Producing a complete nuclear data library for adjoint Monte Carlo simulations

Abstract

Radiation shielding applications related to reactor design typically involve situations where the source region (the core) is much larger than the detector region (a dosimeter). In such cases, the efficiency of Monte Carlo simulation might be significantly increased by solving the adjoint transport equation: adjoint particles are born from the detector, undergo adjoint ('reversed') flights and collisions, and accumulate their tallies in the source region. A key prerequisite to sample the adjoint collision events is the preparation of adjoint nuclear data. In this work, we propose a general method able to handle the diversity of nuclear reactions available in modern neutron data evaluations, and ultimately create a full adjoint nuclear data library. This is a stepping stone in view of implementing adjoint sampling schemes in TRIPOLI-5 __ , the next-generation Monte Carlo code developed by CEA and ASNR. We validate this strategy based on a relevant continuous energy benchmark configuration involving mixtures of heavy and light nuclides, and we compare our results to those obtained by standard forward Monte Carlo simulations.