[FLASH-USERS] Unsplit MHD Solver PPM problem

[dongwook at flash.uchicago.edu]

Dear Sam,

What kind of physics do you include in your simulation?
For example, do you have diffusion terms, gravity,
or just hyperbolic system?

In general, PPM has very small numerical dissipations
compared to the MUSCL-Hancock scheme, and the PPM reconstruction
can generate negative states.

If you have strong shocks in your simulation especially,
you may need artificial viscosity for PPM solver to add gain
more numerical stability in strong shock regions.

Unfortunately, this artificial viscosity implementation
has not been released in FLASH3.3. If you want to try
this feature, please let me know.

Also, if you can let me know what flags you already tried,
it will be useful for me to figure out the issue.

We may take further email communications offline to address
some more specific issues in your setup.

Best,
Dongwook


> Dear FLASH,
>
> I'm running simulations with magnetized shocks and I wanted to try using
> the higher order solvers in the unsplit MHD solver.  FLASH3.3 includes a
> 3rd order PPM solver.  I set up my problem in 2D Cartesian coordinates
> (using AMR) and the simulation crashes after about 60 timesteps.
>
> I run the same simulation with the 2nd order MUSCL-Hancock solver and the
> simulation runs no problem.  I tried playing around with various flags,
> but nothing seemed to help.  Any ideas / suggestions?
>
> Sam
>