dolfin team mailing list archive

[Andy Ray Terrel <aterrel@xxxxxxxxxxxx>]

Anders Logg wrote:
> On Thu, Feb 14, 2008 at 08:39:07AM +0100, Dag Lindbo wrote:
>   
> > Anders Logg wrote:
> >     
> > > On Wed, Feb 13, 2008 at 08:52:29PM +0100, Dag Lindbo wrote:
> > >       
> > > > > > On Tue, Feb 12, 2008 at 10:23:09AM +0100, Dag Lindbo wrote:
> > > > > >             
> > > > > > > Hello all!
> > > > > > >
> > > > > > > In the stokes demo (taylor-hood), I notice spurious pressure spikes
> > > > > > > near
> > > > > > > the corners. What is causing these?
> > > > > > >
> > > > > > > I changed to a paraboloc inflow profile (consistent with the no-slip
> > > > > > > BCs),
> > > > > > > but the problem remains. Further, I ran the same problem on a
> > > > > > > unitsquare
> > > > > > > mesh (see plot attached). The pressure spikes are large. Is the demo
> > > > > > > problem ill posed?
> > > > > > >
> > > > > > > Thanks!
> > > > > > > Dag Lindbo, KTH
> > > > > > >               
> > > > > > I would also suspect the boundary conditions for the velocity
> > > > > > to be the problem, but if you still see the spikes with a parabolic
> > > > > > inflow profile (which is zero at the no-slip boundaries) then I have
> > > > > > no idea.
> > > > > >
> > > > > >             
> > > > > Is this in all corners? Could you send a plot?
> > > > >
> > > > > /Johan
> > > > >           
> > > > Problem solved. Thanks for the help! The error is that the facets adjacent
> > > > to the corners no get properly marked for noslip BC.
> > > >
> > > > Fix to get the correct subdomain markers, in demo/mesh/subdomain/main.cpp:
> > > >
> > > > class Noslip : public SubDomain
> > > >   {
> > > >     bool inside(const real* x, bool on_boundary) const
> > > >     {
> > > >       return (x[1] > 1.0 - DOLFIN_EPS || x[1] < DOLFIN_EPS) && on_boundary;
> > > >     }
> > > >   };
> > > >
> > > > /Dag & Shilpa
> > > >         
> > > Then you don't get no-slip on the dolphin which makes the solution
> > > look pretty dull, don't you think?
> > >
> > >       
> > Right. For the dolfin you can put
> > return ((x[1] > 1.0 - DOLFIN_EPS || x[1] < DOLFIN_EPS) ||
> > 	(x[0] > DOLFIN_EPS && x[0] < 1.0 - DOLFIN_EPS) ) && on_boundary;
> >
> >    Dag
> >     
>
> Turns out to be even simpler:
>
>   return on_boundary
>
> does the trick. Inflow and outflow bcs will be applied after and then
> overwrite where we should not have no-slip.
>
>   
Don't you think that would be a little opaque to a new user?

-- ART

--

====================
Andy Terrel
Computer Science Dept
University of Chicago
aterrel@xxxxxxxxxxxx
---------------------

Science is what we understand well enough to explain to a computer. 
Art is everything else we do.
                                                     -Donald Knuth