Benchmarking and Theory-Guided Allocation of Monte Carlo Particle Histories in Neutron Transport with Thermal-Hydraulic Feedback

Rapid development of Monte Carlo (MC) transport methods and high-performance computing has enabled coupled MC neutronics simulations with thermal-hydraulics (TH) coupling simulations to be widely researched and applied in reactor designs and analyses. Various coupling strategies have been developed to enhance the convergence stability and efficiency of the coupling iterations. However, comparisons of MC neutron transport/TH coupling methods are difficult to interpret when particle budgets, parameter grids, reference solutions, and error metrics differ. In this work, a controlled benchmark was established, and an optimization problem was designed to reveal both efficiency and lower-tail robustness of the methods. Several existing coupling strategies, including the traditional relaxation method with fixed coefficients, the Robbins-Monro method, and the asymptotic batch coupling method, are tested on the benchmark. In addition, a new family of particle history allocation strategies is derived from the general relaxation framework and compared with the existing methods. The results show that the new growing-population strategies can achieve comparable or better efficiency than existing methods while maintaining better robustness and insensitivity to the method coefficients. Further studies and optimizations in the field of neutrons and TH coupling can be explored based on this strategy and benefit from the established benchmark and the optimization problem.