Assessment of RANS Turbulence Models on Simplified Geometries Representative of Turbine Blade Tip Shroud Flow

Abstract

RANS turbulence models are assessed here in simulations of simplified configurations representative of geometries and flows found in tip shrouds of low-pressure turbines rotor blades. The complex flow inside a tip shroud is highly turbulent, its boundary layer separates at the sharp angled tangential fins and reattaches in the inter-fin cavity after a recirculation. The flow in the tip seal of the rotor blade, and the jet mixing flow at the exit of the shroud cause pressure losses and influence substantially the aerodynamic performance of the turbine. Such flows are quite challenging for the RANS models. The main aim of this paper is to assess the capability of the used turbulence models to predict tip shroud representative physics of detached, reattached and jet mixing flows. The simplified geometries consist of a backward facing step, a rib roughened channel and a jet in cross-flow. The Reynolds numbers of these configurations are close to those estimated in real tip shrouds. Experimental data are available in each case to evaluate the accuracy of the simulations and understand the behavior of the models. The evaluated turbulence models include several widely used eddy-viscosity models based on Boussinesq’s hypothesis and a differential Reynolds stress model. Comparisons with measured components of the Reynolds stress tensor, turbulent kinetic energy and velocity profiles are presented. The velocity profiles are in good agreement with the measurements except in the recirculation regions, and eddy-viscosity models overpredict the level of experimental turbulent kinetic energy. Analysis of the strengths and weaknesses of the models for each flow are discussed.

RANS prediction of the aerodynamic losses in a linear turbine cascade with an upstream cavity and a purge flow

“…The creation of this vortex structure can be attributed to the production of turbulent kinetic energy in the shear layer at the intersection of the two flows. Uncu et al [21] found a similar behaviour of overproduction of TKE for the k − l model in a jet in crossflow. It is worth mentioning the inlet turbulence decay for the RSM model in the channel which is not visible for the two other models according to the modelling choice.…”