Study on Monte Carlo iterative K-source (IKS) method for frequency domain neutron noise calculation

• Efficiency Gains in Practical Regimes : Validated IKS method on a simplified UOX 2D fuel assembly and adapted IAEA 3D PWR benchmark. It raises FOM 2–10 times higher than the traditional method at plateau frequencies (e.g., 1 Hz), with balanced RSD in complex 3D geometries. • Robust Convergence Criteria : Proposes a novel convergence criterion combining Shannon entropy (real/imaginary components) and phase via stochastic oscillators, avoiding false convergence in complex-valued flux noise problems. • Clarified Limitations and Future Directions : Identifies IKS limitations at extreme frequencies (0.01/100 Hz): k s →1.0 causes slow convergence and k s fluctuation sensitivity; clarifies future optimizations (e.g., refined k s stabilization). The solution to the neutron noise equation is critical for on-line reactor monitoring and fault diagnosis via neutron noise analysis. This study applies the Monte Carlo Iterative K-Source (IKS) method to frequency domain neutron noise calculations, aiming to enhance computational efficiency and clarify its distinctions from the traditional fixed-source direct method. The theoretical derivation of the IKS method is performed and this method is implemented in the NECP-MCX code. The numerical validation is carried out based on a simplified UOX 2D fuel assembly and an adapted IAEA 3D PWR benchmark. The numerical results demonstrate that the IKS method achieves significant efficiency gains in the plateau frequency region, with a higher Figure of Merit (FOM) compared to the direct method—particularly in complex 3D geometries, where it exhibits more balanced relative standard deviation (RSD) distributions. The proposed convergence criteria, integrating Shannon entropy of real and imaginary parts and phase via stochastic oscillators, effectively judges iteration convergence, with phase emerging as a critical indicator for noise problems. However, the IKS method faces challenges at extreme frequencies where the k s factor approaches 1.0: convergence slows markedly, requiring large particle numbers per cycle to suppress k s fluctuations, and result accuracy becomes sensitive to k s instability due to the 1/(1- k s ) normalization. This work establishes the IKS method as a promising alternative for efficient neutron noise simulation, providing technical support for reactor monitoring and fault diagnosis applications.