[MITgcm-support] problem with free-slip boundary conditions

[Martin Losch]

Ivane,
has anyone answered you yet?
What does your bathymetry file look like? If you don't put in walls  
along either eastern or western AND northern or southern boundaries  
your domain will be periodic (by default). In matlab such a  
bathymetry would be generated by
h0 = -5000; % meters
h = h0*ones(nx,ny);
h(1,:) = 0;
h(:,1) = 0;
fid=fopen('bathy.box','w','b');
fwrite(fid,h,'real*4/8');
fclose(fid);

Martin

On 26 Mar 2007, at 18:37, pairaud wrote:

> Dear all,
>
> We are currently trying to simulate a very simple two-dimensional  
> flow,
> with free-slip boundary conditions (this is a single vortex in a pi^2
> domain, with constant background temperature). The problem is that we
> are not able to impose free-slip conditions on all four boundaries.
> Indeed, on the east and top boundaries, a non zero velocity component
> appears normal to the boundary, which leads to a blow-up of the code
> after a few iterations (everything is ok on the two other boundaries).
> Here is what we use:
>
>       in data :
>           _ exactConserv=.FALSE.
>           _ implicitFreeSurface=.FALSE.,
>           _ rigidLid=.TRUE.,
>           _ nonHydrostatic=.TRUE.,
>           _ no_slip_sides=.FALSE.,
>           _ no_slip_bottom=.FALSE.,
>
>      in data.pkg :
>           _ useOBCS   = .FALSE.,
>
>      in OBCS_OPTIONS.h :
>           _ #undef ALLOW_ORLANSKI
>           (everything is #undef actually in this file).
>
> Apparently something is missing because the flow behavior is incorrect
> as just explained. We would very much appreciate getting your help  
> about
> this problem.
>
> Regards,
>
> Ivane
>
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support