Turbulence Modelling for Internal Cooling of Gas-Turbine Blades
Numerical simulations of geometrical configurations similar to those present in the internal cooling ducts within gas turbine blades have been performed. The flow within these channels are characterized by heat transfer enhancement ribs, sharp bends, rotation and buoyancy effects. On the basis of investigations on rib-roughened channel it is concluded that the frequently employed two-equation turbulence models (.kappa -.epsilon., .kappa.- .omega.) cannot predict heat transfer in separated regions with a correct Reynolds number dependency. Extensions to non-linear models, such as EARSM, do not alters this inaccurate tendency. The importance of the length-scale determining equation for this behaviour is discussed, however without any solution given. A low-Reynolds number (LRN) .kappa.- .omega. turbulence model, with improved heat transfer predictions, is proposed. The new model includes cross-diffusion terms which enhances free-shear flow predictability. A new method to reduce the mesh sensitivity for LRN turbulence models is proposed. Within the concept of finite volume codes it is shown that through a carefully treatment of the integrations for the first interior node, minor additions results in a significant reduced grid dependency for non-near-wall sensitive parameter. The latter modification in conjunction with the new .kappa.- .omega. turbulence model results in an accurate and robust method for simulating large and complex geometries within the frame of internal cooling of turbine blades.