[Getdp] enquiry about the use of GetDP
Christophe Geuzaine
geuzaine at acm.caltech.edu
Mon Mar 10 08:17:16 CET 2003
raul conejeros wrote:
>
> I have tried to trace some of the how to´s from the reference maula and
> the mailing list, but there are still some things which are unclear to
> me and i hope you could help me.
>
> The first concerns the definition of boundary conditions:
>
> I understood, a Dirichlet boundary condition is defined as i use it in
> the constrant Fronteras of my file, but that case i fail to understand
> why it seems not to work for the region tue indicated in the same
> constraint which i fixed to 1.0.
Raul - your mesh is wrong: the line elements in the group 'tue' are not
connected to the triangles in the surface mesh.
>
> On the other hand i defined a Neumann BC as indicated in some of the
> cases of the mailing list as :
>
> Equation {
> Galerkin { Dt [ rho[]*Dof{Tmp}, {Tmp}];
> In todo ; Integration TmpInt; Jacobian TmpJacSur; }
> Galerkin {[ epsil[] * Dof{Grad Tmp}, {Grad Tmp} ];
> In todo ; Integration TmpInt; Jacobian TmpJacSur; }
> Galerkin {[ - nbc , {Tmp} ];
> In Pro ; Integration TmpInt; Jacobian TmpJacLin; }
> }
>
> where nbs is fixed to a value of 10
>
> In this case it seems that it does not matter how much i change the
> value of nbc, the temperature does not seem to vary more around the
> boundary Pro.
Contrary to a Dirichlet condition where you directly set the value of
some degrees of freedom, here you define a new integration term on the
boundary, where the unknowns have to be interpolated and integrated. You
thus need to add the region on which you do that integration to the
'Support' of your 'FunctionSpace':
FunctionSpace {
{Name TmpSpc; Type Form0;
BasisFunction {
{Name sn; NameOfCoef Tmpn; Function BF_Node;
Support Region[{Supe,Pro}]; Entity NodesOf[All];}
}
Constraint {
{NameOfCoef Tmpn; EntityType NodesOf; NameOfConstraint Fronteras;}
{NameOfCoef Tmpn; EntityType NodesOf; NameOfConstraint Inicial;}
}
}
}
> The same happens if i try to convert the region tue into a
> source or into a sink region.
Same mesh issue as above.
> As a last point i would like to ask about the way to define an
> anisotropic parameter epsil with a value in the x direction and a
> different one in the y direction and use it in the equation to indicate
> a different themal condution in each axis direction.
Just define epsil as a tensor, e.g. epsil[] = TensorDiag[valx,valy,0].
Cheers,
Christophe
--
Christophe Geuzaine
Tel: (626) 395-4552 http://www.geuz.org
Fax: (626) 578-0124 mailto:geuzaine at acm.caltech.edu