Skip to content

Combines open source packages to produce an automated fusion specific neutronics workflow

License

Notifications You must be signed in to change notification settings

fusion-energy/fusion_neutronics_workflow

Repository files navigation

CI with docker build

CI with install

docker-publish-release

Fusion Neutronics Workflow

fusion neutronics workflow

Diagram showing the connectivity of software packages that make up the Fusion Neutronics Workflow. Software that the user interacts with directly are shown in blue.

This repository contains a containerized neutronics workflow for carrying out standard neutronics simulations in a repeatable manner.

There are workflows for producing standard neutronics simulations such as dose maps, DPA tallies, neutron/photon spectra, 3D VTK visualizations and the aim is to provide working examples for mainstream fusion neutronics analysis. Contributions are welcome.

This repository combines the several packages to create neutronics workflow that is:

  • Automated - the entire workflow can be run automatically. Care has been taken to code out all manual processes so that the neutronics analysis can be performed in an API manner. As a result it is therefore possible to drive the workflow with machine learning, optimization or parameter space sampling methods.
  • Scalable - by using efficient interfaces (Embree, Double Down) and accelerated geometry (DAGMC) together with scalable Monte Carlo transport (OpenMC) the resulting workflow scales well with computational power.
  • Deployable - By integrating open source software preferentially and ensuring compatibility with containerization the combined workflow is straight forward to deploy. Instances have been run everywhere from cloud computing to local compute.
  • Modular - individual packages created as part of the workflow are all available to install via PyPi pip installers or Conda installs. This allows separate components of the workflow to be easily installed as well as the workflow asa whole. All interfaces are exposed and there is no vendor lock in for any aspect.
  • Documented - documentation of installation procedures, use cases, examples, tutorials and even demonstration videos are all available.

Dockerfile

Quick start

The Dockerfile can build locally or a prebuilt one can be downloaded with

docker pull ghcr.io/fusion-energy/fusion-neutronics-workflow

The docker image can then be run with

docker run -it ghcr.io/fusion-energy/fusion-neutronics-workflow

Sharing files

To map a local folder to the docker container a volume mount can be used. The following example mounts the local folder that the command is being run from with the /home/fusion-neutronics-workflow/examples that is inside the docker images

docker run -it -v $PWD:/home/fusion-neutronics-workflow/examples ghcr.io/fusion-energy/fusion-neutronics-workflow

Performance builds

There are additional higher performance docker images available if your CPU supports vectorization instructions available Embree.

Download a prebuilt docker image with Embree and Double Down enabled

docker pull ghcr.io/fusion-energy/fusion-neutronics-workflow:embree

Download a prebuilt docker image with Embree compiled with AVX instruction set and Double Down enabled

docker pull ghcr.io/fusion-energy/fusion-neutronics-workflow:embree-avx

Software Packages included

Links to the packages that are utilized by the fusion-neutronics-workflow

  • Open source projects created and maintained

    • Paramak - automated production of fusion reactor CAD models (stp and stl files) from parameters.

    • stl_to_h5m automated conversion of STL files to h5m files compatible with DAGMC enabled particle transport codes.

    • openmc_dagmc_wrapper allows one to quickly utilise the h5m geometry files in a range of standard OpenMC based neutronics simulations.

    • neutronics-material-maker Create and customise materials from an internal database or from your own recipe, export to a wide range of neutronics codes.

    • openmc_tally_unit_converter converts the units of common tally such as heating, DPA, effective dose into user specified units.

    • openmc_data_downloader performs on the fly downloading of nuclear data needed for OpenMC neutronics simulations.

    • openmc_data downloading and processing of nuclear data needed for OpenMC neutronics simulations.

    • openmc-plasma-source Creates a plasma source as an openmc.source object from input parameters that describe the plasma

    • spectrum_plotter A Python package for creating publication quality plots for neutron / photon / particle spectrum

    • openmc_mesh_tally_to_vtk A Python package for converting OpenMC mesh tallies to VTK files and optionally converting the units

    • regular_mesh_plotter A Python package for plotting regular mesh tally results from neutronics simulations.

    • dagmc_geometry_slice_plotter A minimal Python package that produces slice plots through h5m DAGMC geometry files

    • dagmc_bounding_box Finds the bounding box and related properties of a DAGMC geometry

    • remove_dagmc_tags A python package and command line tool for removing DAGMC tags such as graveyard and vacuum

    • dagmc_h5m_file_inspector Extracts information from DAGMC h5m files including volumes number, material tags

    • brep_to_h5m Converts Brep CAD geometry files to h5m geometry files compatible with DAGMC simulations

    • brep_part_finder A Python package to identify the part ID number in Brep format CAD files

  • Open source projects that are utilized and contributed to

    • OpenMC The OpenMC project, a Monte Carlo particle transport code based on modern methods.

    • DAGMC Direct Accelerated Geometry Monte Carlo Toolkit

    • SphinxCadQuery An extension to visualize CadQuery 3D files in your Sphinx documentation

  • Open source projects utilized in the software stack

    • MOAB and pymoab the Mesh-Oriented datABase MOAB is a component for representing and evaluating mesh data.

    • CadQuery A python parametric CAD scripting framework based on OCCT

    • Double-Down A double precision interface to Embree via the Mesh Oriented dAtaBase (MOAB).

    • Embree high-performance ray tracing kernels

    • Gmsh A three-dimensional finite element mesh generator with built-in pre- and post-processing facilities

    • Scipy, Numpy, Plotly, Pint and other standard scientific Python packages