Interactive sessions are recommended only for pre- and postprocessing. Larger problems should be run as batch jobs for either serial runs or parallel runs of Fluent.
Taking Fluent into use
On murska.csc.fi and hippu.csc.fi the whole Fluent.Inc suite (Fidap,Fluent,Gambit) is taken into use giving the command
module load fluent/fluent.inc
The standard compiler suite on murska.csc.fi and hippu.csc.fi, the Portland Group compiler, is not supported by ANSYS-Fluent. We thus recommend to switch to the GNU compiler environment
module switch PrgEnv-pgi PrgEnv-gnuif the user wishes to compile User Defined Functions (UDF). Else the last step is not necessary.
Launching Fluent
Thereafter commands for starting interactive sessions is
fluent [options]
The user may check the possible command line options by entering the command
fluent -help
Among others, the following options for the solver Fluent can be given at the command line:
| 2d |
starts two dimensional, single precision solver |
| 2ddp | starts two dimensional, double precision solver |
| 3d | starts three dimensional, single precision solver |
| 3ddp | starts three dimensional, double precision solver |
| -g |
run without GUI and graphics (important for batch jobs) |
| -i journalfile | run given journal script journalfile (important for batch jobs) |
| -gu |
run without GUI (graphic plots still active) |
| -gr |
run without graphics (GUI still active) |
| -driver driver |
run specific graphics driver. driver can be gl | opengl | null | pex | sbx | x11 | xgl. In most cases one has to choose -driver x11 |
| -post |
run post processing version only - same as command flpost |
| -t n |
enable parallel processing using n processors |
| -r release |
start older release of Fluent. Currently on corona installed releases are: 6.1 and 6.2 (launched by default) of Fluent 6. Available releases are listed if the option is given without an argument |
Commands inside Fluent can either be accessed through the pop-up menus of the GUI or through the text based interface.
Some useful hints
- When you are using a ssh connection it is recommended to set the flag -f. This flag makes the connection faster.
- Information about available versions, releases, etc. before starting up the solver can be obtained by typing fluent -help
- if a Fluent job hangs and does not react on any input to the GUI or the text-menu, the user may use the script kill-fluentpid, where pid
is the pid of the Fluent launch process. This might be of particular
use in parallel runs, where else a quite large amount of single
processes would have to be manually stopped by the user.
To give an example: If the interactive Fluent session of the GUI in the picture above and all the subsequently launched processes should be stopped the following sequence of commands is the solution for doing so:
cd $HOME
chmod u+x kill-fluent18397
$HOME/kill-fluent18397 - In case of not using X11-tunneling (via settings in ssh-client
on Windows, and -x option for ssh on UNIX systems), remember to set the
variable for the graphics output, DISPLAY, properly, e.g.,
If you get an error message of style 'Xlib: Client is not authorized to connect to Server' give the permission to direct the output to your terminal. If you are running X-windows this is done by typing the commandsetenv DISPLAY your.address:0.0
at the shell prompt.xhost +corona.csc.fi
- In general it is recommended to have X11-tunneling enabled. On Windows based environments this is set in the Preferences of the ssh-client. On Linux/UNIX based systems running X11 the option -x has to be added to the ssh-command.
- If your hardware does not support accelerated graphics (OpenGL) then the option -driver x11 should be added to the Fluent launch command
- Some special models (e.g. special versions of turbulence models) are only to be invoked via the text interface.