Experimental assessment of presumed filtered density function models

V. Stetsyuk, N. Soulopoulos, Y. Hardalupas, A. M. K. P. Taylor

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

Measured filtered density functions (FDFs) as well as assumed beta distribution model of mixture fraction and “subgrid” scale (SGS) scalar variance z′′2⎯⎯⎯⎯⎯⎯⎯z′′2¯, used typically in large eddy simulations, were studied by analysing experimental data, obtained from two-dimensional planar, laser induced fluorescence measurements in isothermal swirling turbulent flows at a constant Reynolds number of 29 000 for different swirl numbers (0.3, 0.58, and 1.07). Two-dimensional spatial filtering, by using a box filter, was performed in order to obtain the filtered variables, namely, resolved mean and “subgrid” scale scalar variance. These were used as inputs for assumed beta distribution of mixture fraction and top-hat FDF shape estimates. The presumed beta distribution model, top-hat FDF, and the measured filtered density functions were used to integrate a laminar flamelet solution in order to calculate the corresponding resolved temperature. The experimentally measured FDFs varied with the flow swirl number and both axial and radial positions in the flow. The FDFs were unimodal at flow regions with low SGS scalar variance, z′′2⎯⎯⎯⎯⎯⎯⎯<z′′2¯< 0.01, and bimodal at regions with high SGS variance, z′′2⎯⎯⎯⎯⎯⎯⎯>z′′2¯> 0.02. Bimodal FDF could be observed for a filter size of approximately 1.5-2 times the Batchelor scale. Unimodal FDF could be observed for a filter size as large as four times the Batchelor scale under well-mixed conditions. In addition, two common computational models (a gradient assumption and a scale similarity model) for the SGS scalar variance were used with the aim to evaluate their validity through comparison with the experimental data. It was found that the gradient assumption model performed generally better than the scale similarity one.
Original languageEnglish
Article number065107
Number of pages19
JournalPhysics of Fluids
Volume27
Issue number6
Early online date10 Jun 2015
DOIs
Publication statusPublished - Jun 2015
Externally publishedYes

Fingerprint

Probability density function
Large eddy simulation
Turbulent flow
Reynolds number
Fluorescence
Lasers

Cite this

Stetsyuk, V. ; Soulopoulos, N. ; Hardalupas, Y. ; Taylor, A. M. K. P. / Experimental assessment of presumed filtered density function models. In: Physics of Fluids. 2015 ; Vol. 27, No. 6.
@article{7db191ba3ddd4d169aadb93457e452b5,
title = "Experimental assessment of presumed filtered density function models",
abstract = "Measured filtered density functions (FDFs) as well as assumed beta distribution model of mixture fraction and “subgrid” scale (SGS) scalar variance z′′2⎯⎯⎯⎯⎯⎯⎯z′′2¯, used typically in large eddy simulations, were studied by analysing experimental data, obtained from two-dimensional planar, laser induced fluorescence measurements in isothermal swirling turbulent flows at a constant Reynolds number of 29 000 for different swirl numbers (0.3, 0.58, and 1.07). Two-dimensional spatial filtering, by using a box filter, was performed in order to obtain the filtered variables, namely, resolved mean and “subgrid” scale scalar variance. These were used as inputs for assumed beta distribution of mixture fraction and top-hat FDF shape estimates. The presumed beta distribution model, top-hat FDF, and the measured filtered density functions were used to integrate a laminar flamelet solution in order to calculate the corresponding resolved temperature. The experimentally measured FDFs varied with the flow swirl number and both axial and radial positions in the flow. The FDFs were unimodal at flow regions with low SGS scalar variance, z′′2⎯⎯⎯⎯⎯⎯⎯z′′2¯> 0.02. Bimodal FDF could be observed for a filter size of approximately 1.5-2 times the Batchelor scale. Unimodal FDF could be observed for a filter size as large as four times the Batchelor scale under well-mixed conditions. In addition, two common computational models (a gradient assumption and a scale similarity model) for the SGS scalar variance were used with the aim to evaluate their validity through comparison with the experimental data. It was found that the gradient assumption model performed generally better than the scale similarity one.",
keywords = "Large Eddy Simulation, spatial filtering, rotating flows, spatial scaling, density measurement",
author = "V. Stetsyuk and N. Soulopoulos and Y. Hardalupas and Taylor, {A. M. K. P.}",
year = "2015",
month = "6",
doi = "10.1063/1.4922169",
language = "English",
volume = "27",
journal = "Physics of Fluids",
issn = "1070-6631",
publisher = "American Institute of Physics Publising LLC",
number = "6",

}

Experimental assessment of presumed filtered density function models. / Stetsyuk, V.; Soulopoulos, N.; Hardalupas, Y.; Taylor, A. M. K. P.

In: Physics of Fluids, Vol. 27, No. 6, 065107, 06.2015.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Experimental assessment of presumed filtered density function models

AU - Stetsyuk, V.

AU - Soulopoulos, N.

AU - Hardalupas, Y.

AU - Taylor, A. M. K. P.

PY - 2015/6

Y1 - 2015/6

N2 - Measured filtered density functions (FDFs) as well as assumed beta distribution model of mixture fraction and “subgrid” scale (SGS) scalar variance z′′2⎯⎯⎯⎯⎯⎯⎯z′′2¯, used typically in large eddy simulations, were studied by analysing experimental data, obtained from two-dimensional planar, laser induced fluorescence measurements in isothermal swirling turbulent flows at a constant Reynolds number of 29 000 for different swirl numbers (0.3, 0.58, and 1.07). Two-dimensional spatial filtering, by using a box filter, was performed in order to obtain the filtered variables, namely, resolved mean and “subgrid” scale scalar variance. These were used as inputs for assumed beta distribution of mixture fraction and top-hat FDF shape estimates. The presumed beta distribution model, top-hat FDF, and the measured filtered density functions were used to integrate a laminar flamelet solution in order to calculate the corresponding resolved temperature. The experimentally measured FDFs varied with the flow swirl number and both axial and radial positions in the flow. The FDFs were unimodal at flow regions with low SGS scalar variance, z′′2⎯⎯⎯⎯⎯⎯⎯z′′2¯> 0.02. Bimodal FDF could be observed for a filter size of approximately 1.5-2 times the Batchelor scale. Unimodal FDF could be observed for a filter size as large as four times the Batchelor scale under well-mixed conditions. In addition, two common computational models (a gradient assumption and a scale similarity model) for the SGS scalar variance were used with the aim to evaluate their validity through comparison with the experimental data. It was found that the gradient assumption model performed generally better than the scale similarity one.

AB - Measured filtered density functions (FDFs) as well as assumed beta distribution model of mixture fraction and “subgrid” scale (SGS) scalar variance z′′2⎯⎯⎯⎯⎯⎯⎯z′′2¯, used typically in large eddy simulations, were studied by analysing experimental data, obtained from two-dimensional planar, laser induced fluorescence measurements in isothermal swirling turbulent flows at a constant Reynolds number of 29 000 for different swirl numbers (0.3, 0.58, and 1.07). Two-dimensional spatial filtering, by using a box filter, was performed in order to obtain the filtered variables, namely, resolved mean and “subgrid” scale scalar variance. These were used as inputs for assumed beta distribution of mixture fraction and top-hat FDF shape estimates. The presumed beta distribution model, top-hat FDF, and the measured filtered density functions were used to integrate a laminar flamelet solution in order to calculate the corresponding resolved temperature. The experimentally measured FDFs varied with the flow swirl number and both axial and radial positions in the flow. The FDFs were unimodal at flow regions with low SGS scalar variance, z′′2⎯⎯⎯⎯⎯⎯⎯z′′2¯> 0.02. Bimodal FDF could be observed for a filter size of approximately 1.5-2 times the Batchelor scale. Unimodal FDF could be observed for a filter size as large as four times the Batchelor scale under well-mixed conditions. In addition, two common computational models (a gradient assumption and a scale similarity model) for the SGS scalar variance were used with the aim to evaluate their validity through comparison with the experimental data. It was found that the gradient assumption model performed generally better than the scale similarity one.

KW - Large Eddy Simulation

KW - spatial filtering

KW - rotating flows

KW - spatial scaling

KW - density measurement

UR - http://aip.scitation.org/journal/phf

U2 - 10.1063/1.4922169

DO - 10.1063/1.4922169

M3 - Article

VL - 27

JO - Physics of Fluids

JF - Physics of Fluids

SN - 1070-6631

IS - 6

M1 - 065107

ER -