[FLASH-USERS] setting up a user BC

[Klaus Weide]

On Tue, 3 Aug 2010, turcotte wrote:

> Dear all,
> 
> I am trying to set-up an inflow entering the computational domain from the lower 
> y-axis boundary in 2-D cylindrical coordinates.
> Initially I used the same set-up as the WindTunnel example, modifying  
> Grid_applyBCEdge.F90 to have a uniform inflow across the lower boundary.
> For example, the density of the inflowing material is assigned like:
>         case (DENS_VAR)
>            do i = 1,guard
>             dataRow(i)= max(rho_initial,sim_rho_ambient)
>            end do
> 
> I need to implement a more complex BC however. To do so I define the inflow in  
> Grid_bcApplyToRegionSpecialized.F90. 
> My first step was to try to reproduce with Grid_bcApplyToRegionSpecialized the 
> simple uniform inflow.
> I expected that something like this:
>          case (DENS_VAR)
>             do i = 1,guard
>                regionData(i,:,:,DENS_VAR)= max(rho_initial,sim_rho_ambient)
>             end do
> would yield the same inflow as before but it's not the case, the inflow is not 
> uniform. There are variations that are aligned with block boundaries. 
> 
> All else is identical in the two formulations, such as EOS, physical parameters 
> of the inflow, etc...
>  
> What am I missing?

Sylvain,

Not sure, at a first glance what you are doing should work, but the code 
snippets don't give a lot of context.

Some thoughts:

Why do you need a select/case (and, I assume, a surrounding loop over 
variables) at all?  (Maybe there is a good reason, I am just curious.)
Are you making sure to define regionData for ALL defined variables?
(You should have a 'case default'.)

Are you sure that max(rho_initial,sim_rho_ambient) is constant (at least
for a given time)?

What do the variations look like, what are their relative sizes, and in 
which variables?

In the default WindTunnel configuration you would use the 
Grid_bcApplyToRegionSpecialized from the "OneRow" directory. To narrow 
down the problem, maybe it would be helpful to start with a copy of that 
version of Grid_bcApplyToRegionSpecialized.F90, modifying (initially at 
least) only one or a few variables in regionData after the calls to
Grid_applyBCEdgeAllUnkVars / Grid_applyBCEdge ?


Klaus