Development of CFD-Based Tip Clearance Loss and Deviation Models for Axial-Flow Turbines

Accurate loss correlations are crucial for reliable turbine performance predictions with low-fidelity tools. However, conventional tip clearance models disregard some important loss effects, e. g. from the relative motion between blades and endwall, and their natural spanwise distributions, which are valuable for through-flow methods. The objective of this thesis is the development of more realistic tip clearance models by means of high-fidelity CFD simulations, which capture all three-dimensional flow effects and provide highly resolved results. Therefore, a numerical model of a single-stage, high-pressure turbine was set up, verified and validated with experimental data. The model was adapted to allow deliberate parameter variations at the rotor. This model was used in a numerical parameter study, in which all significant tip clearance parameters were varied individually. The simulations were evaluated concerning the mixed-out tip clearance losses and outflow angle deviations to formulate partial correlations for each parameter, which were eventually combined to a set of CFD-based tip clearance correlations. In addition, a downstream progression model was developed, which corrects the correlation results with respect to the local conditions at downstream blade rows. To account for interrelations between the effects, Kriging surrogate models were developed for the three primary parameters gap height, blade loading and incidence. The space-filling technique was applied to refine the models with additional CFD simulations at unprobed locations having high Krige variance. Compared to the CFD-based correlations, the surrogate models show differences at large gap heights and low blade loadings. Both developed tip clearance models were implemented into a through-flow program. Furthermore, the initial mathematical model, which distributes the losses and deviations in spanwise direction for the through-flow method, was supplemented by the more realistic distributions from the CFD simulations. For verification and validation, additional CFD simulations at different operation points were evaluated and performance calculations of a four-stage air turbine were conducted. In comparison to the conventional correlations, the new models yield more accurate tip clearance losses and deviations and increase the agreement with the efficiency and pressure ratio data of the turbine experiments. In particular the performance results obtained with the Kriging surrogate model at part load operation improved. Moreover, applying the CFD spanwise distributions enhances the local flow field prediction on the meridional stream surface. In conclusion, the numerical approach succeeded in developing more accurate and realistic tip clearance models.