Star-CCM+ — Design Manager¶

Design Manager is used when you want to run multiple Star-CCM+ cases that differ by geometry, boundary conditions, or parameters — such as parameter sweeps, design of experiments (DOE), or optimization studies. Instead of manually submitting each case, Design Manager coordinates and tracks them all within a single job allocation.

For running a single simulation, see Single Case.

Prerequisites¶

Before submitting, make sure you have: - A configured Design Manager project file (.dmprj) - Your .sim file referenced inside the Design Manager project - A valid license key or license server address - "Linux Cluster" selected as the compute resource inside the Design Manager project file — this is required for the pre-allocation method used on Vega

How Design Manager Works on Vega¶

Vega requires the pre-allocation method, meaning you request all resources upfront in your job script. Design Manager then uses that allocation to run multiple cases in parallel, up to the number of simultaneous jobs you configure.

For example, if you request 2 nodes (384 cores) and configure Design Manager to run 4 simultaneous jobs at 96 cores each, it will keep up to 4 cases running at a time until all cases in the project are complete.

Note

There is an alternative method called General Job Submission where the Design Manager controller submits individual jobs to the scheduler as cases complete. This is not supported on Vega. Documentation for this method is coming soon.

User Inputs¶

The USER INPUTS section at the top of the script is the only part you should need to edit:

Variable	Description
`DM_PROJECT`	Name of your `.dmprj` file
`CASELOG`	Name of the output log file
`NPROC_PER_JOB`	Number of cores allocated to each individual case
`NSIMULT_JOB`	Number of cases to run simultaneously
`PODKEY`	Your Star-CCM+ pod key (if using pod licensing)
`LICS`	License server address (if using a license server)
`LICOPT`	Comment/uncomment to switch between pod key and license server

Make sure your total cores requested (nodes × ppn) is at least NPROC_PER_JOB × NSIMULT_JOB. For example, 4 simultaneous jobs at 96 cores each requires at least 384 cores (2 nodes).

What the Script Does¶

1. Setup The script uses sed to update NumComputeProcesses and NumSimultaneousJobs directly in the .dmprj file to match what you specified in the user inputs. It then creates a scratch folder, copies your project files over with rsync, and moves into the scratch directory.

2. Execution Design Manager is launched using starlaunch, which handles distributing cases across the allocated compute nodes. It reads the node list from $PBS_NODEFILE and manages job placement automatically. Output is redirected to your log file and elapsed time is recorded to stat.out.

3. Cleanup Once all cases are complete, results are copied back from scratch to your home directory and the scratch folder is removed.

Script¶

Download or reference the full script here: scripts/design_manager.sh

Next: Vega Hardware Reference