[Getdp] Microwave Example, 3D, sources

Daryl Van Vorst dvpub at telus.net
Mon Feb 5 21:18:32 CET 2007


Hi again,

I made some progress... I was having convergence problems for a
frequency domain solution, but increasing NB_fill fixed that and it now
gives me a very reasonable looking solution for the case of a volume
current source (J_s[]).

But using J_s[] in the time domain gives me a null right-hand side. I
suspect this is because the formulation takes the time derivative of
J_s[], but is unable to while time stepping. How can this be remedied?
Can I simply remove the Dt and replace J_s[] with its time derivative?
(The implications of this are not entirely clear to me.)

The J_s[] term in the formulation:

      Galerkin { Dt [ J_s[] , {e} ]; 
                 In Vol_S; Integration I1; Jacobian JVol;  }

Thanks,

Daryl.

On Sat, 2007-03-02 at 14:08 -0800, Daryl Van Vorst wrote:
> Hi,
> 
> I'm trying to get the MWE example from the wiki to work for a simple 3D
> problem (and then will hopefully apply it to a more complex problem once
> things things are clear to me).
> 
> Some simple questions to help me figure out what I'm misunderstanding:
> (Any helpful comments would be greatly appreciated)
> 
> 1. Can I impose a time varying volume current (J_s[]) in a small length
> of the centre conductor? This would be the only source term. I have
> tried this, but am unable to get any meaningful results. I'll put
> details in another e-mail if this is something that should work.
> 
> 2. There is a slight difference between MW_e_2Dt and MW_e_3Dt.    
> 
> The 2D one has:
>  Operation { 
>       InitSolution[A] ;
>       InitSolution[A] ;
>       TimeLoopNewmark[t_min,t_max,dt,0.25,0.5] {
>         Generate[A] ; Solve[A] ; 
>         Test[ SaveFct[] ] { SaveSolution[A] ; }
>       } 
>     } 
> 
> The 3D one has:
>     Operation { 
>       InitSolution[A] ; SaveSolution[A] ;
>       TimeLoopNewmark[t_min,t_max,dt,0.25,0.5] {
>         Generate[A] ; Solve[A] ; 
>         Test[SaveFct[]] { SaveSolution[A] ; }
>       } 
>     }
> 
> For the 3D case getdp complains: "Not enough initial solutions for
> TimeLoopNewmark". So I presume that either I'm missing a crucial step in
> setting up the problem, or this should be changed to be similar to the
> 2D case. Which is right?
> 
> 3. Can I impose an electric field on a surface (in the xy-plane) with
> something like:?
> 
> Constraint {{Name e; case { { Region source ; Value Vector [1.,0,0] ;
> TimeFunction TimeFct[];} } } }
> 
> Doing this does not appear to result in the desired field. It does
> appear to impose something on the surface, but not an x-directed
> electric field of magnitude 1. (I've also tried a radially directed
> field which is more appropriate for a coax.)
> 
> 4. What kind of resolution should I be using for a transient problem
> with a time varying J_s[]? And would this be a different resolution for
> a time varying field constraint (like above)?
> 
> Thanks,
> 
> Daryl.
> 
> 
> 
> _______________________________________________
> getdp mailing list
> getdp at geuz.org
> http://www.geuz.org/mailman/listinfo/getdp