SU2

SU2 is an open-source suite for computational fluid dynamics and PDE-constrained optimization, widely used in aerodynamics and aerospace, backed by remote compute. This image bundles SU2 8.0.1 with MPI; the container stays up so you can run solver and optimization commands against cases mounted at /data.

Configuration

name: su2
tags:
    - simulation
steps:
    - name: app
      platform: docker
      mode: parallel
      image: ghcr.io/dxflow-ai/su2:latest
      command:
          - tail
          - -f
          - /dev/null
      volumes:
          - name: volume
            host: ./volume
            container: /data
      resources:
          cpu: "16"
          memory: 32G

[volume]
app.volume = ./volume

[resource]
app.cpu = 16
app.memory = 32G

{
    "arch": ["amd64"],
    "image": "ghcr.io/dxflow-ai/su2:latest",
    "version": "8.0.1",
    "minimum": {
        "cpu": 8,
        "memory": "16G",
        "storage": "50G"
    }
}

Usage

1. Deploy

dxflow workflow create --identity su2 su2.yml
dxflow workflow start su2

The container stays up so you can run SU2 commands against data mounted at /data.

2. Run a case

Put your .cfg config and .su2 mesh under /data, then run inside the workflow container:

cd /data

# Serial run
SU2_CFD case.cfg

# Parallel run (N ranks)
mpirun -np 16 SU2_CFD case.cfg
# or the Python driver
parallel_computation.py -f case.cfg -n 16

3. Retrieve results

Everything under /data persists — restart files, surface/volume solutions, and history are written there. Visualize with the ParaView image.

Common tools

Notes

• SU2_RUN is set and the SU2 binaries and Python drivers are on PATH / PYTHONPATH, so SU2_CFD and parallel_computation.py work out of the box.
• MPI is allowed to run as root (OMPI_ALLOW_RUN_AS_ROOT), so mpirun -np N SU2_CFD runs without extra flags.

References

• Website: SU2code
• Documentation: SU2 Docs
• Tutorials: SU2 Tutorials