examples are presented showing that the proposed technique is effective to design suitable inductors.
Anglais

Journal of Computational Physics (J. Comput. Phys.)
Publisher Elsevier
ISSN 0021-9991 (eISSN : 1090-2716)
01/05/2011
internationale
Elsevier
230
9
3570-3588