Pittsburgh Supercomputing Center 

Advancing the state-of-the-art in high-performance computing,
communications and data analytics.

FLUENT

FLUENT is a computational fluid dynamics solver that provides a wide array of advanced physical models for turbulence, combustion, and multiphase applications. 

Access to FLUENT is restricted. 

FLUENT is an ANSYS product, and you must have access to ANSYS to use FLUENT.

Corporate users must negotiate licensing with ANSYS, Inc. and arrange for an allocation with PSC. Contact the PSC Corporate Relations office to discuss an allocation of time.

Academic users must have a valid PSC account and be at an institution which has a local ANSYS license.

  • If you don't have a valid PSC account, see the instructions about applying.
  • If you don't know whether or not your institution has a local ANSYS license, contact your departmental or campus computing services to find out.

When you meet these requirements, please complete this form to request access to ANSYS.

Installed on blacklight.

Blacklight Usage

There are several steps you must follow to run Fluent.

First, you must start your job. Fluent will normally be run in an interactive manner. Our sample script runs Fluent in a batch mode, but this is just to provide a basic demonstration of the operation of Fluent. However, blacklight's interactive jobs must be run on blacklight's compute nodes. They should not be run on a blacklight frontend.

To run an interactive job on blacklight's compute nodes issue the command

    qsub -I -l ncpus=16 -l walltime=30:00:00

The -I option indicates that this an interactive job run on a blacklight compute node. This qsub command asks for 16 cores and 30 minutes of walltime. You can ask for up to 512 cores, because 512 is the number of Fluent licenses the PSC has. More information on the qsub command and on running interactive jobs on blacklight is available in the "Running Jobs" section of the blacklight user document.

Once you get the system prompt that indicates you are running interactively on blacklight's compute nodes you must issue the module command

    module load ansys/15.0

Fluent may be in run in either serial or parallel mode. Fluent is part of the ANSYS suite of software packges. Thus, you have to load an ansys module to Fluent. Information about the use of the module command is available online.

Now you can issue your fluent command. Your fluent command will look similar to

    fluent 3ddp -g -t$PBS_NCPUS -i pipe.in

In this command '3ddp' indicates the version of Fluent to use. Here, you are specifying the 3D double precision version. There is a also a 3D single precision version---specfied as '3d'---and 2D single and double precision versions---specified as '2d' and '2ddp' respectively.

The -g option specifies that you want to use the text user interface to Fluent not the graphical user interface. For performance reasons you should use the text user interface on blacklight.

The -t option indicates on how many cores you want to run. Here, you are specifying the number of cores you requested in your qsub command with the -l ncpus specification. To run a serial job omit the -t option.

The -i option is used to specify your Fluent journal file. The -help option lists the available options to the Fluent command. What other options you should use is dependent on your job.

After you enter the above command you will be placed in the text user interface to Fluent. You now have access to the text user interface's hierarchical menu of commands. Hitting <return> will display a list of commands available at each level. Top level commands are: adapt, display, define, file, grid, plot, report, solve, surface, view, and exit.

Entering one of these commands takes you into that command's submenu. The prompt changes to indicate where you are in the menu hierarchy. Again, typing <return> lists all available (sub)commands at that level. For example, entering the view command, and then hitting <return> produces this list: (user input in italics)

> view
/view> <return>
auto-scale         delete-view       read-views
camera/            list-views        save-view
default-view       restore-view      write-views

The "/" after "camera" in the above list indicates that "camera" is not a command but another submenu. The remaining choices from the view menu are commands. Typing "camera" at the "/view>" prompt produces this:

/view> camera
/view/camera> <return>
dolly-camera        position          up-vector
field               projection        zoom-camera
orbit-camera        roll-camera
pan-camera          target