A new paradigm for simulation of turbulent combustion in realistic gas turbine combustors using LES

This paper presents a new paradigm for numerical simulation of turbulent combustion in realistic gas turbine combustors. Advanced CFD methods using Large Eddy Simulation (LES) turbulence models are central to this paradigm in fluid dynamics where engineers can apply the full predictive abilities of numerical simulations to the design of realistic gas turbine combustors. The use of LES models is motivated by their demonstrated superiority over RANS to predict turbulent mixing. The subgrid scale models incorporated in LES are based on the dynamic approach where the model coefficients are computed rather than prescribed by the user. This has provided unparalleled robustness to modern turbulent flow computations using LES. A new numerical algorithm was derived that is discretely energy conserving on hybrid unstructured grids, thus allowing numerical simulations at high Reynolds numbers corresponding to operating conditions without using artificial numerical dissipation. This paper deals specifically with the simulation of the gas phase flow through realistic gas turbine combustors and the implementation of combustion and spray models that are needed to predict and control the combustion phenomena in these geometries. Results from several simulations and comparison with experimental data are used to validate this approach. In particular, a complete simulation of the unsteady flow field in a realistic combustor geometry is carried out. Some preliminary results for reacting flow simulations in gas turbine combustors are also discussed. We discuss several challenges related to large-scale simulations of the flow in realistic combustors, including methods to further accelerate the algorithm's convergence (e.g., use of multigrid techniques), improvement of the parallel performance of the flow solver for two-phase flow simulations (e.g., use of dynamic load balancing that accounts for the additional CPU time spent in the spray module when particles are present in the cells).