Large eddy simulation of turbulent swirling flames

Large eddy simulation (LES) is attractive as it provides a reasonable compromise between accuracy and cost, and is rapidly evolving as a practical approach for many engineering applications. This thesis is concerned with the application of large eddy simulation to unconfined swirl in turbulent non-premixed flames and isothermal flows. The LES methodology has been applied for the prediction of turbulent swirling reacting and non-reacting flows based on laboratory scale swirl burner known as the Sydney swirl burner, which has been a target flame of the workshop series of turbulent non-premixed flames (TNF). For that purpose a LES code was developed that can run wide range of applications. An algorithm was developed for LES of variable density reacting flow calculations. Particular attention was given to primitive conservation (mass, momentum and scalar) and kinetic energy of the flow and mixing field. The algorithm uses the primitive variables, which are staggered in both space and time. A steady laminar flamelet model which includes the detailed chemical kinetics and multi component mass diffusion, has been implemented in the LES code. An artificial inlet boundary condition method was implemented to generate instantaneous turbulent velocity fields that are imposed on the inflow boundary of the Cartesian grid. To improve the applicability of the code, various approaches were developed to improve stability and efficiency. LES calculations for isothermal turbulent swirling jets were successful in predicting experimentally measured mean velocities, their rms fluctuations and Reynolds shear stresses. The phenomenon of vortex breakdown (VB) and recirculation flow structures at different swirl and Reynolds numbers were successfully reproduced by the present large eddy simulations indicating that LES is capable of predicting VB phenomena which occurs only at certain conditions. For swirling flames, the LES predictions were able to capture the unsteady flow field, flame dynamics and showed good agreement with experimental measurements. The LES predictions for the mean temperature and major species were also successful