Litcius/Paper detail

Koopman reduced-order modeling and analysis of flag flapping in the wake of a cylinder

Haokui Jiang, Jean-Lou Pfister, Daniel Zhengyu Huang, Shunxiang Cao

2025Physical review. E13 citationsDOI

Abstract

We develop a Koopman reduced-order model (ROM) to analyze the instability mechanism and predict the hydrodynamic behavior for the flag flapping in the wake of a cylinder. The Koopman ROM is constructed using a kernel dynamic mode decomposition method and enhanced through a residual dynamical mode decomposition algorithm, which improves accuracy by identifying and eliminating spurious modes. Our analysis reveals a flow transition from the "2S" mode in the periodic phase to the "2P" mode in the quasiperiodic phase, with the main Koopman mode M_{1} providing insights into the instability mechanism. In the case of chaotic flapping at a Reynolds number of Re=1200, the Koopman ROM demonstrates high accuracy in predicting the chaotic fluid-structure interaction flow when comparing the fractal dimension d_{c} and the maximum Lyapunov exponent λ_{max} of the true and reconstructed flow. Additionally, we observe similar flag flapping and vortex shedding characteristics in the near-structure region throughout the investigated Reynolds number range Re∈[500,1200], leading to similar vorticity patterns for M_{1}. The flag flapping has a local minimum displacement at position x_{0}≈3.0 in the flag's displacement envelope. Notably, this position closely matches the critical position x_{c} obtained from global linear instability analysis at low Reynolds numbers. This conclusion aligns with the performance of the Koopman ROM using sparse measurement probes attached to the flag. The position of the probes influences the accuracy of the ROM, where higher accuracy corresponds to a larger displacement of the eigenfunction.

Topics & Concepts

WakeFlappingFlag (linear algebra)Order (exchange)CylinderMechanicsMathematicsEngineeringPhysicsAerospace engineeringGeometryEconomicsAlgebra over a fieldWingPure mathematicsFinanceModel Reduction and Neural NetworksFluid Dynamics and Vibration AnalysisFluid Dynamics and Turbulent Flows