[FLASH-USERS] combining MHD and thermal conduction

[Klaus Weide]

On Fri, 11 Aug 2017, Slavin, Jonathan wrote:

> Hi all,
> 
> I've been trying to do supernova remnant evolution runs that include
> electron thermal conduction and MHD using the unsplit staggered mesh unit.
> I can do runs that include either electron thermal conduction or MHD but
> when I try to combine them I get a rapid (after 43 steps) failure with:
>  ... |initial RHS|^2=3.6549+240 ...

This indicated that the variable you are trying to diffuse - temperature, 
I assume - is extremely high already on entry to the diffusion solver.
This could be either from previous calls of the diffusion solver, or
from something else having gone wrong in previous time steps.

It would be interesting to know whether temperature blows up only in 
certain locations. I suggest looking at some plotfiles (or checkpoints) 
before step #43.

If you find that the problem is caused only in low-density regions, you
can try disabling thermal diffusion there by setting runtime parameter
'diffusion_cutoff_density'.

Klaus