Accuracy of RANS CFD Methods for Design Optimization of Turbine Blade Tip Geometries

摘要

In the never-ending search for higher engine performance, the optimization of the blade tip geometry provides an opportunity to control the tip leakage flow, mitigate the rotor mixing losses and manage the heat load distributions. However, the setup of numerical flow simulations of a high-pressure turbine stage for the purpose of blade tip geometry optimization is subjected to conflicting requirements: the computational demands have to be low enough to run a sufficiently large number of direct evaluations within a limited period of time, whereas the numerical model must guarantee sufficient accuracy in the prediction of the relevant flow physics that drives variations in the performance objectives. This paper evaluates the options for the setup of numerical simulations suitable for CFD-based blade tip optimization and assesses the impact of the CFD methods on the computation accuracy with emphasis on rotor tip flow prediction. The numerical study is performed with the Numeca FINE/Turbo and FINE/Open commercial RANS solvers on a set of optimised blade tip geometries in a high-pressure gas turbine stage. The domain is meshed with either an unstructured or a multi-block structured grid in order to assess the influence of domain discretization methods on performance parameters quantification. The location of the high-pressure stage after the combustor poses increased demands on turbulence modelling due to the high inlet turbulence intensity. This work compares the performance of the one-equation Spalart-Alllmaras (SA) and the two-equation k-ε and k-ω SST. Another section is dedicated to methods for modelling of stator-rotor interaction. Domain restriction to rotor-only is attractive for its low computational demands, but it does not allow to include the stator-rotor interaction effects. This method is compared against a full-stage setup, empolying either the commonly used steady computation with mixing plane or the Non-Linear Harmonics method which allows to capture the unsteady blade-passing effects. The computations are validated by experimental data available from a large scale high-speed turbine facility, employing a rainbow rotor approach to allow the simultaneous aerothermal testing of multiple optimized blade tip geometries.