Modeling and simulation of neutron detectors for the transient reactor test facility

Abstract

The Transient REActor Test (TREAT) facility was restarted and will be used to test accident-tolerant fuels to improve nuclear reactor safety. In this work, alternative neutron detectors for use in core and with the hodoscope at the TREAT facility were modeled and simulated using different computational tools to understand the underlying physics. The Hornyak button scintillation detector used in the original TREAT hodoscope to detect fast neutrons and its variants were evaluated using Geant4 to simulate the coupled nuclear and optical physics. The Hornyak-button model predicted an intrinsic efficiency of 0.35% for mono-directional fission neutrons and strong gamma-induced Cherenkov noise, which agree relatively well with the reported experimental observations. The proposed variants use silicon photomultipliers to reduce Cherenkov noise and have optimized layered or homogenized scintillation volumes. The layered and homogenized variants with 5-cm length were predicted to achieve neutron-detection efficiencies of 3.3% and 1.3%, respectively, at a signal-to-noise ratio of 100. Another candidate devices for the hodoscope are the actinide and hydrogenous microstructured semiconductor neutron detectors (MSNDs) evaluated using Geant4 and MCNP. With a sufficient rejection of the gamma noises, the U235 -filled and the hydrogenous MSNDs were predicted to yield neutron-detection efficiencies of 1.2% and 2.5%, respectively, at the length of 2 cm. The micro-pocket fission detectors (MPFDs) were developed to detect in-core neutrons, and the electron collection process in such devices was evaluated using Garfield++-based computational routine. The high-performance Garfield++ application was developed using the built-in, optimized element-search techniques and a hydrid MPI and OpenMP parallelization scheme. The preliminary results indicated that the averaged deposited energy per fission fragment was 7.15 MeV, and the induced current occured within 400 ns.