American Nuclear Society

Home / Store / Journals / Electronic Articles / Nuclear Technology / Volume 168 / Number 2

Coupled Multigroup Proton/Neutron Cross Sections for Deterministic Transport

Charles T. Kelsey IV, Anil K. Prinja

Nuclear Technology

Volume 168 / Number 2 / November 2009 / Pages 257-263


Member Price:$27.00
Member Savings:$3.00

The limited availability of coupled multigroup proton/neutron cross-section libraries has hampered the use of deterministic transport methods for solving shielding problems involving energetic proton sources. Libraries are developed from evaluated nuclear data for low-energy transport and the physics models of MCNPX for intermediate-energy transport. They allow deterministic solutions of orbiting spacecraft shielding problems. Evaluated cross sections for protons and neutrons are available for many nuclides up to 150 MeV. NJOY99 is used to produce coupled multigroup proton/neutron cross sections from these. For higher energies, MCNPX is run in its cross-section calculation mode where the XSEX3 program is used to tally double-differential cross sections. The XSEX3 program was modified to discretize the cross sections in energy and output Legendre expansions for angular dependence. The NJOY99 and modified XSEX3 output are combined to produce cross-section libraries for energies up to 400 MeV. The libraries are used to solve trapped proton flux shielding problems using the discrete ordinates transport code Attila. High-order Legendre expansions (P39) are required to accurately describe the highly anisotropic scattering. Attila applies the extended transport correction allowing accurate three-dimensional solutions at much lower degrees. Particle flux solutions for orbiting spacecraft shielding problems obtained with Attila and MCNPX compare favorably. Coupled multigroup proton/neutron cross-section libraries, for use with deterministic transport codes, can be prepared using NJOY99 and MCNPX. Our results using the Attila code demonstrate that multigroup deterministic methods are computationally efficient alternatives to Monte Carlo simulation.

Questions or comments about the site? Contact the ANS Webmaster.