Application of an efficient SPOD method to the flow characteristics observation of 2 × 2 fuel rod bundles

A fast spectral proper orthogonal decomposition (F-SPOD) method is proposed for the efficient analysis of large-scale flow datasets and is applied to investigate unsteady flow structures around an exposed 2 × 2 nuclear fuel rod bundle. Spectral proper orthogonal decomposition (SPOD) is widely used to extract coherent structures across different frequencies, particularly in periodic flow phenomena such as vortex shedding and coolant-induced instabilities. However, classical SPOD becomes computationally expensive in three-dimensional cases due to the high spatial resolution required. To overcome this limitation, F-SPOD first applies singular value decomposition (SVD) to the data matrix to extract orthogonal spatial modes and corresponding temporal coefficients. Spectral analysis is then performed in the reduced temporal space, significantly improving computational efficiency and memory usage. Validation results confirm that F-SPOD retains excellent accuracy compared to classical SPOD while dramatically reducing computational cost. When applied to the fuel rod case, F-SPOD reveals that high-frequency pressure fluctuations dominate near the inlet and evolve into oblique low-frequency waves downstream. Tangential velocity fluctuations are concentrated near wall surfaces and exhibit a rotational pattern along the flow direction. Additionally, the analysis highlights the importance of monitoring point density: insufficient spatial sampling can lead to oversmoothed modal energy spectra and the omission of critical frequency components, thereby reducing the reliability of modal interpretations. The adoption of the F-SPOD algorithm is thus crucial for improving computational efficiency in such analyses.