Nonlinear frequency-domain reduced-order modelling of turbulent flows
Nonlinear frequency-domain reduced-order modelling of turbulent flows
We present a frequency-domain nonlinear reducedorder modelling technique to find pproximate periodic orbits (APOs) in turbulent flow problems and predict physical quantities and dynamical features. Although finding Unstable Periodic Orbits (UPOs) is feasible in low-dimensional dynamic systems or low-Re turbulent flows, computing costs become expensive for high-Re turbulent flows. Rather than finding UPOs in full-state space, we propose to build a ReducedOrder Model (ROM) to circumvent such impediment while the benefit of periodicity is maintained by considering the ROM in the frequency domain, which naturally accommodates time-periodic velocity fields. To this end, the space-time basis functions of the low-order space are extracted using Spectral Proper Orthogonal Decomposition (SPOD). The NavierStokes equations are converted into a low-order algebraic system via Galerkin projection onto selected SPOD modes. Numerical solutions of the ROM are achieved using gradientbased optimization of the amplitude coefficients, letting the solutions in the ROM satisfy conservation laws. The proposed approach is demonstrated in chaotic flows in a 2D liddriven cavity at Re=20,000. Numerical results show that the frequency-domain ROM admits multiple solutions that capture dominant dynamical flow features, such as vorticity structures, and predicts well statistical quantities, such as mean turbulent kinetic energy.
reduced-order model (ROM), spectral proper orthogonal decomposition, Turbulent flows (TF)
Li, Xiaodong
a471b477-f562-422e-94be-706100614bad
Lasagna, Davide
0340a87f-f323-40fb-be9f-6de101486b24
June 2024
Li, Xiaodong
a471b477-f562-422e-94be-706100614bad
Lasagna, Davide
0340a87f-f323-40fb-be9f-6de101486b24
Li, Xiaodong and Lasagna, Davide
(2024)
Nonlinear frequency-domain reduced-order modelling of turbulent flows.
13th International Symposium on Turbulence and Shear Flow Phenomena (TSFP13), , Montreal, Canada.
25 - 28 Jun 2024.
6 pp
.
Record type:
Conference or Workshop Item
(Paper)
Abstract
We present a frequency-domain nonlinear reducedorder modelling technique to find pproximate periodic orbits (APOs) in turbulent flow problems and predict physical quantities and dynamical features. Although finding Unstable Periodic Orbits (UPOs) is feasible in low-dimensional dynamic systems or low-Re turbulent flows, computing costs become expensive for high-Re turbulent flows. Rather than finding UPOs in full-state space, we propose to build a ReducedOrder Model (ROM) to circumvent such impediment while the benefit of periodicity is maintained by considering the ROM in the frequency domain, which naturally accommodates time-periodic velocity fields. To this end, the space-time basis functions of the low-order space are extracted using Spectral Proper Orthogonal Decomposition (SPOD). The NavierStokes equations are converted into a low-order algebraic system via Galerkin projection onto selected SPOD modes. Numerical solutions of the ROM are achieved using gradientbased optimization of the amplitude coefficients, letting the solutions in the ROM satisfy conservation laws. The proposed approach is demonstrated in chaotic flows in a 2D liddriven cavity at Re=20,000. Numerical results show that the frequency-domain ROM admits multiple solutions that capture dominant dynamical flow features, such as vorticity structures, and predicts well statistical quantities, such as mean turbulent kinetic energy.
This record has no associated files available for download.
More information
Published date: June 2024
Venue - Dates:
13th International Symposium on Turbulence and Shear Flow Phenomena (TSFP13), , Montreal, Canada, 2024-06-25 - 2024-06-28
Keywords:
reduced-order model (ROM), spectral proper orthogonal decomposition, Turbulent flows (TF)
Identifiers
Local EPrints ID: 497143
URI: http://eprints.soton.ac.uk/id/eprint/497143
PURE UUID: 304bb574-2513-4d4a-8990-c478292a7dfa
Catalogue record
Date deposited: 14 Jan 2025 18:05
Last modified: 16 Jan 2025 03:11
Export record
Contributors
Author:
Xiaodong Li
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics