A computational study of turbulent premixed flames including direct numerical and large -eddy simulations

Ohmsatya Prasad Ravi, Purdue University


Several issues in numerical modeling and simulations of turbulent premixed flames are addressed in this thesis. In the first part of this study, the reaction progress variable and flame surface density combustion models are extended for nonisenthalpic premixed flames. The performance of five state-of-the-art premixed combustion models is evaluated for two premixed jet flames using the boundary layer approximation. Overall, the proposed modification gives good predictions of mean temperature and axial velocity profiles for nonisenthalpic jet flames. ^ In the second part of this study, results from direct numerical simulations (DNS) are used to develop and assess filtered reaction rate models. A model based on the reaction progress variable approach, traditionally used in time-averaged calculations, is proposed as a subgrid combustion model for large eddy simulations (LES). Both a priori and a posteriori studies show that the new model gives excellent results. Four different subgrid turbulence models are also studied to evaluate their accuracy in predicting the transitional jet flames. ^ The reaction progress variable based subgrid combustion model is implemented in the commercial CFD code FLUENT/UNS. Using this code, LES of lean premixed combustion in an axisymmetric dump combustor are conducted. Cold flow simulations of this combustor give excellent predictions of mean and radial velocity compared to the experimental data. The reacting flow simulations also give good predictions of mean axial velocity and intensity of axial velocity fluctuations. The 3D LES has shown the large scale, asymmetric, flame front flapping about the axis of the combustor. ^




Major Professors: Jayavant P. Gore, Purdue University, Steven H. Frankel, Purdue University.

Subject Area

Engineering, Chemical|Engineering, Mechanical

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server