Data driven feature identification and sparse representation of turbulent flows

File Description SizeFormat 
TSFP2019.pdfAccepted version2.17 MBAdobe PDFView/Open
Title: Data driven feature identification and sparse representation of turbulent flows
Authors: Beit-Sadi, M
Krol, J
Wynn, A
Item Type: Conference Paper
Abstract: dentifying coherent structures of fluid flows is of greatimportance for reduced order modelling and flow control.Finding such structures in a turbulent flow, however, canbe challenging. A number of modal decomposition algo-rithms have been proposed in recent years which decom-pose snapshots of data into spatial modes, each associatedwith a single frequency and growth-rate, most prominentlydynamic mode decomposition (DMD). However, the num-ber of modes that DMD-like algorithms construct may beunrelated to the number of significant degrees of freedomof the underlying system. This provides a difficulty if onewants to create a low-order model of a flow. In this work,we present a method of post-processing DMD modes forextracting a small number of dynamically relevant modes.This is achieved by first ranking the DMD modes, then us-ing an iterative approach based on the graph-theoretic no-tion of maximal cliques to identify clusters of modes and,finally, by replacing each cluster with a single (pair of)modes.
Issue Date: 30-Jul-2019
Date of Acceptance: 20-Apr-2019
URI: http://hdl.handle.net/10044/1/73152
Journal / Book Title: Proceedings of TSFP-11 (2019)
Copyright Statement: © 2019 The Author(s)
Sponsor/Funder: Engineering & Physical Science Research Council (EPSRC)
Engineering & Physical Science Research Council (EPSRC)
Funder's Grant Number: EP/K503381/1
EP/N015398/1
Conference Name: Eleventh International Symposium on Turbulence and Shear Flow Phenomena (TSFP11)
Start Date: 2019-07-30
Finish Date: 2019-08-02
Conference Place: Southampton, UK
Appears in Collections:Aeronautics



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

Creative Commons