Structural Optimization Of A Vertical Axis Wind Turbine With Aeroelastic Analysis

Abstract

Currently, the price per kW of offshore wind energy is 55% larger than onshore [38, 52]. Of this price, the rotor corresponds to 22%. To reduce the price of wind energy, it is necessary to investigate wind turbine concepts with scales above 10 MW. The commonly known Horizontal Axis Wind Turbine (HAWT) requires offshore a large support structure. If the turbine is designed to be floating, a deep floater is needed to limit the tilt angle. A possible concept to meet this challenges is the lift-driven Vertical Axis Wind Turbine (VAWT). This thesis aims to optimize the structural design of a VAWT rotor blade and to decrease the mass to area ratio by varying blade shape and structural layout. The choice of mass to rotor area ratio as an optimization function follows from the fact that this area is directly proportional to the energy output while mass drives production and installation costs. The VAWT is defined by an axis perpendicular to the unperturbed flow direction. The rotor geometry is described through a Troposkein shape. It is assumed that the blades carry their own weight leading to a reinforced root region. During operation the blades experience aerodynamic and inertia forces, which are deflecting the blades outwards, leading to an alternation of the aerodynamic loads.The interplay of load alternation and blade deflection could lead to a diverging flutter motion. After a fitting design is obtained, the blade motion has to be inspected for a safe use during operation. The rotor is designed with an adjusted optimizer, originally written by M. Schelbergen [62]. The optimizer uses the Matlab optimization toolbox in combination with Nastran. The modification allows a smooth transition of the thickness of skin, shear web and girder. The airfoil section is varied along the blade. The optimization is based on load cases such as a parked rotor and the maximum up- and downwind forces. These loads are simplified and assumed to vary neither by the motion nor the deflections of the blade. In addition, an aeroelastic model is required to observe the blades’ motion. Through out this
thesis two aeroelastic codes were used. The VAWT AeroElastic Multibody Panel Solver (VÆMPS) was created by coupling Sandia National Laboratories OWENS and the near wake panel solver UMPM. However, its computational performance was not satisfying and it was decided to use HAWC2 coupled with an actuator cylinder model to determine the induction.