Analytical transonic flow simulation in streamline curvature methods for axial-flow compressor design

Abstract

Aircraft gas-turbine engine component design and analysis is still a pleasant art to be further explored to satisfy the optimum engineering limits. One of the challenges for aero-engines is to provide high efficiencies while being light and compact. As a responding solution for these requisites, modern axial-flow transonic fans and compressors provide high shock-induced single-stage pressure ratios, reducing weight and size. The problem becomes thus to attain this within acceptable isentropic efficiencies and working ranges, at least comparable to those from the subsonic axial-flow compressors. The constant and crescent demand to obtain more accurate turbomachinery blading performance in the analysis and design process has led the designer to explore different levels of simulation fidelities and optimisation strategies. The extensive use of Computational Fluid Dynamics (CFD) methods in aerodynamics has made the three-dimensional (3-D) Reynolds-averaged Navier-Stokes (RANS) numerical simulations the preferred technique for turbomachinery analysis. Despite their high-order resolution and extensive flow field information that can be collected, it comes at intolerably high computational costs in terms of time and resources, especially if they are used as solvers within an optimisation framework. In contrast, two-dimensional (2-D) through-flow methods such as streamline curvature (SLC), keystone in the turbomachinery design, provide a rapid flow solution whilst offering accurate results. In the context of transonic axial-flow fans and compressors, previous 2-D SLC tools have failed to replicate the real physics related to compressible flow. More specifically, the prediction of the highly-complex shock-system shape and location for an accurate estimation of shock-associated losses has always been assumed and oversimplified. The situation aggravates, when the assumed overall shock configuration applies only for design point at unstarted operations, requiring of empirical correlations to estimate the shock-loss coefficient for off-design operations. The overall performance prediction of the fan and compressor is thereby highly-dependant on the shock modelling quality. For this reason, an analytical transonic-flow simulation package was developed and implemented into an existing in-house 2-D SLC compressor performance simulator with the aim of enhancing the aerodynamic prediction for transonic axial-flow fans and compressors. The novel toolkit to handle transonic flow and fully coupled to the 2-D SLC software consists of the following contributions: (1) a 3-D blade-element-layout method; (2) an adaptation of the full radial-equilibrium equation (REE) to handle the effects of 3-D blade shaping; (3) a physics-based shock- structure and loss model for unstarted and started operations that uses an iterative-solution method to locate the choke-induced passage-shock; (4) a choking mass flow redistribution model. In this way, shock losses were determined throughout the blade span and for various off-design operating conditions, including those at choking, where the mass flow was limited and redistributed spanwise according to the unique incidence. 2-D SLC simulations were conducted for the NASA Rotor 67 and the NASA Advanced Duct Propulsor (ADP) Fan to calibrate and validate the models accordingly against experimental rig-test data and 3-D CFD results. The analytical shock- loss and structure model improved the shock-loss prediction between 40-50% with respect of the state-of-the-art models, and showed satisfactory agreement against measured data within 0.6% at the blade tip and 0.3% at mid-span sections. A parametric study was conducted for the NASA Rotor 67 to demonstrate design trends when varying 2-D and 3-D blade-element parameters utilizing the transonic-flow toolkit in its entirety, evidencing the impact on the shock-loss radial distribution. A new cutting-edge variable-pitch fan (VPF) was designed using the NASA ADP Fan 2-D SLC model, which was further optimised using an evolutionary genetic algorithm (EGA) to demonstrate the application of the novel transonic-flow package. The single-objective optimisation consisted in the variation of sweep and lean angles with the purpose of studying the pure influence of shock losses on the overall isentropic efficiency. The optimised geometry resulted in a backward-swept and mostly pressure-side leaned VPF that incremented the isentropic efficiency by 0.4% with respect of the baseline configuration