Low Reynolds Number Effects on the Endwall Flow Field in a High-Lift Turbine Passage

Abstract

Understanding the endwall flow phenomena surrounding low-pressure turbine blades is key to improving performance, as these flow features contribute significantly to loss generation at low Reynolds number cruise. It is well documented that a horseshoe vortex system forms at the junction of the endwall and turbine blade. The vortices develop and gain significant strength in the passage and contribute to total pressure losses. During low Reynolds number conditions, the flow through a low-pressure turbine passage can be greatly impacted by a number of factors, including Reynolds number and incoming turbulence. The focus of this paper is on significant changes to the endwall flow field observed in experimental measurements and an accompanying implicit large-eddy simulation of the flow through a linear cascade of high-lift front loaded low-pressure turbine blades at low Reynolds number. Results show a significant effect on both the time-averaged endwall flow topology and the unsteady vortical flow characteristics when the Reynolds number based on inlet conditions was decreased to 30,000. Various techniques, such as spectral proper orthogonal decomposition, were used to analyze and compare both high-speed particle image velocimetry measurements and numerical results in order to extract the dominant structures and their unsteady behavior. The total pressure loss development through the passage was assessed in order to better understand how the observed changes in endwall flow structures contribute to the overall losses.

A Numerical Test Rig for Turbomachinery Flows Based on Large Eddy Simulations With a High-Order Discontinuous Galerkin Scheme—Part III: Secondary Flow Effects

“…With the general reduction of core engine sizes, the importance 2 of understanding and predicting losses generated by secondary flow 3 effects is increasing. These effects have been reviewed in the lit-4 erature for both compressors [1,2] and turbines [3] using predomi-5 nantly experimental methods or, due to the computational expense, [4][5][6][7][8][9][10]. 20 For example, Cui et al [4] presented an extensive analysis of 21 both midspan and end wall flow physics for the T106A LPT with 22 parallel end walls at a Reynolds number of 160,000 using an incompressible second order accurate, unstructured Finite Volume (FV) 24 method.…”

“…It has to be noted, though, that the inlet boundary layer had a thickness of only 3% of the channel height. Another notable series of papers focuses on the front-loaded L2F LPT cascade at different Reynolds numbers employing state-ofthe-art measurement techniques and numerical simulations with a compressible ninth order accurate FV solver with weighted essentially non-oscillatory (WENO) discretization [8,12]. Robison et al [10] investigate the difference between wakes generated by bars compared to an LPT profile (50% reaction stage) at a Reynolds number of 160,000 and a Mach number of 0.1.…”