[dune-pdelab] RE : Instationary Navier-Stokes equations with Finite Volume method

[CERRITO Samuel]

Hello,

First of all, thanks for your prompt answer !

Once I have written my own operator and run my simulation, do you have any tools (functions ?) for debugging ? By debugging, I mean dumping the values of a matrix in case the maths are wrong

Thanks again
Samuel


-------- Message d'origine--------
De: Christian Engwer [mailto:christian.engwer at uni-muenster.de]
Date: jeu. 11/09/2014 14:36
À: CERRITO Samuel
Cc: dune-pdelab at dune-project.org; Xavier.KUHN at cea.fr
Objet : Re: [dune-pdelab] Instationary Navier-Stokes equations with Finite Volume method
 
Hi,

> I use classes from stokesdg.hh  as local operators and P0LocalFiniteElementMap as finite element map for pressure and velocity components. My idea was to turn discontinuous Galerkin case into finite volume method  by using a zero order polynomial (P0LocalFiniteElementMap).
> 
> This method doesn't seem to work, do you think the classes I use are appropriate to solve these equations ? I am aware of the existence of StokesDGVectorFEM but I'm not quite sure about how it works.

the problem is not DUNE related, but due to the maths. The p0/p0 pair
does not fulfill the inf-sup condition and thus you get wrong
results. Furthermore the DG method is not a FV method, although they
have a lot of similarities. In particular, you have to be careful
regarding the definition of your fluxes.

> Are there any other local operators I can try ?

you should write your own operator. And! more important first think
about which discretization you want to use.

You'll have to either use a pair like P1/P0, or use a stabilization,
like LPS, or other techniques.

Ciao
Christian


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.dune-project.org/pipermail/dune-pdelab/attachments/20140915/9257d6f0/attachment.htm>