Turbulence Transport Modelling in Gas Turbine Related Applications

Doktorsavhandling, 2006

Computational fluid dynamics is a cornerstone in gas turbine engine design. It is used to optimize shapes of turbine and compressor airfoils, to predict heat transfer to gas turbine hot parts, to reduce the amount of pollutants that form when fuel is burnt, to reduce gas turbine noise and so on. There are still, however, areas where the computational methods lack in reliability and need further refinement. One is the modelling of turbulence transport effects on mean flow characteristics and is the main subject of this thesis. The thesis focuses on RANS predictions of turbulence and heat transfer, where the unknown turbulence transport terms are closed using turbulence models based on the eddy-viscosity concept. The potential of using the V2F turbulence model for heat transfer predictions in complex flows is illustrated by computing a three-dimensional stator vane passage flow. It is shown that the V2F model is able to predict the effects of turbulence on the secondary flow field in the stator passage, and, that the secondary flow field is largely what determines the heat transfer to the vane endwalls. The use of the realizability constraint to prevent unphysical growth of turbulence kinetic energy is also thoroughly discussed. There are however problems with the V2F model as well. It is in principle unable to resolve turbulence anisotropy and furthermore suffers from predicting laminar to turbulent boundary layer transition too rapidly. It is shown that the former problem can be dealt with by employing the nonlinear eddy-viscosity model of \cite{reif00}. This model is tested in the asymmetric diffuser flow and proves to be capable of very accurate Reynolds stress predictions. This study also highlights the strong sensitivity of the mean flow to turbulence closure and suggests that the near wall modelling is of the utmost importance. In an effort to improve the performance of the V2F model in transitional flows the ideas behind the transition modelling approach of Walters & Leylek (2004) are adapted to the V2F model. Also provided is an overview of the existing literature on the subject of transition modelling.