Abstract
The current energy challenges in the field of aircraft propulsion demands a better understanding of turbine flows. The complexity of the flows and the geometrical configurations at play limit the feasibility of experimental investigations in this field. Building predictive numerical methods to capture accurately the flow physics is thus important, even if it still constitutes a challenge.
This paper focuses on numerical simulations of high-speed low-pressure turbine blades (HS-LPT), a major component in the framework of high-efficiency geared turbofan engine designs. In this framework, Wall Resolved Large-Eddy Simulations (WRLES) of flows in HS-LPT transonic cascades are performed, using a high order Finite Volume Method (FVM). The Explicit Compressible Solver (ECS) of the massively parallel code “YALES2” is used here. The study focuses on two different cases of low-pressure turbine cascades. First, the well-known T106-C cascade benchmark is studied in order to assess the YALES2 solver for compressible turbomachinery flows. The predictions matches fairly well with those of other numerical codes in the literature and experiments. The second test case investigated is the “SPLEEN” cascade, a next-generation high-speed low-pressure turbine cascade developed by Safran aircraft engines and the von Karman institute for fluid dynamics as a part of a large scale collaborative project. The aim here is to assess the solver for transonic flows, in terms of prediction of the SS separation bubble, the separation-induced laminar-turbulent transition and the wake deficit, using a very detailed experimental database. The results are thus compared with high-fidelity experimental data. This paper shows that the method presented here is able to provide predictive results that can be further used to help in the design LPT blades.
