14.01.2013 Views

Abstracts - Dipartimento di Elettronica Applicata

Abstracts - Dipartimento di Elettronica Applicata

Abstracts - Dipartimento di Elettronica Applicata

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Meta 2010 & FEM 2010 – Rome, 13-15 December 2010<br />

A Comparison between Hybrid Methods:<br />

FEM-BEM versus FEM-DBCI<br />

G. Aiello, S. Alfonzetti, S. A. Rizzo, and N. Salerno<br />

Università <strong>di</strong> Catania, <strong>Dipartimento</strong> <strong>di</strong> Ingegneria Elettrica, <strong>Elettronica</strong> e dei Sistemi<br />

(DIEES), Catania, Italy – E-mail: alfo@<strong>di</strong>ees.unict.it<br />

In the literature several methods have been devised to enable the Finite Element<br />

Method (FEM) to solve static and quasi-static electromagnetic field problems in openboundary<br />

domains. Among these there are the hybrid FEM/BEM (Boundary Element<br />

Method) method [1], and the hybrid FEM-DBCI (Dirichlet Boundary Con<strong>di</strong>tion<br />

Iteration) method proposed by the authors [2]. This paper compares these two hybrid<br />

methods by referring to simple electrostatic field problems.<br />

Consider an electrostatic system made of voltaged conductors, <strong>di</strong>electric objects and<br />

charge <strong>di</strong>stributions embedded in air. In the FEM-BEM method a truncation boundary<br />

�F enclosing the system is introduced. On �F an unknown Neumann con<strong>di</strong>tion<br />

�r�v/�n=q is assumed. The FEM-BEM leads to the global system [1]:<br />

� A A F 0 � � v � �b<br />

0 �<br />

� t<br />

� � �<br />

�<br />

� �<br />

�<br />

A F A FF C<br />

� �<br />

v F � �<br />

0<br />

�<br />

(1)<br />

��<br />

0 H � G��<br />

��<br />

q ��<br />

��<br />

0 �<br />

F �<br />

where: v and vF are the vectors of the unknown values of the potential v in the nodes<br />

inside the domain and on �F, respectively, A, AF and AFF are sparse matrices of<br />

coefficients, b0 is the known term vector due to the conductor potentials and sources,<br />

C is a sparse matrix of coefficients, and qF is the vector of the unknown values of q,<br />

evaluated in nodes other than those of v [1].<br />

In the FEM-DBCI method a Dirichlet boundary con<strong>di</strong>tion is assumed on �F. The<br />

global system is [2]:<br />

� A A F � � v � �b0<br />

�<br />

� � � � � � � (2)<br />

��<br />

G'<br />

H'<br />

� �v<br />

F � � 0 �<br />

Comparing the two methods the following considerations can be made. First, by<br />

analyzing the <strong>di</strong>mensions of the various matrices, it can be shown that FEM-DBCI<br />

requires less memory than FEM-BEM. Moreover, the greater complexity of (1) with<br />

respect to (2) makes FEM-BEM more time-consuming than FEM-DBCI.<br />

From the point of view of accuracy, it can be noted that in FEM-DBCI a numerical<br />

derivative of the potential is performed on the integration curve, whereas this is not<br />

necessary in FEM-BEM. FEM-BEM can therefore be expected to give more accurate<br />

results than FEM-DBCI.<br />

These considerations have been verified by means of a set of examples, exhibiting analytical<br />

solutions.<br />

References<br />

[1] S. Alfonzetti, N. Salerno, “A non-standard family of boundary elements for the hybrid FEM-BEM<br />

method,” IEEE Trans. Magn., 45, 1312-1315, 2009.<br />

[2] G. Aiello, S. Alfonzetti, “Charge iteration: a procedure for the finite-element computation of<br />

unbounded electrical fields,” Int. J. Num. Meth. Engng, 37, 4147-4166, 1994.<br />

44

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!