Session: 08-03-01 Fundamental Vortex Induced Vibrations & Cylinder Hydrodynamics

Submission Number: 181504

Wake Dynamics via Dynamic Mode Decomposition and Lift Oscillation in Cylinder Flow at Moderate Reynolds Numbers 

The flow past a circular cylinder is a classical problem in fluid dynamics, extensively studied due to its fundamental vortex dynamics and its relevance to engineering applications such as offshore structures. Understanding the evolution of flow structures across different Reynolds number (Re) regimes is crucial both for practical applications and for advancing the fundamental knowledge of flow stability, wake dynamics and fluid-structure interaction.

A key feature of this flow is the unsteady alternate shedding of vortices, which induces an oscillatory lift force on the cylinder, quantified by the root-mean-square (RMS) of the lift coefficient (C_L,rms). Despite its simple geometry, this configuration exhibits complex behavior in the variation of C_L,rms, with different regimes depending on the Reynolds number. Within the moderate subcritical regime (300 < Re < 10000), a well-documented phenomenon occurs: after an initial increase, the lift oscillation drops, reaching a minimum around Re = 1000 (sometimes referred to as the "first lift crisis"), before gradually increasing again beyond Re = 4000. This phenomenon is our object of study.

In this work, we employ computational fluid dynamics (CFD) simulations using the NEKTAR++ solver to investigate the dynamics in the range 300 < Re < 10000. To accurately capture the C_L,rms curve across this entire range, a hybrid approach was utilized. Direct Numerical Simulations (DNS) were performed in a 2.5D configuration to resolve the transitional wake structures for Re up to 1000. For Reynolds numbers beyond this range, where the computational cost of full DNS is prohibitive do to the fine mesh requirements, we incorporate a vanishing viscosity turbulence model. The combined simulations match the experimental C_L,rms curve reported in the literature.

To investigate the physical mechanisms underlying this transition and the drop in C_L,rms, we perform a detailed analysis focusing on three representative Reynolds numbers: Re = 300, Re = 1000, and Re = 10000, using high-resolution 2.5D DNS. Although all cases feature vortex shedding, they display fundamentally different wake dynamics and force characteristics. Flow structures were identified using instantaneous and time-averaged fields of vorticity and the Q-criterion. Furthermore, to extract coherent spatiotemporal patterns and identify the dominant dynamical modes governing the transition, we applied Proper Orthogonal Decomposition (POD) and Dynamic Mode Decomposition (DMD) to the flow fields.

We expect our results to further the understanding of this flow which is present in many areas, allowing for improvements in its modeling, such as by calibrating simpler and computationally cheaper models by using these resource-intensive simulations as a reference.

Presenting Author: Marlon Mathias Universidade de S&#227;o Paulo

Presenting Author Biography: Professor at the Mechanical Engineering Department of Escola Politécnica at the University of São Paulo. Works in Physics-Informed Machine Learning and high-fidelity fluid simulation. Master and Doctor in Mechanical Engineering from the São Carlos School of Engineering, USP (EESC-USP), having worked in the hydrodynamic instability that leads to the transition to turbulence. Aeronautical Engineer also from EESC-USP.

Authors:

Marlon Mathias Universidade de São Paulo
Enrique Takagi Guimarães Universidade de São Paulo
Eric Lodi Gomes Universidade de São Paulo
Vitor Bortolin Universidade de São Paulo
Bernardo Luiz Harry Diniz Lemos Universidade de São Paulo
Rodrigo Amaral Universidade de São Paulo
Alessando Alberto De Lima Universidade de São Paulo
Julio Meneghini Universidade de São Paulo