dolfin team mailing list archive

Thread
Date

Re: i would like to solve a magnetostatic problem with dolfin

To: Jens Nie <JNie@xxxxxxxxxxxxxxxxxxx>
From: Anders Logg <logg@xxxxxxxxx>
Date: Wed, 30 Mar 2005 09:11:33 -0600
Cc: Discussion of DOLFIN development <dolfin-dev@xxxxxxxxxx>
In-reply-to: <D6A0366D4D3A1241809C75B35D89B879355C9B@lin0017.roseninspection.net>
Mail-followup-to: Jens Nie <JNie@xxxxxxxxxxxxxxxxxxx>, Discussion of DOLFIN development <dolfin-dev@xxxxxxxxxx>
Reply-to: Discussion of DOLFIN development <dolfin-dev@xxxxxxxxxx>
User-agent: Mutt/1.5.6+20040907i

Hi!

There is currently no ready-made solver in DOLFIN for your problem,
but DOLFIN can be used to build a solver for your problem.

DOLFIN does not provide a meshing tool, so you need to generate your
mesh somewhere else (Tetgen, Triangle, PDE toolbox, FEMLAB, ...) and
save it/convert it to DOLFINs XML format.

You then need to enter your equations in weak form (the variational
formulation) using FFC (http://www.fenics.org/ffc/).

Both FFC and DOLFIN should be considered as experimental and as such
there is very little documentation, so if you need something more
stable you should try another package.

Best wishes
/Anders

On Wed, Mar 30, 2005 at 03:39:25PM +0200, Jens Nie wrote:

>    Hello.
> 
> 
>    Being a complete newbie to dolfin I need some advise on how to start
>    modeling a magnetostatic problem. All this for the case that it is
>    possible using dolfin at all.
> 
> 
>    What I would like to model is the magnetic flux leakage field of a
>    defect in a steelplate. My model is based on the assumption that the
>    magnetic field applied is defined by a magnetic scalar potential
>    gradient along a the steelplate. Actually the geometry of interest
>    would look like this:
> 
> 
>    Background region of air, modeled using a cube of 40 x 40 x 40 mm.
> 
>    Located at the backgrounds center a steelplate sized 40 x 40 x 6 mm
> 
>    And in the center of the steelplate the defect with a size of 10 x 10
>    x 6 mm.
> 
> 
>    In order to create the magnetic field, magnetic scalar potentials are
>    to be defined at the front and the rear of the model. For example 0
>    and 800 A. The Boundary conditions should be normal magnetic on these
>    faces, tangential magnetic on all other faces. To have an effect from
>    the steel plate the material parameters for this region have to be
>    defined. At a first try a relative permeability of 75 should be ok.
> 
> 
>    Having studied some of the examples the following questions arise:
> 
> 
>    - Which solver would be the right one to choose?
> 
>    - How to create the geometries of the model, how to create the mesh?
> 
>    - How can the material properties be defined in the model?
> 
>    - How can nonlinear material properties (B-H-Curve for steel) be
>    modeled with this solver?
> 
> 
> 
>    Kind regards
> 
> 
>    Jens Nie