The Quest for Clean Energy: IAEA Challenge on Materials for Fusion


Modelling Damage in a Fusion Reactor

Nuclear fusion is the process by which the nuclei of two light atoms combine to produce a single nucleus whilst releasing a large amount of energy. A star like our own Sun shines because of energy released from a set of fusion reactions which convert hydrogen to helium.

A nuclear fusion power station would have many advantages over the nuclear fission reactors currently in use across the world (nuclear fission is the process of splitting heavy atoms of uranium or plutonium to release energy). A fusion reactor has the potential to produce cheap, clean energy without the dangers of long-lived radioactive waste or weaponisation.

In order to get two nuclei to fuse, they have to hit each other at immense speed (since both are positively-charged, there is a strong electrostatic repulsive force between them). In most designs of experimental reactor, the nuclei involved are two isotopes of hydrogen, deuterium (2D) and tritium (3T) which combine at a temperature of 100 million oC to form a helium nucleus (4He) and release an energetic neutron (1n):

2D + 3T → 4He + 1n

At such high temperatures, matter exists as a plasma of positively-charged nuclei and negatively-charged electrons. This plasma must be confined by a magnetic field to keep it from damaging the reactor walls and the details of maintaining a stable plasma confinement are the subject of ongoing research.

A further obstacle to a practical fusion reactor design concerns the neutrons which carry most of the energy that will ultimately drive turbines to produce electricity. Being uncharged, they are not confined by the magnetic field but pass through the reactor “first wall” to deposit their energy in a specially-designed “blanket” which efficiently slows them, releasing their kinetic energy.

This challenge is concerned with the material that might make an effective first wall: in the extreme environment near to the fusion plasma, such a material is exposed to high temperatures and bombardment by the highly-energetic neutrons and other particles. Candidate materials include tungsten, steel and beryllium. Since experiments on physical samples are difficult and expensive to carry out, scientists have turned to computational models to simulate the behaviour of a material.

Molecular Dynamics is the method used to simulate the physical movement of the atoms inside a solid material after the disruptive impact of a high-energy particle such as the neutron resulting from a fusion reaction. This impact can be profound for a local region, deep inside the material, creating a cascade of displaced atoms and creating voids, defects and dislocations in its structure. By following the movement of the atoms in the material, the damage created can be simulated and analysed. Different metals or compositions, impact energies and temperatures can be explored in this way and can help with the search for an effective first wall material.

The Data

The data provided are the positions of the atoms of either tungsten, W, or iron, Fe, after a collisional cascade molecular dynamics simulation run for 40 picoseconds. The initial state of the material is taken to be a perfect crystal, and the final state is provided as the (x, y, z) locations of the atoms after the energy of the impacting neutron has been absorbed. Four different impact energies are considered for each material and at least seven different simulations are run (for different impact directions) for each energy.

The Challenge

Participants are invited to come up with innovative ways to visualise, analyse and explore the provided data. The precise nature of the software solution to the challenge is deliberately left open to invite as many novel approaches as possible, but may involve one or more of the following:

  • Novel software for visualizing the material damage represented by the data files in a way that aids its qualitative and quantitative assessment.
  • New software tools to rapidly and reliably identify, classify and quantify new patterns and structures of particular kinds in the data sets.
  • Efficient algorithms to depict and summarise the statistical distribution of atom displacements and to analyse the effect of impact energy on this distribution.

It is envisaged that analysis techniques applied in the domains of tomography, medical imaging, protein crystallography, or computer vision may be applied in a modified form to address the proposed Challenge.


Participants shall provide the following deliverables by the challenge end date:

  • A completed solution description form submitted via this link outlining their analysis of the data and explaining the technical, mathematical or algorithmic approach to their solution.
  • The means to obtain a working executable version of the software developed, with full instructions for its compilation (if appropriate), installation and deployment. Submissions utilising open source technologies are particularly encouraged.
  • A well-described test suite of example input files to the submitted software application and their expected outputs.
Prizes and Incentives

The individual or team with the best submission, as judged by the advisory panel, will be awarded €5,000 and invited to IAEA Headquarters in Vienna to present their solution.


26 April 2018 Challenge opens

Participants register with the IAEA Challenge, providing their contact details, and subscribe to the Challenge, indicating their intent to participate. Registered participants will have access to a page providing additional technical details and instructions on how to submit a proposal. The FAQ & Contact page allows participants to send clarification questions; the IAEA team will answer these and regularly update a FAQ list until the close of the challenge to ensure that all participants have the same information.

23 June 2018 Challenge closes

Participants shall provide the deliverables described above in a suitable form by this date.

June – July 2018 Evaluation and Judging

Submissions will be evaluated and, where appropriate, compared by the advisory panel. Code provided may be executed on further data sets of simulated material damage. The following criteria will apply:

  • Scientific correctness, as benchmarked against existing software tools;
  • Visual impact in the depiction of important features in the data;
  • The innovative, efficient and effective application of algorithmic analysis techniques to the data;
  • Conciseness, ease of deployment and use


31 July 2018 Winner announced

The winning individual or team will be contacted, and an announcement made online and through the IAEA’s public communication channels. The winners will be invited to IAEA Headquarters in Vienna to present their solution.

Read full terms and conditions here