[Getdp] Just a litle help ..

Christophe Geuzaine geuzaine at acm.caltech.edu
Sat Mar 13 09:33:01 CET 2004


Nacho Andres de la Fuente wrote:

> Hi everybody, I am starting to seriously use gmsh + getdp and I am
> pretty amazed about its capabilities. In principle everything looks
> quite simple but nevertheless doubts always arise. I hope someone can
> help me on this matter:
> 
> If you do not mind, I would like to frame the questions into this
> example, since I have calculated the analytical solution:
> 
> "Given two concentric conducting spheres (a spacecraft) in outer space,
> solve Div[sigma.Grad(Phi)]= 0 with given Neumann b.c. in the outer
> surface of the external sphere." 
> (Conductivities of sphere 1 and sphere 2 are different => b.c. at the
> discontinuities as well)
> 
> 1) Are the b.c. at the discontinuities absorbed by the weak formulation?

Yes.

> Or do I have to separate all the domains and impose all the Neumann
> b.c.? 
> 
> 2) My problem is an electro-quasistatic one with an equation which
> includes the Neumann conditions like:
> 
> Formulation {
>   { Name PhEl_v; Type FemEquation;
>     Quantity {
>       { Name v; Type Local; NameOfSpace PhEl_v_Ele; }
>     }
>     Equation {
>       Galerkin { [ sigma[] * Dof{d v} , {d v} ]; In ConductingDomain;
>                  Jacobian JVol; Integration GradGrad; }
>       Galerkin {
> [IllPhotoFunc[{v},Colat[X[],YNew[X[],Y[]],ZNew[X[],Y[]]]] , {v} ];
>                  In SurfIll; Jacobian JSur; Integration I1; }
>       Galerkin { [ DarkPhotoFunc[{v}] , {v} ]; In SurfDark;
>                  Jacobian JSur; Integration I1; }
>     }
>   }
> }
> 
> Do I miss something here? because I am just getting zeros as a solution
> (and the analytical solution yields a different solution). In case that
> you need it, I have attached both the *.geo and the *.pro files.

It is probably because you forgot to put the surfaces SurfIll and
SurfDark (on which you impose the Neumann boundary conditions) in the
support of your basis functions.

> 
> 3) Concerning the axilsymmetry conditions in the problem? I am using
> cartesian coordinates since the geometry of my problem is the real one
> (i.e. the gmsh file is made of two concentric spheres). In principle the
> conditions should be (in spherical formulation):

Just use the "VolAxi" Jacobian on a 2D mesh (having the Y-axis as the
axis of symmetry).

Christophe

-- 
Christophe Geuzaine
Applied and Computational Mathematics, Caltech
geuzaine at acm.caltech.edu - http://geuz.org