On-the-fly sampling of temperature-dependent thermal neutron scattering data for Monte Carlo simulations

• Method of on-the-fly sampling S(α,β) data at any temperature is proposed.
• Temperature dependence of thermal energy and momentum transfer CDFs is studied.
• Functional fits are found to the data using a regression analysis.
• Fits can construct the CDFs at any temperature at thermal energies.
• Storage of fitting coefficients is much less than the current ACE data.

Temperature can strongly affect the probabilities of certain neutron interactions (fission, capture, scattering, etc.) with materials. These probabilities are referred to in the nuclear community as ‘cross sections’ and are used as inputs for computer simulations. During the lifetime of a nuclear reactor, the core and its surrounding materials will experience a wide range of temperatures. To simulate the neutronic behavior in a realistic core, it is required to pre-store a large amount of cross section data to encompass the entire temperature range a neutron may experience. In recent years, methods have been developed to reduce data storage and obtain the cross section at the desired temperature ‘on-the-fly’ during radiation transport simulations using Monte Carlo codes. At thermal energies, however, the scattering of neutrons is complicated by their relatively small wavelengths, making molecular binding and lattice effects significant. Current approaches typically require nuclear data file sizes of tens to hundreds of MB per temperature, which can be prohibitive for realistic reactor physics simulations. To reduce the storage burden, a fitting approach in temperature is investigated that allows for the efficient evaluation of the thermal neutron scattering physics at an arbitrary temperature within a predefined range. The physics for thermal neutron scattering in graphite and hydrogen in water are evaluated with this approach. In both cases, the functional fits are able to accurately reproduce the scattering probabilities. The data storage for the fitting approach requires only a few 100 kB, which is a significant memory savings over the existing methods. These data can be used to sample a neutron's outgoing energy and scattered angle at an arbitrary temperature with minimal errors.