The premixer is the most important part of DLN/DLE gas turbine combustion chambers because it is responsible for velocity and fuel concentration fields at the flametube inlet. Most currently used premixers were developed using long expensive experimental research efforts. At present, progress in high performance computing allows application of a new approach to the development process utilizing extensive numerical simulation up to the first tests of combustion chamber. However, it is necessary to be sure that the most important processes which take place inside premixers and downstream such as interaction of vane wakes with the main flow, fuel/air mixing, vortex breakdown and vortical structures generation are correctly predicted. In other words, verified methodology of the aerodynamic design of the premixers is required. The present work continues the previously performed analysis of non-reacting swirling flow inside premixers and it is directed towards more accurate determination the effect of different computational model parameters on the results of numerical simulation. In the first phase of this work, results of uncompressible swirling flow measurements were used. The effect of premixer geometry (full or simplified), grid resolution, turbulence model (k-e RNG, SST, SSG RSM, DES, SAS), number of timesteps and initial conditions on time-averaged flow parameters as well as flow structures such as precessing core were assessed. Some interesting features were found as the result of this work. For example, simplification of premixer geometry slightly effects on the results of RANS computations and entirely changes precessing core structure obtained in URANS modeling. In the second phase of this work compressible swirling flow measurements were used to compare the ability of DES and SST turbulence models to predict velocity and velocity fluctuation fields as well as vortical structures in a swirling flow. As the result of this work the main requirements for the computational fluid dynamics analysis of premixer aerodynamic were developed and validated. They specify parameters of computational model depending on the target flow parameters.

This content is only available via PDF.
You do not currently have access to this content.