Improvement of mode selection criterion of dynamic mode decomposition in a hydrofoil cavitation multiphase flow case

Abstract

With the improvement of cavitation prediction, how to extract key information from massive data of cavitation flow field accurately and quickly has become an urgent requirement. Reduced-order modeling method has been used to extract the key information and simplify the data of restored flow field. However, its applicability to two-phase flow field and how to select main modes have not been well-studied. Therefore, this study uses dynamic mode decomposition (DMD) method to analyze cavitation flow field of a two-dimensional hydrofoil with three cavitation numbers by mode decomposition and reconstruction, and to analyze the applicability of DMD method and two tradition main mode selection criteria for cavitation. The results show that DMD method is effective for the inversion of two-phase flow field, but due to the complexity of unstable cavitating flow, the applicable range of different criteria has been significantly reduced. Therefore, this study combines the advantages of clustering method, presents a new main mode selection criterion based on clustering analysis, which plays an important role in cavitation flow analysis. This method provides a strong guarantee for the use of reduced-order modeling method, and also provides a good basis for the introduction of artificial intelligence in the future.