The goal of the present doctoral thesis was to develop a turbulence-chemistry interaction model for turbulent combustion simulations for high Reynolds number flows of practical interest. Present simulations were carried out using the unsteady Reynolds-averaged Navier-Stokes (URANS) and the Large Eddy Simulation (LES) formulations. The compressible flow treatment was used for both cases in order to model the coupling between hydromechanics and thermodynamics. The transient Navier-Stokes equations were applied to catch up unsteady combustion physics properly. The interaction between turbulence and chemistry was modeled based on the Eddy Dissipation Concept (EDC) with a detailed chemistry treatment. The open source OpenFOAM toolbox was used as the main frame for mathematical modeling and numerical simulations.
First, non-reacting turbulent bluff-body flows were investigated with the goal of validation, verification and understanding of the capabilities of the numerical method using the conventional URANS approach. These results were analyzed in detail and agreed fairly well with experimental data.
Then, the validation of the URANS approach (based on the standard k-ε model) was extended for reacting turbulent flows: the Sandia Flame D, the Sandia Flame CHNa and the Sydney Bluff-Body Flame HM1E. The chemistry was described by the full GRI-3.0 mechanism. There was relatively good agreement between simulations and measurements and it is believed that one of the main reasons for the observed discrepancies was the round-jet anomaly of the standard k-ε model.
Furthermore, the numerical method was extended to a large-eddy simulation model. A sub-critical circular cylinder flow at a Reynolds number Re = 3:9 103 and a Mach number M = 0:2 was simulated to evaluate the applicability of the implemented LES approach. In general, the LES results agreed fairly well with the available experimental and numerical data and gave an indication of the adequacy and the accuracy of the implemented method.
As a next step, LES validation was extended for a modest sub-critical circular cylinder flow at Re = 2x104. Both an incompressible and a compressible (M = 0:2) flow treatment was used. The predicted results revealed significant inaccuracies like spurious oscillations of the compressible flow solution. The incompressible flow results were found to be consistent with the existing LES studies as well as with measurements.
Testing of the non-reflecting boundary conditions was performed for the Aeolian tones aeroacoustic predictions. The laminar flow over a circular cylinder at Re = 140 and M = 0:2 was calculated by direct solution of the unsteady compressible Navier-Stokes equations. The sound generated by a circular cylinder at Re = 2:2 x 104 and M = 0:2 was predicted using LES. The calculated acoustic fields showed a dipole directivity, similar to a natural vortex shedding. The impact of the Doppler effect was investigated and discussed as well. In general, (in spite of spurious oscillations in the near field) the computed aerodynamic and far-field acoustic results were found to be in satisfactory agreement with measurements and analytical relationships.
Finally, the method was extended for the turbulent reactive flow predictions using LES. The LES formulation of the Eddy-Dissipation Concept was proposed. The validation was performed for the Sandia Flame D, where reasonable agreement between predicted and measured data was achieved. It is believed, that the observed discrepancies were related with the lack of grid resolution and inaccurate boundary conditions for the turbulence at inlet boundaries.
NTNU: Skipnes Kommunikasjon as , 2014.