Questions - Valparaíso - Chile

Christophe Geuzaine Christophe.Geuzaine at
Wed Sep 6 11:41:32 CEST 2000

Alejandro Angulo wrote:
> Christophe:
> I write to make you a little question : How I coupling
> electrical circuit with electrostatic floatig
> potencial 2D problem? . I try to make this, but I
> don't know how introduce de floating conductor in the
> circuit constraint, because this can't be consider
> like "branch". I can define it as "node"?.

Hi Alejandro,

To write a Network constraint including FEM domains, you just have to
include the finite element region into the branch. Floating potentials
are exactly like the global U quantity appearing in the examples I sent

> Now, I'am trying to solve a problem in time domain,
> but the time parameter seems not change when it
> appears in constaint term (my voltage source always
> apperars like DC source !!!). The constraint that I
> used is:
> Vf = 110.*Sqrt[2.] ;
> Freq = 50. ;
> Fase = 2.*(Pi/3.) ;
> { Name Voltage ; Type Assign ;
> Case {
> { Region Ebt ; Value Vf * F_Cos_wt_p [] {2.*Pi*Freq ,
> Fase} ; }
> }
> }

You should split the value in two parts:

Case {
 { Region Ebt ; Value Vf ; TimeFunction F_Cos_wt_p [] {2.*Pi*Freq ,
Fase} ; }

> Also, I got little problems when I use the stranded
> conductor formulation that you send me in a past
> e-mail with Electric Circuit . Is possible to couple
> both problems??????

Yes, see above.

> When I couple 2D a-v formulation with circuit
> equation, the 'large' of the domain is include only in
> the GlobalTerm???  I write some like this:
> Axysimmetric Case.
> GlobalTerm { [ -(2*Pi) * Dof{I} , {U} ] ; In BT ; }
> Planar Case.
> GlobalTerm { [ -(Large) * Dof{I} , {U} ] ; In BT ; }
> It's correct or no???

I don't know your problem sufficiently to answer. Just develop your
equations on paper in a rigorous way. If you come up with the terms I
sent you in my e-mail in the axisymmetrical case, the results you
obatain by reasonning in the same way for planar problems should also be

> I'm still try to solve my problem of power tranformer
> in appropiate form. Now I have found the following
> problem:
> In my transformer, the LV coil is shortcircuit and it
> haven't any problem when it's coupling whit circuit
> constraint. On the other hand, the HV coil is fed with
> nominal current.  The real coil has two conductor in
> parallel (473 turns X 2 conductors), but initially I
> model it like a single coil with twice number of turns
> (946 turns X 1 conductor).  With this formulation
> GetDP runs OK, but when I try to formulate the
> transformer with two coils in parallel GetDP show me a
> Warning "Iterative solver terminated with code =
> -9"????.
> I send you my new formulation and I request you please
> you revise it and indicate me my errors.

Did you try to print the matrix/RHS in a file and see if the system you
build is still valid? Did you check the linear solver options (maybe try
to solve the problem on a coarse grid with a direct solver ---see LU in
the SOLVER.PAR file)? I'll be out of my office for a week, so I won't
have time to check your problem immediately.

> Thanks and regards.
> Alejandro
> PD: - The mesh generate is a 'Coarse Mesh'!!!!!! Does
> this bring problems????
>     - Wich is the maximum size of the problem that
> GetDP can solve?. I ask you this question , because in
> this same problem, when I try to solve it with more
> refined mesh, appeared again the message: 'Couldn't Reallocate....'.

The size of the problem is only limited by the memory of your computer.
The choice of the linear solver dramatically changes the memory
consumption: you should therefore investigate which method (and which
preconditionner) is best to solve your problem.

My best,

Christophe Geuzaine

Tel: +32-(0)4-366.37.10    mailto:Christophe.Geuzaine at
Fax: +32-(0)4-366.29.10