TY - CONF

T1 - Joint scalar probability density function modeling for multiscalar turbulent mixing

AU - Perry, Bruce A.

AU - Mueller, Michael E.

N1 - Funding Information:
The authors gratefully acknowledge valuable support in the form of computational time on the TIGRESS high performance computing center at Princeton University, which is jointly supported by the Princeton Institute for Computational Science and Engineering and the Princeton University Office of Information Technology's Research Computing Department. B.A.P. is thankful to be supported by the NSF Graduate Research Fellowship Program under Grant No. DGE1148900.
Funding Information:
The authors gratefully acknowledge valuable support in the form of computational time on the TIGRESS high performance computing center at Princeton University, which is jointly supported by the Princeton Institute for Computational Science and Engineering and the Princeton University Office of Information Technology’s Research Computing Department. B.A.P. is thankful to be supported by the NSF Graduate Research Fellowship Program under Grant No. DGE1148900.

PY - 2017

Y1 - 2017

N2 - Large eddy simulation of turbulent reacting flows requires closure of the filtered chemical source term, which can be achieved using a presumed form of the joint subfilter probability density function (PDF) of the controlling scalars. In the flamelet/progress variable approach, the controlling scalars are the mixture fraction and progress variable, and a beta distribution is most often used to model the marginal subfilter PDF of the mixture fraction. Recent advances on this class of model allow for multiple inlets by defining additional mixture fractions and therefore require a model for the joint subfilter PDF of these mixture fractions. Several models have been proposed, including the statistically-most-likely distribution, the Dirichlet distribution, and a more general bivariate form of the b-distribution. In this work, direct numerical simulations of multiscalar mixing in isotropic homogeneous turbulence are used to provide a physical basis for selecting an appropriate subfilter PDF model. This is further supported mathematically by applying the concept of statistical neutrality for the proposed distributions.

AB - Large eddy simulation of turbulent reacting flows requires closure of the filtered chemical source term, which can be achieved using a presumed form of the joint subfilter probability density function (PDF) of the controlling scalars. In the flamelet/progress variable approach, the controlling scalars are the mixture fraction and progress variable, and a beta distribution is most often used to model the marginal subfilter PDF of the mixture fraction. Recent advances on this class of model allow for multiple inlets by defining additional mixture fractions and therefore require a model for the joint subfilter PDF of these mixture fractions. Several models have been proposed, including the statistically-most-likely distribution, the Dirichlet distribution, and a more general bivariate form of the b-distribution. In this work, direct numerical simulations of multiscalar mixing in isotropic homogeneous turbulence are used to provide a physical basis for selecting an appropriate subfilter PDF model. This is further supported mathematically by applying the concept of statistical neutrality for the proposed distributions.

KW - Large eddy simulation

KW - Subfilter PDF

KW - Turbulent mixing

UR - http://www.scopus.com/inward/record.url?scp=85048955069&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85048955069&partnerID=8YFLogxK

M3 - Paper

AN - SCOPUS:85048955069

T2 - 10th U.S. National Combustion Meeting

Y2 - 23 April 2017 through 26 April 2017

ER -