Abstract:
Accurate analysis of reactor core behavior to unlock traditional design margins and enhance the economic efficiency of nuclear power plants has become a key research direction in current reactor design and development. Based on advanced multi-physics coupling computational analysis techniques and extensive research experience, this paper develops the MORE multi-physics coupling framework for reactors. MORE enables visual construction of various coupling modes, supports high-precision and high-efficiency mapping between ultra-large-scale multi-type grids, and provides more stable and efficient nonlinear iterative algorithms for strongly coupled complex transient calculations, addressing critical challenges in coupling program development such as complex grid mapping, data transmission, and process control. Using the MORE multi-physics coupling framework, a core physics-thermal hydraulic coupling calculation program was developed. In terms of grid mapping, efficiency is enhanced through multiple technologies including mapping algorithm optimization, stretched grids, and MPI-OpenMP hybrid parallelization. Three-dimensional stretched grids significantly improve mapping efficiency, reducing mapping time by two orders of magnitude. For billion-scale unstructured grids to billion-scale Cartesian grids, the thousand-core mapping time is approximately 70 seconds, with a thousand-core mapping efficiency exceeding 50%. In iterative acceleration, through in-depth research on methods such as JFNK and their key technologies—including core modules like the inexact Newton iteration module, Krylov subspace computation module, Jacobian-Free technology module, global convergence function module, forcing term selection module, and perturbation selection module—a multi-physics coupling iteration algorithm library (MORE-ALGO) was developed, covering mainstream explicit, semi-implicit, and fully implicit coupling methods at home and abroad. A JFNK hybrid iteration method using Picard iteration as the initial value was proposed to accelerate coupling calculations. The coupling program was validated against benchmark problems such as the NEACRP and IAEA rod ejection accidents. Comparisons of parameters including
keff, critical boron concentration, steady-state power peaking factor, power peak time, and transient relative power peak demonstrate the accuracy of the calculation results. Tests on the NEACRP rod ejection benchmark problem show that the JFNK method improves the convergence rate by 4-9 times compared to the Picard method, demonstrating superior stability and computational efficiency. Finally, preliminary engineering application analysis was conducted for the HPR1000 reactor. Results show that the power peak in a 100% power rod ejection accident at the beginning of life is reduced by 8.8% compared to traditional analysis methods, providing important support for further exploring safety margins in reactivity insertion accident analysis and evaluation.