Sensitivity Analysis in 3D Turbine CFD

A better understanding of turbine performance and its sensitivity to variations in the inletboundary conditions is crucial in the quest of further improving the efficiency of aero engines.Within the research efforts to reach this goal, a high-pressure turbine test rig has been designedby Rolls-Royce Deutschland in cooperation with the Deutsches Zentrum für Luft- und Raumfahrt(DLR), the German Aerospace Center. The scope of the test rig is high-precision measurement ofaerodynamic efficiency including the effects of film cooling and secondary air flows as well as theimprovement of numerical prediction tools, especially 3D Computational Fluid Dynamics (CFD).A sensitivity analysis of the test rig based on detailed 3D CFD computations was carried outwith the aim to quantify the influence of inlet boundary condition variations occurring in the testrig on the outlet capacity of the first stage nozzle guide vane (NGV) and the turbine efficiency.The analysis considered variations of the cooling and rimseal leakage mass flow rates as well asfluctuations in the inlet distributions of total temperature and pressure. The influence of anincreased rotor tip clearance was also studied.This thesis covers the creation, calibration and validation of the steady state 3D CFD modelof the full turbine domain. All relevant geometrical details of the blades, walls and the rimsealcavities are included with the exception of the film cooling holes that are replaced by a volumesource term based cooling strip model to reduce the computational cost of the analysis. Thehigh-fidelity CFD computation is run only on a sample of parameter combinations spread overthe entire input parameter space determined using the optimal latin hypercube technique. Thesubsequent sensitivity analysis is based on a Kriging response surface model fit to the sampledata. The results are discussed with regard to the planned experimental campaign on the test rigand general conclusions concerning the impacts of the studied parameters on turbine performanceare deduced