Improving neutron simulations

Experts from both (a) neutronics (ISIS, PSI, DTU, ESS, ESS-B) and (b) Monte Carlo ray-tracing (DTU, ILL, NPI, UCPH) have come together in SINE2020’s E-tools work package. They have been developing accurate, integrated simulation software to simulate the path of neutrons from their source to sample which will lead to more efficient use of beam time and help with the design of new instruments.

The work has involved developing the codes and software packages that simulate the production and transport of neutrons from the target, through moderators and reflectors (neutronics, e.g. MCNP), or simulate the movement of neutrons along guides and their interaction with detectors, samples and other materials in the beamlines (Monte Carlo ray-tracing, e.g. McStas).

MCNP and McStas are two of the main software packages that are currently used for simulations. The have different functionalities to offer a user but, unfortunately, they can not easily be applied together. However, the Monte Carlo Particle List developed by Thomas Kittelmann is helping to solve this problem. This is a way of storing particle data, along with its metadata, to allow particle-data to be transferred between one software package to another.

During the SINE2020 project, the E-tools work package team have been adapting several Monte-Carlo codes to be compatible with the MCPL particle list and hence allow them to become more integrated. In neutronics-software, MCPL is now supported by both MCNP, Geant4 and PHITS, and in Monte Carlo ray-tracing the format has been adapted by McStas, RESTRAX/SIMRES and Vitess.

With this task, there has also been a good opportunity to increase the functionality of McStas and MCNP, including the inclusion of unique features present in RESTRAX/SIMRES, for example. Read more about this here. DTU in Denmark has been the main hub of developing McStas features and MCNP has been developed by colleagues based in ESS-Bilbao in Spain, PSI in Switzerland and DTU in Denmark. As an example, the development team have been able to extend MCNP simulation capabilities to include supermirrors and neutron guides to work in combination with important variance reduction methods in MCNP. As a result, the modified MCNP allows the determination of both reflected and non-reflected intensity in a reasonable computation time.

With the developed methods and software interconnection capabilities, simulations of instruments and shielding can now estimate the experimental signal-to-noise, taking into account the physical properties of all of the material that together make up a neutron scattering instrument.

Benchmarking experiments have been performed on the BOA beamline at PSI and the ChipIR beamline at ISIS. The measurements were generally in good agreement with results of simulations, and hence provide important validation of the developed simulation methods.

In conclusion, neutron experiment simulations are much better understood, more functional and more useful than before. There is now a wealth of well-documented software tools and methods that instrument builders can use to plan their future neutron instruments and experiments, and will thus pave the way for optimized use of the ESS from day one.

Acknowledgments: Peter Willendrup, DTU