[FLASH-USERS] Large numerical diffusion between chamber gas and target species in LaserSlab problem

[Richard Liska]

David,

what you see is caused by the fact that physically correct formulation 
would have moving boundary with outer vacuum, as it is typically used in 
the Lagrangian codes. In Flash you have ambient He instead of vacuum and 
thus instead of a Riemann problem with vacuum on one side on the boundary, 
you have a Riemann problem with material on both sides of the boudary with 
a very different solution, namely in ion temperature. And diffusion in the 
simulation is purely numerical diffusion as is the huge ion temperature. I 
think that your preplasma is from He and the huge ion temperature is 
around the boundary between the original preplasma and ambient He.

I do not think, that you could do something with this in Flash easily. The 
possibility exists, however it is not simple. You might introduce a new 
material - vacuum, however to avoid its mixing you would need to use 
material volume fractions in computational cells and do material interface 
reconstruction inside cells, so that you would have real material 
boundary with vacuum. This might be done for hydro, however it might be 
really difficult to make it work also with all the other physics included 
in Flash.

--Richard

On Tue, 22 Jul 2025, David Montgomery wrote:

> Dear Flash Users and Developers,
>
> We are running a modification of a 2D cylindrically symmetric laser slab 
> problem. The laser is normal incidence in the positive Z (positive Y in 
> hydro), with a supergaussian intensity profile.
>
> We have created a pre-plasma in the simulation_initBlock.F90 routine to 
> help mitigate laser imprint by initializing the He chamber gas to 10 eV 
> and having a hyperbolic tangent (TANH(x)) smoothing function on the 
> laser-irradiated side with width smaller than the target thickness.
>
> Early enough times in the run do not show any or much mixing between the 
> polystyrene target species and the low density Helium gas species, so we 
> don’t think that’s our problem.
>
> However, as the plasma expansion develops, a Sedov-like blast wave is 
> driven through the Helium chamber gas, getting to very high ion 
> temperatures (40 keV), and we see numerical diffusion between the 
> low-density helium chamber gas and the target (see attached plots). 
> Otherwise the electron temperature and electron density plots look 
> reasonable, especially when compared to 1D Lagrangian simulations 
> without the need for a “chamber gas”.
>
> Is there a way to suppress diffusion between the two species? We think 
> the artificially-high Sedov-like blast wave may be driving diffusion 
> between the two species.
>
>
> Thanks,
>
> David
>
>