Numerical Study of the Static and Pitching RISØ-B1-18 Airfoil

1 Introduction
The aim of this work is the better understanding of the physics of the aeroelastic motion of wind turbine blades in order to improve the numerical simulation of such dynamical systems. In previous works, both aerodynamic damping calculations and fully-coupled aeroelastic simulations of wind turbine blades were performed by using the CFD code EllipSys3D as a fluid flow model, and the aeroelastic code HAWC as a structural model [1, 2]. The results were compared with semi-empirical dynamic stall engineering models, such as the Beddoes-Leishman model [3]. However, the lack of experimental results in such configurations made it difficult to conclude which model was performing better. In order to clarify this issue, it was decided to come back to more basic cases for which experimental results exist in the litterature. Not many experimental campaigns exist for which the operational conditions (including Reynolds number, Mach number, etc...) are close to our concern, namely wind turbine applications. One of the rare experimental set-ups that meets these requirements is the measurements performed by Risø in the VELUX wind tunnel [4]. The airfoil profile that has been chosen for our comparative tests is the RISØ-B1-18 airfoil designed by Fuglsang et al [5].

The so-called Direct Numerical Simulation of the fluid flow dynamics (for which all the scales of the turbulent flow are simulated by the numerical code) around a turbine blade is still far out of reach of modern computers. Therefore, turbulence models have to be implemented in the numerical codes in order to reduce the computational costs to an acceptable size. Two types of turbulence modelling, which have been implemented in the Navier-Stokes code EllipSys3D, are considered in this paper. In a first place, the so-called Reynolds Averaged Navier-Stokes (RANS) approach using the k − ! SST turbulence model by Menter [6] has been implemented. The simulation of a wind turbine rotor with this model has proven to give sensible results [7] compared to the well-detailed measurements obtained during the NREL Unsteady Aerodynamics Experiment Phase VI performed by NREL at the NASA-Ames wind tunnel [8]. However, a more detailed study showed that the RANS approach alone was unable to correctly simulate the three-dimensional patterns in the flowfield around the 2D section of a pitching airfoil [9]. Conversely, the so-called Detached Eddy Simulation (DES) [10], which is a combination of a RANS approach in the vicinity of the blade and a Large Eddy Simulation (LES) in the far field, gave promising results [9].  inally, two-dimensional simulations using the k − ! SST model will be performed, mainly due to their low computational costs, and in order to evaluate the discrepancies that can be expected with such simulations compared to three-dimensional and experimental results.