Doctoral thesis, 1998

Numerical simulations of three-dimensional confined swirl flows in a combustion chamber geometry are carried out with improved turbulence and combustion models.
In the first part of the work a comparative study of six turbulence models is done in application to swirl flow. The models selected include four two-equation and two second order closure models. The automatic coordinate transformation and code generation procedure is developed. The results of simulations show an improvement of flow predictions for the second order closure models. The sensitivity of the flow-field to the inflow conditions and small-scale inlet geometry changes is investigated in three different cases. The conclusions are drawn on the importance of various factors such as inlet- and boundary conditions, chamber geometry and the choice of the turbulence model on the flow predictions.
The maximum flow-rate principle is discussed in the context of turbulent swirl flows. The method of determining the Richardson correction constant with the help of the maximum flow-rate principle is proposed and validated. Numerical investigation of two swirl flow cases with heat transfer is presented.
In the second part of the thesis three-dimensional simulations of a premixed swirl-stabilized combustion with and without spray are performed. The effects of swirl are accounted for by the Richardson-number correction technique and the streamline-coordinate correction method. The turbulent-flame-speed closure model, the pressure-density interaction model and Richardson-type corrections are implemented into the Kiva computer program. The turbulent flame speed closure model is modified to account for the effect of partially non-homogeneous mixture. The maximum flow-rate principle is used to determine the Richardson correction constant. The variable density effects are captured by modeling the pressure-density interaction terms. The standard k-epsilon turbulence model and the distributed reactor combustion model are also used for comparison. The jet-A fuel combustion in stoichiometric and lean regimes is analyzed.
A comparative analysis of the results produced with the new and standard techniques is presented. Phenomenology of the flow-field in stoichiometric and lean combustion cases is discussed. The effect of spray combustion computations on the performance of the composite model is analyzed.
The improved turbulent-flame-speed closure combustion model produces a different temperature distribution and predicts a considerably greater flame thickness than the distributed reactor model. The maximum flow-rate principle predicts the correct value of the Richardson correction constant. The lean regime significantly affects the flame structure and stability. The inclusion of spray slows down the computations and leads to a non-steady solution. Numerical simulations of three-dimensional confined swirl flows in a combustion chamber geometry are carried out with improved turbulence and combustion models.
In the first part of the work a comparative study of six turbulence models is done in application to swirl flow. The models selected include four two-equation and two second order closure models. The automatic coordinate transformation and code generation procedure is developed. The results of simulations show an improvement of flow predictions for the second order closure models. The sensitivity of the flow-field to the inflow conditions and small-scale inlet geometry changes is investigated in three different cases. Conclusions are drawn on the importance of various factors such as inlet- and boundary conditions, chamber geometry and the choice of the turbulence model on the flow predictions. The maximum flow-rate principle is discussed in the context of turbulent swirl flows. The method of determining the Richardson correction constant with the help of the maximum flow-rate principle is proposed and validated. Numerical investigation of two swirl flow cases with heat transfer is presented. In the second part of the thesis three-dimensional simulations of a premixed swirl-stabilized combustion with and without spray are performed. The effects of swirl are accounted for by the Richardson-number correction technique and the streamline-coordinate correction method. The turbulent-flame-speed closure model, the pressure-density interaction model and Richardson-type corrections are implemented into the Kiva computer program. The turbulent flame speed closure model is modified to account for the effect of partially non-homogeneous mixture. The maximum flow-rate principle is used to determine the Richardson correction constant. The variable density effects are captured by modeling the pressure-density interaction terms. The standard k-epsilon turbulence model and the distributed reactor combustion model are also used for comparison. The jet-A fuel combustion in stoichiometric and lean regimes is analyzed. A comparative analysis of the results produced with the new and standard techniques is presented. Phenomenology of the flow-field in stoichiometric and lean combustion cases is discussed. The effect of spray combustion computations on the performance of the composite model is analyzed. The improved turbulent-flame-speed closure combustion model produces a different temperature distribution and predicts a considerably greater flame thickness than the distributed reactor model. The maximum flow-rate principle predicts the correct value of the Richardson correction constant. The lean regime significantly affects the flame structure and stability. The inclusion of spray slows down the computations and leads to a non-steady solution.

Chalmers, Department of Thermo and Fluid Dynamics

Mechanical Engineering

Physical Sciences

91-7197-643-4

Doktorsavhandlingar vid Chalmers tekniska högskola. Ny serie: 1406