Numerical Simulations of Cooling Concepts Related to Gas Turbine Combustors

The mean by which a gas turbine combustor is cooled is of major importance for the efficiency and emission performance of the machine. The trends of increasing top cycle temperature and leaner combustion have also been accompanied by a need of efficient use of the cooling air. Traditional cooling schemes such as film cooling are associated with high coolant flow rates and unacceptable emission levels due to the interaction of low coolant temperature and combustion gases. It is therefore of interest to develop and evaluate computational tools for analysis of improved or new cooling technologies, which may be used for design. This thesis presents numerical simulations of two concepts relevant for future gas turbine combustors; forced convection and impingement cooling. The computational approach is based on solution of the governing differential equations for fluid flow and heat transfer using the Finite Volume Method. As cost effective and robust computational tools are vital for engineering applications, Reynolds averaging in conjunction with various two-equation turbulence closures are applied. The turbulence models include linear, non-linear and explicit algebraic stress formulations. In the thesis this approach is evaluated and compared to experiments. The study on forced convection cooling is focused on ribbed surfaces, which have a high heat transfer performance due to boundary-layer disruption and extensive mixing of the coolant flow. In the real application the pressure drop also has to be considered. A calculation method for optimization is presented, where a typical gas turbine combustor channel is investigated for different rib heights, pitches and configurations at a low computational cost. It is shown that a V-shaped rib configuration supplies superior performance and the influence of rib blockage ratio is small, for a given mass flow rate and pressure drop of the cooling air. However, due to limitations in manufacturing techniques and wear, the shape of the ribs also may vary. As this possibly affects the thermodynamic performance of a cooling duct, a detailed study on relevant rib shapes is performed. The results indicate that the deformation of square ribs may increase heat transfer and reduce the pressure drop in the cooling duct. For the detailed investigations of ribbed surfaces, it is also found that the predictive ability of the computational method is strongly dependent on flow and geometry conditions, and that the turbulence modeling in the vicinity of the ribbed wall is crucial. For impingement cooling applications, an extensive validation study of the present computational approach is performed with respect to geometry and flow conditions. The effects of jet Reynolds number, nozzle to wall distance, nozzle configuration and crossflow are investigated. It is shown that simple configurations like a jet issuing from a fully developed pipe flow may be handled reasonably well, if an ad-hoc realizability constraint is applied. For the more complex flow configurations with throttling and crossflow interference of the jet, the predictions show larger deviation from experiments. This is believed to be strongly associated with the modeling of turbulence