satcomx00/fds repository overview

⁠Fire Dynamics Simulator (FDS) Dockerfiles

Fire Dynamics Simulator (FDS) is a large-eddy simulation (LES) code for low-speed flows. This repository contains Dockerfiles for the Fire Dynamics Simulator (FDS) by NIST, aimed at providing a convenient and consistent FDS simulation environment across various operating systems.

Note: This image is maintained independently and is not officially provided by NIST.

⁠Docker Repository

All versions of the FDS Docker image can be found at satcomx00/fds on Docker Hub⁠.

⁠Supported Tags

The Docker image supports several tags, each corresponding to different versions of FDS. You can pull the latest version using the latest tag, or specify a version directly (e.g., 5.5.3, 6.5.3, etc.).

^*1 Running with Docker Desktop based on Hyperkit (Mac OS) or WSL 2 (Windows) which are lightweight virtualization solutions for Linux Docker containers.

^*2 Running with warning⁠. This should not affect the functionality of FDS.

⁠Prerequisites

• Docker installed on your system
• Basic knowledge of Docker command-line tools
• Understanding of FDS simulations if running specific simulation tasks

⁠Installation

Pull the Docker image from Docker Hub:

docker pull satcomx00/fds:latest

⁠Usage

⁠Basic Testing

To test if FDS is working properly (without running a simulation):

docker run --rm satcomx00/fds fds

To test a specific version:

docker run --rm satcomx00/fds:6.7.3 fds

⁠Running Simulations

To run an FDS simulation, you'll need to provide several environment variables and mount your working directory:

docker run --rm \
  -e FDS_FILE_PATH="/workdir/simulation.fds" \
  -e MPI_PROCESSES=4 \
  -e OMP_NUM_THREADS=2 \
  -e INPUT_ARCHIVE_DIR="/workdir/output" \
  -v $(pwd):/workdir \
  satcomx00/fds:latest fds /workdir/simulation.fds

⁠Required Environment Variables

⁠OpenMP Configuration

The container automatically configures OpenMP with the following settings:

• OMP_PROC_BIND="close" - Binds threads close to the master thread
• OMP_PLACES="cores" - Places threads on physical cores
• OMP_DYNAMIC="FALSE" - Disables dynamic adjustment of threads
• OMP_SCHEDULE="static" - Uses static scheduling policy

⁠Output

After simulation completion, the container will create a zip archive containing the simulation results. The archive will be named <simulation_name>-output.zip and will be located in your mounted working directory.

⁠Container Specifications

• Unlimited stack size for MPI optimization
• Unlimited virtual memory for MPI optimization
• Automatic zipping of output folder

⁠Notes

• The FDS input file must contain a &HEAD CHID= parameter
• The container uses hybrid parallelization with MPI and OpenMP
• Total cores used = MPI_PROCESSES × OMP_NUM_THREADS
• All paths in environment variables should be relative to the container's /workdir directory

⁠Support

For issues and feature requests, please visit the GitHub repository⁠.