The adverse pressure gradient and aerodynamic flow interactions within axial compressors lead to suction surface flow separation and recirculation at the rotor trailing edge hub corner. Corner flow separation appearance and size variation, with respect to flow incidence, affects blade loading, blockage area, and entropy generated performance loss during both on- and off-design operation. Accurate prediction of this complex non-equilibrium turbulent flow typically requires eddy resolving simulations (DES, LES, etc.). However, the computational cost associated with these models for industry relevant Reynolds number flows is unacceptable in the design process. As such, lower cost RANS models are a standard in the component design process despite known difficulties predicting the non-equilibrium flow phenomena characterizing separated flows.
A recent collaboration between researchers at the Notre Dame Turbomachinery Laboratory (NDTL) and IHI Corporation has led to the development of a RANS model calibration framework based on Bayesian inference methods. The Bayesian calibration framework was successfully tested for prediction of linear cascade corner flow separation. Calibrated parameter solutions significantly increased prediction accuracy and experimental matching of wake Mach number and stagnation pressure. Numerical simulations also demonstrated improved prediction accuracy of the corner flow separation for a broad range of off-design incidence angles outside of the calibration range.
Figure 1: Pitchwise profiles obtained from measurement (exp.), simulation using default model parameters (default), simulation using calibrated model parameters (calibrated), and simulation using reference calibrated model parameters (ref.) at 90% span.
Additional details can be found in the forthcoming GPPS Journal special issue, Data-driven Modelling and High Fidelity Modelling, expected April/May 2021 (JGPPS-00101-2020-02). Please send inquiries to firstname.lastname@example.org
By Ethan Perez, Ryan T. Kelly, and Aleksandar Jemcov
Published by Jasmin Avila