High-Fidelity Simulations and RANS Analysis of a Low-Pressure Turbine Cascade in a Spanwise-Diverging Gas Path - End-Wall Flow Analysis and Loss Generation Mechanisms

In order to increase the impact of cascade tests on the design of low-pressure turbines (LPT) at engine level, the end-wall geometry must be taken into account. Even if, given the high aspect ratio of LPT bladings, profile loss is expected to be the major loss source, the evolving secondary flow phenomena in the end-wall region can contribute up to 40% of loss generation. In addition, the growth of end-wall boundary layers introduce blockage effects on the otherwise undisturbed, free-stream flow away from the end-walls. Experimental test facilities are able to capture some of the resulting flow phenomena and scale-resolving computational fluid dynamics have in the past shed light on the mechanisms driving the evolution of vorticity and loss generation in the end-wall region. The majority of those past studies, however, have been conducted with spanwise-parallel end-walls. The present paper considers the MTU-T161 LPT cascade with spanwise-diverging end-walls, which feature many real engines. The diverging gas path adds another layer of complexity to the flow field. To study the aerodynamic performance of this cascade, wall-resolved large-eddy simulations and Reynolds-Averaged Navier-Stokes (RANS) analyses are performed at engine-relevant conditions of isentropic exit Reynolds number of 90,000 and isentropic exit Mach number of 0.6 with proven computational fluid dynamic solvers. In separate simulations, the momentum thickness of the incoming boundary layers is systematically varied, as the resulting end-wall flow downstream of the blade is highly dependent on the state of the incoming end-wall flow. Along with validating obtained results against experimental data, the data-rich results are used to deepen the understanding and decomposition of loss generation mechanisms in LPT flows in an engine-like, spanwise-divergent gas path. Lastly, a rigorous comparison between high-fidelity simulations and RANS shows the capability of RANS to accurately reproduce important flow features and its implications as a standard, industrial design tool.