Seminars

Recently, large eddy simulation (LES) methodology has emerged as a useful tool in modeling turbulent combustion. LES, in resolving the large-scale mixing process, offers a more accurate footing for the small-scale models that dictate combustion phenomena. Because of this, LES codes are seeing increased exposure in complex industrial applications. To this end, the primary objectives of this dissertation are to develop a versatile CFD solver for application to gas turbine combustors and to apply said solver to several computational and experimental results with varied physics in order to validate its use in these problems. The solver was developed using the open source C++ library, OpenFOAM. Modifications and additions include a low-Mach number pressure-projection algorithm, a flamelet/progress variable approach (FPVA) based combustion model, dynamic coefficient calculation, and a Lagrangian particle tracking tool for evaporating fuel droplets in an FPVA framework.  Amongst the problems examined in this work is a direct numeric simulation (DNS) of a turbulent channel, through which a transient and operationally dangerous phenomena known as boundary layer flame flashback is studied. The solver is validated in this simplified geometry before being compared against two experimental gas turbines combustors. The first is a commonly seen configuration known as a swirl burner, in which flashback is revisited in the context of more complicated flow features and an unstructured, irregular mesh. The second uses an experimental design known as an ultra compact combustor (UCC), which directs the flow azimuthally to encourage rotational mixing rather than axial. The resulting increases in G-loading strongly affect the reaction rates and flame stabilization. This problem examines the capabilities of the models to handle complex reaction mechanisms and evaporating Lagrangian particle fuel droplets. Computational results in both cases will be compared against experiments, with quantities of interest including velocity and temperature statistics.