The method of discrete ordinates is investigated for neutron noise simulations in the frequency domain. For this purpose, the solver NOISE-SN is developed and used to simulate two neutron noise problems that are respectively derived from the two-dimensional systems described in the neutron transport simulation benchmarks C4V and C5G7. In the first problem based on the C4V system, NOISE-SN is compared to the diffusion-based simulator CORE SIM+. These results show that NOISE-SN and CORE SIM+ calculate similar spatial distributions of neutron noise, although significant differences can be found at the location of the perturbation and at locations with strong variations of material properties, where the discrete ordinates method is expected to be more accurate than diffusion theory. Then NOISE-SN calculations are performed to test different $S_N$ approximations, and the fictitious source method that may be applied to mitigate possible numerical artifacts, known as the ray effect. In the second problem based on the C5G7 system, the choice of a low order of discrete ordinates in NOISE-SN leads to unphysical values of the neutron noise because of the ray effect. The increase of the order of discrete ordinates or introducing a fictitious source in the equations to be solved alleviates the issue. The second option is shown to remove the ray effect without a high order of discrete ordinates and thus without too expensive calculations, even though the strength of the fictitious source needs to be tuned carefully to avoid very slow convergence rates.

研究了离散纵坐标法在频域中进行中子噪声模拟。为此，开发了求解器 NOISE-SN 并用于模拟两个中子噪声问题，这些问题分别源自中子输运模拟基准 C4V 和 C5G7 中描述的二维系统。在基于C4V系统的第一个问题中，将NOISE-SN与基于扩散的模拟器CORE SIM+进行比较。这些结果表明，NOISE-SN 和 CORE SIM+ 计算出相似的中子噪声空间分布，尽管在扰动位置和材料特性变化很大的位置可以发现显着差异，其中离散纵坐标方法预计会更多比扩散理论准确。然后进行 NOISE-SN 计算以测试不同的${秒}_N$近似值，以及可用于减轻可能的数值伪影（称为射线效应）的虚构源方法。在基于 C5G7 系统的第二个问题中，由于射线效应，在 NOISE-SN 中选择低阶离散纵坐标会导致中子噪声的非物理值。增加离散坐标的阶数或在待求解的方程中引入一个虚构的源可以缓解这个问题。第二个选项显示在没有高阶离散坐标的情况下消除光线效应，因此不需要太昂贵的计算，即使需要仔细调整虚拟源的强度以避免收敛速度非常慢。