Large-Eddy Simulation of Turbulent Flow in an Axisymmetric Dump Combustor
|Title||Large-Eddy Simulation of Turbulent Flow in an Axisymmetric Dump Combustor|
|Publication Type||Journal Article|
|Year of Publication||2008|
|Authors||Afshari, A, Jaberi, FA|
|Keywords||combustion, Gas turbine; modeling; combustion chamber; Monte Carlo method; Lagragian Method; turbulent flow; large eddy simulation|
A hybrid Eulerian–Lagrangian, mathematical/computational methodology is developed and evaluated for large- eddy simulations of turbulent combustion in complex geometries. The formulation for turbulence is based on the standard subgrid-scale stress models. The formulation for subgrid-scale combustion is based on the filtered mass density function and its equivalent stochastic Lagrangian equations. An algorithm based on high-order compact differencing on generalized multiblock grids is developed for numerical solution of the coupled Eulerian–Lagrangian equations. The results obtained by large-eddy simulations/filtered mass density function show the computational method to be more efficient than existing methods for similar hybrid systems. The consistency, convergence, and accuracy of the filtered mass density function and its Lagrangian–Monte Carlo solver is established for both reacting and nonreacting flows in a dump combustor. The results show that the finite difference and the Monte Carlo numerical methods employed are both accurate and consistent. The results for a reacting premixed dump combustor also agree well with available experimental data. Additionally, the results obtained for other nonreacting turbulent flows are found to be in good agreement with the experimental and high-order numerical data. Filtered mass density function simulations are performed to examine the effects of boundary conditions, subgrid-scale models, as well as physical and geometrical parameters on dump-combustor flows. The results generated for combustors with and without an inlet nozzle are found to be similar as long as appropriate boundary conditions are employed.