[FLASH-USERS] Problem with applyBCEdge

[James Guillochon]

Hi Klaus, sorry about the sparse details.

- I have included that line in the Config file, I can verify that OneRow is included in the compile. The xl boundary is set as "user".
- Those squares are blocks, the slice shows the whole y-z extent of the simulation, but it's a slice through a 3D simulation that extends into the x-direction as well.
- I showed a slice after it had evolved for several time-steps, but the pattern is present even at time-step 1.
- The boundary is supposed to be a constant velocity in the +x direction, with a fixed density/temperature. The boundary is almost identical to the WindTunnel problem except that I am using the Helmholtz EOS to calculate the derived variables from the given density/temperature. I am including the following modules in the Config file:

REQUIRES Driver                                               
REQUIRES physics/Hydro                                        
REQUIRES physics/Eos/EosMain/Helmholtz                        
REQUIRES Simulation/SimulationComposition                     
REQUIRES Multispecies
REQUIRES Grid/GridBoundaryConditions/OneRow
REQUIRES physics/Gravity/GravityMain/Poisson/Multipole

-- 
James Guillochon
Department of Astronomy & Astrophysics
University of California, Santa Cruz
jfg at ucolick.org

On Dec 1, 2009, at 12:55 PM, Klaus Weide wrote:

> On Tue, 1 Dec 2009, James Guillochon wrote:
> 
>> Hi all,
>> 
>> So I am using the Grid_applyBCEdge.F90 from the WindTunnel problem to 
>> include a fixed boundary condition in my simulation (-X boundary). It 
>> seems to partially work on the interface, but every other block doesn't 
>> seem to be applying the boundary condition, which results in a bizarre 
>> checkerboard pattern (attached to this e-mail). I think the issue must 
>> be in the looping indices in the region specialized function, which I 
>> have not modified. I am using FLASH 3.2, and the problem exists with the 
>> default ppm4 and ppm4dev. Any ideas? The attached image shows a slice 
>> through the grid cells just interior to the boundary cells.
> 
> James,
> 
> The checkerboard pattern looks quite ... interesting, but without more 
> information about the context it it hard to interpret.  Are those 
> checkerboard squares cells or blocks?  Are you showing one block
> or the whole domain? (Are they actually the same thing?) How many time
> steps does it take for this pattern to develop?
> 
> One very basic thing to check: since you took the Grid_applyBCEdge.F90 
> form the WindTunnel simulation, make sure you have also copied relevant 
> lines from the Config file.  In paricular, you should have copied the
> following line from WindTunnel's Config:
> 
> REQUIRES Grid/GridBoundaryConditions/OneRow
> 
> Let me know if this helps!
> 
> Klaus
> 
> !DSPAM:10135,4b1582c478871625623900!
>