[FLASH-USERS] E-MHD velocities and dt computation BUG

[Loffelmann, Jiri]

Hello FLASH users,

I discovered what seems to be an implementation bug in the new FLASH version 4.7.
It concerns the computation of flux velocities and corresponding timesteps in the Extended MHD model. This is carried out in the function "hy_uhd_computeDtExtMHD.F90" inside the folder ".../source/physics/Hydro/HydroMain/unsplit".
The bug is on line 191, where an uninitialised variable 'tauCent' is used for the computation of the magnetization 'chi'. This results in the chi being zero. The same problem is on line 204 for the computation of 'etatau', which also ends up always being zero.
Further use of the variables 'tauCent', 'chi' or 'etatau' in the following computations results in wrong values (usually zero) of some of the flux velocities (Nernst, Hall, Resistive, ..) and incorrect computation of the time step limitation. Finally, this can lead to instability during the simulation for some problems.

I used a quick fix that corrected the computation of the velocities and erased the instability in my simulation:
Insert line "tauCent = tau" after line 189.

I am happy to provide further details and the setup of the simulation where the instability appears. But I believe that the problem in the implementation should be obvious.

Regards,
Jiří Löffelmann

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://flash.rochester.edu/pipermail/flash-users/attachments/20230411/41d0a057/attachment.htm>