Preliminary aerodynamic predictions in highly loaded S-shaped ducts by viscous-inviscid interaction

In large civil aeroengines, the drive for improved efficiency, reduced fuel-burn and emissions has led to engine architectures with compact cores. For the S-shaped transition ducts, connecting the larger radial offset between the compressors whilst minimising length has become an increasing challenge as a complex aerodynamic flow field is developed in response to streamline curvature, making performance difficult to rapidly assess in preliminary designs without time-consuming high-fidelity simulations. The present work therefore aims to create an application enabling a more agile design process.

The tool foundations (part I) are first discussed through the coupled pressure, velocity, stress, and turbulence flow fields in the ducts based on experimental measurements. Equations capturing the dominant curvature effects (e.g. radial and streamwise static pressure gradients) are next derived from appropriate simplifications of the Navier-Stokes equations, and performance criteria such as the minimum skin friction coefficient are selected for design space exploration.

Following this groundwork, a novel viscous-inviscid interaction method (part II) is proposed for rapid performance evaluation with a semi-analytical iterative process for two-dimensional axisymmetric incompressible steady flows. A potential core driven by streamline curvature and implicit velocity profile by spline reconstruction is linked to a momentum integral/entrainment scheme reproducing the growth of turbulent boundary layers up to separation. The shear stress distribution is generated by an adapted mixing length turbulence model finely calibrated by Ordinary Kriging, and skin friction closure is accomplished with a composite law-of-the-wall sensitive to the biased core velocity and the local pressure gradients. The viscous-inviscid coupling is finally made by core displacement. Furthermore, an innovative extension is introduced for strutted S-shaped ducts. A quasi-three-dimensional equivalent inviscid core is rebuilt by orthogonal superimposition of simpler two-dimensional computations. The shear layer developments are realised along three-dimensional curvilinear streamlines in the superimposed flow field.

Despite the flow complexity, the hub pressure loading of unstrutted ducts is, for instance, predicted within ±5% compared to current CFD methods and convergence is achieved about 20 times faster. The produced performance maps can thus be used to inform detailed CFD models with better defined engine boundary conditions for final geometry optimisations. Improvements focusing on accuracy and numerical robustness (e.g. doubly curvilinear coordinate system) are recommended in part III to refine the encouraging results of the strutted analysis. Options for tackling compressor-generated phenomena (e.g. wakes) are lastly explored by generalising successfully-applied concepts (e.g. superimposition).