Angular distribution of fast neutrons outside iron shielding

Abstract

where E (g) is the total collision cross section; d is the thickness of the layer. The particles remaining were assigned a weight [2s/E](1-P), where ~ is the scattering cross section. Histories were terminated when the particle weight became less than a given quantity, which usually was 0.1% of the penetration probability for unscattered radiation. The angutar neutron spectra outside a layer were obtained from a probability analysis over energy and direction of the neutrons escaping from the layer. The program wl~ch was made up in accordance with this scheme provided a determination of the coordinates of the point of neutron escape from the layer and atso output of data in the form of a 16-group energy spectrum for each of eight angular intervals (these intervals were selected with uniform steps in the cosine df th~ angle of calculations). For neutron energies g - 4 MeV, the statistical model of the nucleus was assumed valid, the excitation temperature of the nucleus being assumed to be T = 0.80 MeV [10], On the basis of experimental data in [11], corrections for direct interaction of neutrons with nucleons in the iron were introduced for E > 7 MeV. Results of the Monte-Carlo calculations were compared with multigroup calculations by the "transmission matrix" method in the 2P 7 approximation (ROZ program [2]). The angular and energy spectra from a plane, monodirectional source with the energy spectrum of a D--D source (Ema x = 3.4 MeV) outside a layer of iron 10 cm thick were calculated. The angular dependence of a given energy group was remarkably close in both cases, but some difference were observed in the relationship between groups, i.e., in the energy spectra. The Monte-Carlo calculation gave a softer spectrum. In the two cases, the maximum difference in the absolute number of neutrons escaping from tile layer in each group did not exceed 25%. Hence, it follows that the ROZ program, which requires considerably less machine time, can be used successfuUy in the one-dimensional case. In that case, it is expedient to monitor mass calculations with the ROZ program by periodic Monte-Carlo calculations. Note that the advantage in machine time increases with increasing thickness since the time used increases practically in proportion to attenuation in the Monte-Carlo method, and is proportional to thickness in the ROZ calculation.