[Getdp] 3D magnetostatic convergence

Ruth V. Sabariego r.sabariego at ulg.ac.be
Wed Mar 18 14:03:06 CET 2009


You should increase Nb_Fill much more, try with 500, 1000...
I am using a Getdp version compiled with Petsc and a direct solver 
(umfpack), with the number of unknowns you have that could be also feasible.

Ruth


svhb at telenet.be wrote:
> hi all,
>
> I have a working description of a magnetostatic 3D problem with B-conforming formulation and around 100.000 tetrahedra in the mesh. My goal is to put 3 systems next to each other to see the neighour influence from one to another electromagnet. This blows the mesh to > 200.000 tetrahedra, and convergence is missing when solving. 
>
> I changed the Nb_Fill value in the solver.par file to 80, but still no luck. What is the maximum that can be used for this parameter? 
>
> Are there other parameters that influence the convergence? 
> Is a h-phi formulation better to reach convergence? 
>
> Stefaan
>
>
>
>
>   


-- 
Dr. Ir. Ruth V. Sabariego
University of Liege, Institut Montefiore,
Dept. of Electrical Engineering & Computer Science, 
Applied & Computational Electromagnetics (ACE),
Sart Tilman Campus, Grande Traverse, 10 (B28), B-4000 LIEGE, Belgium
phone: +32-4-3663737 -- fax: +32-4-3662910 -- http://elap.montefiore.ulg.ac.be/