OpenFOAM

OpenFOAM#

OpenFOAM is an open-source computational fluid dynamics (CFD) toolbox widely used for simulating fluid flow, turbulence, heat transfer, and other complex physical processes. For more details, refer to the official OpenFOAM documentation.

Available Versions on Aire#

Version

Command

v2412

module load openfoam/v2412

How to Submit a Job#

Below are examples of job submission scripts for running OpenFOAM on Aire.

Serial jobs#

For small-scale simulations or testing purposes, you can run OpenFOAM in serial mode:

#!/bin/bash
#SBATCH --job-name=openfoam_serial
#SBATCH --time=01:00:00
#SBATCH --mem=4G
#SBATCH --nodes=1
#SBATCH --ntasks=1

module load openfoam/v2412

# Generate the mesh
blockMesh > log.blockMesh

# Run your OpenFOAM case
simpleFoam > log.simpleFoam

Parallel jobs#

For larger simulations, OpenFOAM can be run in parallel using MPI. Below are examples for running a 16-core job on a single node and a multi-node job.

Single-node jobs#

For a 16-core job on a single node:

#!/bin/bash
#SBATCH --job-name=openfoam_parallel
#SBATCH --time=01:00:00
#SBATCH --mem=128G
#SBATCH --nodes=1
#SBATCH --ntasks=16

module load openfoam/v2412

# Decompose the domain for parallel execution
decomposePar

# Run the simulation in parallel
mpirun -np $SLURM_NTASKS simpleFoam -parallel > log.simpleFoam

# Reconstruct the results after the simulation
reconstructPar

Multi-node jobs#

For a multi-node job, adjust the script to request multiple nodes and distribute tasks across them:

#!/bin/bash
#SBATCH --job-name=openfoam_multi_node
#SBATCH --time=04:00:00
#SBATCH --mem=256G
#SBATCH --nodes=2
#SBATCH --ntasks=64
#SBATCH --ntasks-per-node=32

module load openfoam/v2412

# Decompose the domain for parallel execution
decomposePar

# Run the simulation in parallel across multiple nodes
mpirun -np $SLURM_NTASKS simpleFoam -parallel > log.simpleFoam

# Reconstruct the results after the simulation
reconstructPar

Note

  • Single-node job: Ensure --ntasks matches the number of cores available on the node.

  • Multi-node job: Adjust --nodes, --ntasks, and --ntasks-per-node based on the number of nodes and cores per node.

  • Ensure your decomposeParDict file is configured to match the total number of tasks ($SLURM_NTASKS).

Contents