convecting an inert scalar with a robust solver

[Brian]

Hi guys,

I am trying to model a non-reactive case of high pressure hydrogen
being released into air (a pressure ratio of 100 to 1000). If I solve
the EulerEquations I have to use a robust solver like the Godunov
solver or the simulation will fail. The Roe solver generates a NAN
because it cannot handle the high strength shock that is generated in
the simulation. What I would like to do is convect an inert scalar
(I'll call it alpha) in order to track where the two gases are
throughout the release process. The equation I would like to add to
the solver is:

d(rho*alpha)/dt + d(rho*u*alpha)/dx + d(rho*v*alpha)/dy = 0

One approach I have tried is to use the ReactiveEulerEquations set and
use Z as my tracking variable. I made it non-reactive (or inert) by
setting the solver variables Q and K to zero. However I can only find
the Roe solver option in the ReactiveEulerEquations. Is there a
robust solver I can use with the ReactiveEulerEquations in order to
run a simulation that can handle strong shocks, or am I better off to
add the equation above to the EulerEquations via the add2W option? If
so is there a way I can easily integrate alpha using the godunov
solver routine that calculates the other variables (RHO, P, U, V, E)?
Has anyone tried this, or have examples?

Thanks,

Brian

[James Quirk]

Brian,

Try running the example:

amrcp bcg/axs.3

it generates 108 different solvers for the ReactiveEulerEquations.

Also, take a look at:

amrcp contrib/sk.mailit

it shows how you might construct a two material solver,
although note that the script could do with polishing
as t was written way back in 1999 and is a bit klunky.

James

>
> Brian