Direct and large-eddy simulation of interactions
between reacting flows and evaporating droplets
Direct and large-eddy simulation of interactions
between reacting flows and evaporating droplets
Reacting flows interacting with liquid droplets are of practical and scientific importance due
to their appearance in a multiplicity of industrial and domestic applications such as fire
suppression systems and humidified gas turbines. Experimental measurements are usually
limited to global quantities or local quantities at limited spatial locations, which provide
little detailed information for fundamental understanding of complex interactions.
Numerical simulations can overcome these limitations, but are restricted by the available
computer capacity. Therefore, most previous simulations in the field employ simplifications
such as a two-dimensional configuration, the Eulerian description of the dispersed phase, or
the Reynolds averaged Navier-Stokes (RANS) methodology, etc. Such simplified methods
are not appropriate for scrutinizing the local, unsteady interactions embedded in the
realistic multiphase reacting flows from the first principle. To this end, systematic
understanding of the multilateral, multiscale and multiphysics interactions among
turbulent flow, chemical reaction and dispersing droplets is still far from being achieved.
Recently, the rapid development of the supercomputer hardware and software
technologies enables the application of high fidelity numerical techniques, i.e., direct
numerical simulation (DNS) and large eddy simulation (LES), to such complex flows. In
the present study, a hybrid Eulerian-Lagrangian methodology is developed and
implemented to investigate the multiphysics interactions. The well-designed parallel
algorithms enable us to look at both canonical and practical configurations, including a
temporal reacting shear layer, a turbulent reactive jet diluted with droplets and a simplified
small-scale domestic fire suppression system. All these configurations are characterised by
nonuniform droplet loading in the computational domain, fully three-dimensional
simulation and the Lagrangian description for droplets. The number of traced droplets
reaches the magnitude of 106 in some cases. DNS results with various physical parameters
have been obtained, showing self-consistency and correct trends. In LES, to avoid arbitrary
tuning of subgrid model coefficients, fully dynamic procedures have been designed
following the Germano procedure and implemented for the main six subgrid models. LES
of a heated plane jet, a reacting jet diluted with evaporating droplets and a simplified fire
suppression system has been performed and analysed. Droplet effects on turbulence and
combustion are quantified through examining the transport equations for the kinetic
energy and internal energy of the reacting flow.
Xia, Jun
91c183b9-2ad3-40fd-ae51-3ee5851fb1e8
February 2008
Xia, Jun
91c183b9-2ad3-40fd-ae51-3ee5851fb1e8
Luo, Kai H.
86f52a13-fdcd-40e4-8344-a6fe47c4e16b
Kumar, Suresh
71833d28-23b5-4e50-b301-55561e214fb8
Xia, Jun
(2008)
Direct and large-eddy simulation of interactions
between reacting flows and evaporating droplets.
University of Southampton, School of Engineering Sciences, Doctoral Thesis, 183pp.
Record type:
Thesis
(Doctoral)
Abstract
Reacting flows interacting with liquid droplets are of practical and scientific importance due
to their appearance in a multiplicity of industrial and domestic applications such as fire
suppression systems and humidified gas turbines. Experimental measurements are usually
limited to global quantities or local quantities at limited spatial locations, which provide
little detailed information for fundamental understanding of complex interactions.
Numerical simulations can overcome these limitations, but are restricted by the available
computer capacity. Therefore, most previous simulations in the field employ simplifications
such as a two-dimensional configuration, the Eulerian description of the dispersed phase, or
the Reynolds averaged Navier-Stokes (RANS) methodology, etc. Such simplified methods
are not appropriate for scrutinizing the local, unsteady interactions embedded in the
realistic multiphase reacting flows from the first principle. To this end, systematic
understanding of the multilateral, multiscale and multiphysics interactions among
turbulent flow, chemical reaction and dispersing droplets is still far from being achieved.
Recently, the rapid development of the supercomputer hardware and software
technologies enables the application of high fidelity numerical techniques, i.e., direct
numerical simulation (DNS) and large eddy simulation (LES), to such complex flows. In
the present study, a hybrid Eulerian-Lagrangian methodology is developed and
implemented to investigate the multiphysics interactions. The well-designed parallel
algorithms enable us to look at both canonical and practical configurations, including a
temporal reacting shear layer, a turbulent reactive jet diluted with droplets and a simplified
small-scale domestic fire suppression system. All these configurations are characterised by
nonuniform droplet loading in the computational domain, fully three-dimensional
simulation and the Lagrangian description for droplets. The number of traced droplets
reaches the magnitude of 106 in some cases. DNS results with various physical parameters
have been obtained, showing self-consistency and correct trends. In LES, to avoid arbitrary
tuning of subgrid model coefficients, fully dynamic procedures have been designed
following the Germano procedure and implemented for the main six subgrid models. LES
of a heated plane jet, a reacting jet diluted with evaporating droplets and a simplified fire
suppression system has been performed and analysed. Droplet effects on turbulence and
combustion are quantified through examining the transport equations for the kinetic
energy and internal energy of the reacting flow.
Text
THESIS_JX.pdf
- Other
More information
Published date: February 2008
Organisations:
University of Southampton, Astronautics Group
Identifiers
Local EPrints ID: 65719
URI: http://eprints.soton.ac.uk/id/eprint/65719
PURE UUID: a0183f9e-0516-4856-bc8f-d4c201a16311
Catalogue record
Date deposited: 17 Mar 2009
Last modified: 13 Mar 2024 17:49
Export record
Contributors
Author:
Jun Xia
Thesis advisor:
Kai H. Luo
Thesis advisor:
Suresh Kumar
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