A functional digital model of the Dingo thermal neutron imaging beamline.

Abstract

In this work, we extend our previously published Monte Carlo simulation model of the Dingo thermal neutron beamline at the Australian Centre for Neutron Scattering model by (1) including a sapphire crystal filter in the model, and (2) utilising the NCrystal package to simulate thermal neutron interactions with the crystalline structure. In addition to previous experimental measurements performed in the beamline's high-resolution mode, the beam was experimentally characterised in its high-intensity mode upstream from the sample stage (at the tertiary shutter wall exit) and these measurements were used as inputs for the model. The planar neutron distributions were optimised at both the sample stage and tertiary shutter wall exit, and model predictions were validated against experimental gold wire activation measurements. For both configurations-with and without the sapphire filter-we measured neutron fluxes, and performed neutron activation analysis using 11 materials to improve the accuracy of the neutron spectrum in the model relative to the original version. Using the optimised spectrum, we simulated out-of-beam neutron spectra that were further used as the initial input in unfolding code to explore the capability of the current solution to accurately reproduce the experimental results. The normalised neutron planar distribution from the simulation was on average within 2% at the centre, and 6% and 24% at the penumbra of the experimental results at the tertiary shutter wall exit and sample stage, respectively. The specific activities predicted by the refined model were within an average of 13% and 5% of the experimentally measured activities with and without the sapphire filter, respectively. We observed a decrease of around 45% in thermal neutron flux when the sapphire filter is used, which has been reproduced by the model. The maximum value of the logarithm of the ratio of simulated to experimental out-of-beam neutron spectra across 8 locations was 0.6 compared to 2.0 in the previous work, resulting in an average normalised root mean squared error between the unfolded spectrum and experimental measurements of 5% and 9% with and without the filter, respectively. Without the sapphire filter, the optimised predicted in-beam neutron spectrum consists of around 59% thermal, 21% epithermal and 20% fast neutrons, while the addition of the filter provides an almost pure (approximately 98%) thermal neutron beam.