Details

CORRELATIVE STUDIES AND COHERENT STRUCTURES EDUCTION BASED ON PROPER ORTHOGONAL DECOMPOSITION AND LINEAR STOCHASTIC ESTIMATION

by VERFAILLIE, SWANN

Abstract (Summary)
The thrust of this thesis is to introduce and test two Coherent Structures eduction methods utilizing two different databases. The first set of data was collected in the turbulent flow produced by a circular jet that is known to develop high shear layers where large scale coherent structures are usually embedded. The second experiment was conducted on the swirling flow exiting a Triple Annular Research Swirler (TARS). The TARS is used as a fuel injector in a gas turbine combustor simulator and it features three concentric swirling flows that merge into a complex swirling jet. The methods that are exposed in this thesis are the Proper Orthogonal Decomposition (POD) and the Linear Stochastic Estimation (LSE). These techniques that have become more and more famous over the past decades are based on filtering techniques and are built on a solid mathematical background: the POD extracts the most energetic structures from a turbulent flow and essentially relies on projections and linear algebra theory whereas the LSE uses statistical theory to provide an estimate of the instantaneous flow field by reproducing, in space and time, the large-scale coherent structures of the flow using minimum experimental information. The mathematical theory that is needed to understand these techniques has been exposed in a first part and then both POD and LSE were tested and compared for the two databases. Both filtering techniques turned out to be very efficient for extracting Coherent Structures from the turbulent flow fields and for apprehending their dynamic. Although the LSE highly depends on the levels of correlation and consequently raised several issues for the experimental setups, it eventually allowed us reconstructing complete coherent flow fields from the knowledge of only few near-field acoustic signals and provided a tool as powerful in its concept as it is in practice in terms of data reduction.
Bibliographical Information:

Advisor:

School:University of Cincinnati

School Location:USA - Ohio

Source Type:Master's Thesis

Keywords:coherent structures linear stochastice estimation proper orthogonal decomposition turbulent flow swirling coherence correlation

ISBN:

Date of Publication:01/01/2004

© 2009 OpenThesis.org. All Rights Reserved.