English

Body-Free Simulation of Three-Dimensional Turbulent Cylinder Wakes

Fluid Dynamics 2026-04-15 v2 Computational Physics

Abstract

We present a body-free simulation framework for three-dimensional turbulent cylinder wakes, in which the upstream cylinder is not explicitly resolved. Instead, the incompressible Navier--Stokes equations are solved in a simplified rectangular domain, and the inflow is prescribed using velocity profiles extracted from experimental measurements or pre-computed direct numerical simulations (DNS). We show that, for the Reynolds numbers considered here, prescribing low-dimensional inflow information at a single downstream location is sufficient to reconstruct the principal wake dynamics, including three-dimensionality, coherent vortex shedding, Reynolds-stress distributions, and spectral content, for Re=500Re=500, 5,0005{,}000, and 11,00011{,}000. Comparisons with full-body DNS and particle image velocimetry (PIV) measurements show good agreement in mean velocity profiles, Reynolds stresses, and dominant shedding frequencies. A local stability analysis further shows that successful wake reconstruction is obtained when the imposed inflow is selected from the absolutely unstable near-wake region. The results support existing theory that the essential dynamics of bluff-body wakes are governed primarily by the instability of the near-wake profile rather than by the explicit presence of the body itself. The role of the onset of three-dimensionality is elucidated for first time. Relative to full DNS with the cylinder present, the proposed body-free simulation offers a substantial reduction in computational cost, providing a physically interpretable and computationally efficient route for wake reconstruction, reduced-complexity simulation, and flow-control studies.

Keywords

Cite

@article{arxiv.2510.24922,
  title  = {Body-Free Simulation of Three-Dimensional Turbulent Cylinder Wakes},
  author = {Zhicheng Wang and Theo Käufer and Khemraj Shukla and Michael Triantafyllou and George Em Karniadakis},
  journal= {arXiv preprint arXiv:2510.24922},
  year   = {2026}
}