Computational Predictions of Airfoil Roughness Sensitivity

Abstract

The current trend in today’s utility-scale wind industry is driving towards larger, lighter, and more efficient rotor designs. As design margins narrow to support this trend, the need for accurate performance predictions increases significantly. The aerodynamic impact of blade surface roughness on performance is well known and has been documented over the years. In order to address roughness effects, blade design engineers typically attempt to bound the problem by calculating airfoil performance characteristics for both ‘clean’ (maximum laminar flow) and ‘tripped’ (fully turbulent) conditions using computational tools such as RFOIL and/or computational fluid dynamics (CFD). However, accurately capturing the aerodynamic impact over this range of conditions, especially near stall, has proven to be very difficult and can have significant consequences on turbine performance predictions. The current work focuses on accurately predicting roughness effects and improving airfoil performance predictions with regards to various levels of roughness. Wind tunnel tests have been conducted on an 18% thick airfoil with various levels of artificial roughness applied at the leading edge, in addition to the use of zigzag tape. Two-dimensional airfoil computational predictions using RFOIL and the commercial Reynolds-Averaged Navier-Stokes flow solver ANSYS CFX will be presented. In order to approximate the effects of roughness on boundary layer transition, RFOIL uses the e n method while ANSYS CFX offers a wide range of turbulence models, including a wall-roughness formulation, and Langtry & Menter’s correlation-based, bypass transition model to predict transition. The resulting airfoil data polars will be used to calculate the effect on turbine annual energy production using NREL’s dynamic simulation code, FAST. In addition, results from 3D, steady-state CFD predictions will be presented for comparison.