Transforming Laser-Scanned 750 kW Turbine Surface Geometry Data into Smooth CAD for CFD Simulations

Abstract

This paper presents a method for automatically reconstructing and smoothing surfaces from laser-scanned wind turbine blades. The aim is to accurately reconstruct turbine blade surfaces in the absence of an accurate CAD model. The input consists of a series of imperfectly aligned blade point clouds, and the output is a CFD surface mesh. The automatic process starts by segmenting the blade into as many sections as there are points in the spanwise direction of the target CFD mesh. Each segment is prepared for conversion into a periodic B-spline by undergoing angular sorting, application of the Iterative Closest Point algorithm, and light smoothing with the Savitzky-Golay filter. The final surface mesh consists of a series of B-spline airfoils with matching control points fitted on a series of spanwise nonperiodic splines. The smoothed airfoils closely match the noisy point cloud data across the entire blade. Three blades of a single turbine were scanned and meshed. The maximum distance between the blade tips of the three clouds is 2.5 cm (0.1% radius). Minor differences in airfoil profiles were observed, but they had negligible effects on lift and drag. Pitch torques were slightly more affected.