[FLASH-BUGS] Re: Problem with outflow/reflecting BCs

[Dan Sheeler]

I think this is a bug in the solver in that it relies on hydro for its 
periodic bcs.  If you set hydro to outflow, and gravity to isolated, you 
should be fine.  If you set hydro and gravity to periodic, you should be 
fine.  If you set hydro to outflow and gravity to periodic, that will be a 
problem, and I think that's what you're doing in these runs.  The gravity 
gets periodic bcs only when hydro is periodic (because periodic hydro 
constructs the grid such that a block at a domain edge has a neighbor on 
the opposite side of the domain, the gravity bc routines never really get 
activated in the periodic hydro case... if that makes sense).

Do you really need periodic gravity with outflow hydro?  Then the gravity 
bc routines would need to be a little smarter than they are now, but I 
think it could be done.

Dan

  --
Dan Sheeler
ASC Flash Center
sheeler at flash.uchicago.edu
(773) 834-3236

On Tue, 10 Apr 2007, Robi Banerjee wrote:

> Hi,
>
> We encountered a problem in connection with outflow/reflecting
> boundary conditions (still using FLASH 2.5). We stumbled over
> this problem with setups which use the self-gravity module. Although,
> the gravity module does not know about the hydro boundary conditions
> the problem shows up most prominently when including self-gravity.
>
> We prepared a few images on our web-page to illustrate what's going
> wrong:
>
>  http://www.ita.uni-heidelberg.de/~tpeters/flashbug.shtml
>
> Essentially, the graviational potential is erroneous in the case of
> outflow BCs and right if we use perodic BCs. This happens when the refinement 
> structure changed (i.e. new blocks are added). This general
> result is independent of the setup (in collapse calculations, the other 
> quantities are subsequently screwed up).
>
> Could it be that not all guard cells at the edge of the simulation box
> are updated properly when using outflow/reflecting BCs?
>
> Do you know about any problems in connection with the non-periodic BCs?
>
> Thanks for your support.
>
> All the Best,
> Robi
>
> ======================================
> Robi Banerjee, Research Associate
> Institut fuer Theoretische Astrophysik
> Albert-Ueberle-Str. 2
> 69120 Heidelberg
> Germany
> e-mail: banerjee at ita.uni-heidelberg.de
> tel.  : +49 6221 54-8967
> fax   : +49 6221 544221
>