Characterization of Turbine Vane Roughness Variations Using Self-Organizing Maps

Turbine vane roughness and blade roughness produce adverse effects including increased aerodynamic losses and increased heat transfer into the turbine components. Consequently, measuring levels of roughness and characterizing the roughness development as a function of operating time and conditions are critical to developing proper engine maintenance and replacement schedules. Because of real blade and vane three-dimensional geometries, historical roughness studies have employed two-dimensional traverses of surface elevation measurements performed at predetermined blade locations using stylus-based profilometry. While laser-based and white-light surface metrology has advanced considerably in the past decades, capturing the mean component geometry at the resolution required to characterize the roughness has proven to be difficult given the high surface curvature and typical surface reflectivities. Even on two-dimensional cascade vanes, local regions of high surface curvature result in situations where characterizing the nature of surface roughness is challenging. In this study, a single-vane cascade flow tunnel based on the VKI transonic vane shape was constructed to create deposition-like roughness. The roughness was generated using textured paint sprayed into the inlet plenum of the tunnel. The vane was then scanned using a NextEngine 3D Scanner HD laser scanning system. A combined self-organizing map and multi-dimensional statistics approach, which was developed by McClain and Kreeger (2013) for characterizing airframe ice roughness, is employed to represent the mean blade profile, to unwrap the surface topography, and to evaluate the roughness characteristics as a function of surface distance from the leading edge. The approach enables new insights into the variations of roughness parameters along the surface of engine components and test models. Finally, difficulties in extending the approach to real engine components with twist and other three-dimensional features are discussed.