dolfin team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

On Wed, Jun 14, 2006 at 06:17:35PM +0200, Johan Jansson wrote:
> On Tue, Jun 13, 2006 at 10:48:07AM +0200, Johan Hoffman wrote:
> > 
> > Ok. A solid model is an option, but it seems that a membrane model may be
> > sufficient. I'll see what I'll do. Adding a membrane model to Ko may then
> > later be a natural step, which probably does not take too much effort?
> > 
> > This connects also to a previous discussion on obtaining a surface mesh
> > automatically from a volume mesh. With a membrane model the equations
> > would be solved on the surface mesh being the boundary of the fluid volume
> > mesh. The resulting coupling as a fluid-stucture interaction problem could
> > be quite a challange depending on the properties of the solid vs fluid.
> > 
> > /Johan
> > 
> 
> Yes, I think a membrane model will be necessary at some point. I
> haven't looked into it so much yet, I think it's a good principle to
> keep things simple unless absolutely necessary. I guess the main
> property of the membrane model which we haven't considered in Fenics
> is that we have triangular cells in 3D.

Won't be a problem with the new mesh.

/Anders

> I assume the application here is blood vessels or something similar? I
> think the walls in that case are quite thick w.r.t. the diameter,
> something like 5-10%, so I don't think a membrane model is
> applicable. It's probably unavoidable in other applications though,
> where you have a very small thickness and performance is paramount.
> 
>   Johan