Towards an Understanding of Traverse Migration in the High Pressure Stage of a Gas Turbine: Effects of Geometry Fidelity and Turbulence Modelling

Abstract

The aerodynamic design of a turbine stage requires the accurate prediction of radial profiles of pressure, temperature and velocity at various axial locations within the turbine stage. In the case of hot gas path components like the High Pressure Turbine (HPT), which is located downstream of the combustor, the location of the hot spot and its migration through the stage is critical in arriving at an appropriate aerofoil cooling flow requirement and distribution. In addition, the migration of the flow and the evolution of the temperature traverse through the stage impacts the aerodynamic efficiency of the stage. This is predicted using CFD techniques and has been an inevitable part of the design process. Typically, the fidelity of the computational model evolves with the component design. During early design phases, simplistic geometry is used for the simulations and progressively the fidelity is increased to resolve the geometrical features of interest, like that of the end wall film cooling and rim seal cavity geometries. The present paper provides an improved understanding of the temperature evolution in a HP turbine stage, particularly with respect to the geometry fidelity and the choice of turbulence models. Computational analyses are carried out using the Rolls-Royce in-house CFD solver, HYDRA. The geometry fidelity comparisons dealt with are discrete endwall cooling holes vs. equivalent slot and explicit cavity resolution vs. patch surface techniques. In addition, comparisons of traverses predicted using the k-Epsilon realizable turbulence model and SST k-Omega model are presented and debated. The influence of the geometry fidelity and turbulence model on the evolution of radial distribution through the stage is presented along with supporting flow field interpretations. It is concluded that the slot representation of platform cooling flow is satisfactory to replicate the overall traverse at the exit of the High Pressure Nozzle during early stages of design. The near wall temperature gradient would be lower and in the present case the Horse Shoe Vortex (HSV) at the endwalls are not observed with discrete cooling flow modelling which indicates probable aerodynamic impact. The choice of turbulence modelling could have significant impact on the traverse prediction in comparison to the geometry approximations.